The math behind ordering 3 trump, no 2nd Rd hand from S1-R1

Ask questions, discuss and debate your strategies, euchre polls and more
Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Re: The math behind ordering 3 trump, no 2nd Rd hand from S1-R1

Unread post by Wes (aka the legend) » Thu Mar 11, 2021 3:59 pm

jblowery wrote:
Thu Mar 11, 2021 3:28 pm
Would you order up from S1R1 if the upcard is higher than what you have? For example I'm in S1 and holding (Card_K-H) (Card_Q-H) (Card_10-H) (Card_K-C) (Card_K-S)

Upcard is the (Card_A-H)

I'd still do it but your chances of taking this are even more bleak.
Yep I'd call that. I would call QhTh9h9c9s vs a Ah too.

The exception to the rule is vs a Jack upcard. Then I bag except at 9-9. Vs a Jack upcard the likelihood this call makes a point goes down significantly and the chances our opponents call if we pass goes up significantly which mitigates our problem of having no where to go in the 2nd rd.

At 9-9, 3 trump vs the Right is unfortunately as good as it's gonna get for your team so I'd go for it.

I'd also call vs a Jack when up 9-6/9-7. Basically a semi-donate.



Tbolt65
Posts: 820
Joined: Sun Jun 16, 2019 9:14 pm
Location: Las Vegas

Unread post by Tbolt65 » Thu Mar 11, 2021 6:46 pm

You know where I stand. I am not calling for the most part. This is a true no where to go in the 2nd round hand. That I will pass. This is a rare enough hand you dont see alot but when you do. I advocate being more selective when/if your calling.

I do agree with the calling stated here at 9 to 9 or up 9to6, 9to7.

Tbolt65
Edward

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Thu Mar 11, 2021 6:54 pm

Wes (aka the legend) » Thu Mar 11, 2021 3:59 pm

jblowery wrote: ↑
Thu Mar 11, 2021 3:28 pm
Would you order up from S1R1 if the upcard is higher than what you have? For example I'm in S1 and holding (Card_K-H) (Card_Q-H) (Card_10-H) (Card_K-C) (Card_K-S)

Upcard is the (Card_A-H)

WES SAID, "I'd still do it but your chances of taking this are even more bleak.

Yep I'd call that. I would call QhTh9h9c9s vs a Ah too"

Yet, you would not call with QH TH 9H AS KS. You make NO SENSE WES!

Irishwolf

Tbolt65
Posts: 820
Joined: Sun Jun 16, 2019 9:14 pm
Location: Las Vegas

Unread post by Tbolt65 » Thu Mar 11, 2021 7:27 pm

The more bleak comment I believe was from jblowery when mulling over the hand that was just put forth of what they might do.

Tbolt65
Edward

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Thu Mar 11, 2021 7:34 pm

irishwolf wrote:
Thu Mar 11, 2021 6:54 pm
Wes (aka the legend) » Thu Mar 11, 2021 3:59 pm

jblowery wrote: ↑
Thu Mar 11, 2021 3:28 pm
Would you order up from S1R1 if the upcard is higher than what you have? For example I'm in S1 and holding (Card_K-H) (Card_Q-H) (Card_10-H) (Card_K-C) (Card_K-S)

Upcard is the (Card_A-H)

WES SAID, "I'd still do it but your chances of taking this are even more bleak.

Yep I'd call that. I would call QhTh9h9c9s vs a Ah too"

Yet, you would not call with QH TH 9H AS KS. You make NO SENSE WES!

Irishwolf
The situations are not analogous. Come on bro.

Btw I'm slowly doing my own kitchen table sample of the 3S-R1 QhTh9hAsKs hand because I respect your passion. I wanna at least do 450 hands. That's when I reach a 95% confidence interval correct? This is gonna take me around a month. I'm trying to do at least 10 hands a day. I'm 40 hands in and I must say it's not looking good for the pass strategy! To be continued.

I also need to do the same kindve sample for these S1-R1 calls I'm advocating itt.

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Thu Mar 11, 2021 10:18 pm

That is terrific you are doing some hands. I hope played both ways, PASS vs ORDER! And by the way, I did hands on the AH KH 9H 9D 9S, and the euchre rate is closer to 45% by ordering the QH. WHICH, makes the -.10 too low. (-.80 vs -.90) for euchres. I suspect the -.58 is much a much lower number, but that said, all things being equal, it still favors ordering with this hand.

So what I also have to add of the euchre rate of ordering, of the two hands AH KH 9H 9D 9S (S1) vs QH 10H 9H AS KS, the latter is half (~22.5% vs 45%). Of course we know the variance is approximately +/- 3%.

But I also have to say, even good players make errors of decisions on making or passing. So Situational, might make it just as favorable to pass. I see all kinds of errors being made, good and average players. Good players pushing the envelop too far. So the favorite saying, "IT ALL DEPENDS," and avoid being predictable!

"I'm trying to do at least 10 hands a day. I'm 40 hands in and I must say it's not looking good for the pass strategy! To be continued."

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Fri Mar 12, 2021 4:49 am

irishwolf wrote:
Thu Mar 11, 2021 10:18 pm
That is terrific you are doing some hands. I hope played both ways, PASS vs ORDER!
Of course, and I'm logging everything in an excel spread sheet. Just did 10 more hands for a sample of 50 now. After dealing this out 50 times, calling has a +EV of .34. IOW the EO of calling was .2 vs the EO of passing being -.14.

My standard deviation is 1.8026 and thus my variance (S^2) is 3.2494. I got these numbers from imputing the EV of passing for each hand at this site: https://www.calculator.net/standard-dev ... &x=65&y=16
irishwolf wrote:
Thu Mar 11, 2021 10:18 pm
I did hands on the AH KH 9H 9D 9S, and the euchre rate is closer to 45% by ordering the QH. WHICH, makes the -.10 too low. (-.80 vs -.90) for euchres. I suspect the -.58 is much a much lower number, but that said, all things being equal, it still favors ordering with this hand.
Good stuff. That will be the next hand I do a sample on.
irishwolf wrote:
Thu Mar 11, 2021 10:18 pm
So what I also have to add of the euchre rate of ordering, of the two hands AH KH 9H 9D 9S (S1) vs QH 10H 9H AS KS, the latter is half (~22.5% vs 45%). Of course we know the variance is approximately +/- 3%.
In case you're curious, my euchre rate for the 3rd Seat call QH TH 9H AS KS so far is 34%
irishwolf wrote:
Thu Mar 11, 2021 10:18 pm
But I also have to say, even good players make errors of decisions on making or passing. So Situational, might make it just as favorable to pass. I see all kinds of errors being made, good and average players. Good players pushing the envelop too far. So the favorite saying, "IT ALL DEPENDS," and avoid being predictable!

"I'm trying to do at least 10 hands a day. I'm 40 hands in and I must say it's not looking good for the pass strategy! To be continued."
I think it's inevitable that relatively great players will make systematic mistakes if they don't know the math no matter what card game they are playing (assuming a minimum level of complexity). Confirmation bias plagues all in this regard. My mistake is I never realized how easy it is to reach a 95% confidence interval doing a kitchen table sample. I mean it's tedious for sure, but not hard, and not too time consuming if one's modest goal is to say do one sample per month. Doing these kitchen table samples has the potential to plug up a lot of systematic holes. This is exciting to me.

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Fri Mar 12, 2021 9:28 am

Wes,

One thing you have to do is pull out any hands that S2 would assist the dealer. Those were and are significant in my sampling. Curious that you euchre rate is 34%.

Irishwolf

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Fri Mar 12, 2021 11:11 am

irishwolf wrote:
Fri Mar 12, 2021 9:28 am
Wes,

One thing you have to do is pull out any hands that S2 would assist the dealer. Those were and are significant in my sampling. Curious that you euchre rate is 34%.

Irishwolf
Oh I do. Whenever S2 or S1 would've called I cancel the hand and reshuffle. I wouldn't pay too much attention to my euchre rate. Still a small sample.

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Fri Mar 12, 2021 1:30 pm

Eldest ordering, insignificant. He would have to have all three trumps, and that is less than 1.3%. And even then he would not have any cards in next or would pass. Double bowers with even two aces, it's pass. I think I found ONE hand in 300 that S1 would have ordered.

~Irishwolf

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Fri Mar 12, 2021 5:38 pm

Okay Wes, this is what I did not tell you. I or you or anyone, does not need 450 hands. And this is where Eric failed in his book. What I do is do my hands in sets of 25 hands. 25 is not significant, you could do in sets of of 20. I do this then I test for randomness. I like 25 as it fits my notebook. So my sets are in 25 hands to each position + stock. And I deal in 1's to meet "randomness".

Okay, stay with me. So for randomness with 18 cards, you can test now test to see if the Right, Left or any off suit ace meets the expected frequency. So for example, the JH should fall approximately 27.8% to S1 or S2 or S4. For example in a hand it was 7 of 25 times you would expect it to be to any position, and 16.7% to the Stock. Believe me it will vary, like 5 to 9 but over the course of six sets of 25 hands it should be close to EV. You can flip a coin, heads or tails, it will never be exactly 50:50 but should approach the expect frequency.

When done, you now have two sets of data, a compared comparison between Ordering vs Passing. So for 300 hands I have 12 sets (24 data points matched) of results of Order vs Passing of scoring a point. You can now do a Statistical Test called a Paired Comparison Student's t-test. You are testing that there is no statistical difference, generally at the 95 level of significance. But it will tell you what the results is.

And guess what, I don't need 300 hands. I can do the test with 200 hands (16 data points). It is a powerful test. Plus I can slice and dice and do many things with the results (points, euchres, etc). In science or testing anything. How you organize or format your test data is very important. What do I do with the results? But I am anal or OCD, just my nature!

Without the t- test, and you just do hands and now have raw data results. It becomes opinion if the data is significant or not. So I like doing my tests like this.

This is what Eric did not do in his book in his test results. He got results but against what? The problem with euchre, due to inadequate shuffling, the standard deviate is very high (4.00 to 6.00). Plus/minus one standard deviate is 68% of the population, two is 95%. Euchre does follow what is called a normal distribution curve. Two shuffles or over and under never gets the cards randomized.

So chew on that some!

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sat Mar 13, 2021 1:03 am

irishwolf wrote:
Fri Mar 12, 2021 1:30 pm
Eldest ordering, insignificant. He would have to have all three trumps, and that is less than 1.3%. And even then he would not have any cards in next or would pass. Double bowers with even two aces, it's pass. I think I found ONE hand in 300 that S1 would have ordered.

~Irishwolf
True. My sample is now up to 70 and a S1-R1 order has never happened.

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sat Mar 13, 2021 1:50 am

irishwolf wrote:
Fri Mar 12, 2021 5:38 pm
Okay Wes, this is what I did not tell you. I or you or anyone, does not need 450 hands.
I was just going by what you said earlier itt:
irishwolf wrote:
Fri Mar 12, 2021 5:38 pm
For a 95% +/- accuracy you only need 450 samples.
I don't know where or how you got that number. The last time I took a stat class was in 1998 or 99. I just took it upon faith that it was approximately correct.
irishwolf wrote:
Fri Mar 12, 2021 5:38 pm
And this is where Eric failed in his book. What I do is do my hands in sets of 25 hands. 25 is not significant, you could do in sets of of 20. I do this then I test for randomness. I like 25 as it fits my notebook. So my sets are in 25 hands to each position + stock. And I deal in 1's to meet "randomness".
I also deal in 1's becuz I recalled you saying you did that awhile ago, and it made sense to me.
irishwolf wrote:
Fri Mar 12, 2021 5:38 pm
Okay, stay with me. So for randomness with 18 cards, you can test now test to see if the Right, Left or any off suit ace meets the expected frequency. So for example, the JH should fall approximately 27.8% to S1 or S2 or S4. For example in a hand it was 7 of 25 times you would expect it to be to any position, and 16.7% to the Stock. Believe me it will vary, like 5 to 9 but over the course of six sets of 25 hands it should be close to EV. You can flip a coin, heads or tails, it will never be exactly 50:50 but should approach the expect frequency.

When done, you now have two sets of data, a compared comparison between Ordering vs Passing. So for 300 hands I have 12 sets (24 data points matched) of results of Order vs Passing of scoring a point. You can now do a Statistical Test called a Paired Comparison Student's t-test. You are testing that there is no statistical difference, generally at the 95 level of significance. But it will tell you what the results is.

And guess what, I don't need 300 hands. I can do the test with 200 hands (16 data points). It is a powerful test. Plus I can slice and dice and do many things with the results (points, euchres, etc). In science or testing anything. How you organize or format your test data is very important. What do I do with the results? But I am anal or OCD, just my nature!
This stuff is a little over my head. All I am capable of right now is dealing out hands and comparing the EO of calling vs passing to get my overall EV. It was a comforting thought that I only had to deal out 450 hands to reach a 95% confidence interval. I hope that's still true. Other methods or tests that are more efficient than this are outside of my purview as of right now until I learn more. I don't mind being inefficient tho and dealing more hands than theoretically necessary as long as I reach a meaningful answer.
irishwolf wrote:
Fri Mar 12, 2021 5:38 pm
Without the t- test, and you just do hands and now have raw data results. It becomes opinion if the data is significant or not. So I like doing my tests like this.
Here is my strategy to figure out whether my data is significant or not. As mentioned before the last time I took a stat class was before the 21st century. There's a chance I'm doing something horribly or laughably wrong. So I'm gonna show you my method step by step and if I'm f**king something up you can let me know.

1) I deal out the hands to get the EO of calling vs the EO of passing which then gives me the EV of passing which is all entered in my excel spreadsheet. EG: I deal out a hand and calling gets 2 points, then I play the same hand and passing leads to the dealer passing and my P calling and getting euchred in the second round (-2), so the EV of passing for that one hand is -4. Another example: calling gets euchred (-2). If I pass the dealer gets 1 point which is -1 pts for my team. EV of passing for this hand = 1. I.E. passing does 1 pt better than calling for that hand.

2) I enter all my data points (each EV of passing data point) in that box at this website: https://www.calculator.net/standard-dev ... &x=61&y=24

My sample is up to 70 hands now. Here are literally all my data points I put in that box:
1,1,1,1,1,0,1,1,0,1,-4,0,0,-4,1,-4,,-4,1,1,0,-2,-3,-1,1,-4,1,-1,1,-3,1,1,1,-1,1,1,0,0,0,-3,-3,1,1,1,1,1,0,-4,0,-1,1,0,0,1,-1,-1,-3,1,-3,-4,-2,1,1,0,1,1,-3,1,0,0,-3
Once I do that, the site calculates the standard deviation, variance and mean of my sample. The mean of my sample is actually the average EV of the passing strategy (the number I want). I don't need the site for that. My excel gives me that number, but it's nice that I can verify this number is correct using a different source (my excel).

3) Now, equipped with this information, I go to another calculator site: https://www.medcalc.org/calc/test_one_mean.php

And I enter the mean of my sample which again is the same as the EV of passing. So far that number is -.4286. Then I enter the standard deviation. So far that's 1.7740. I enter the sample size (70), and then I set the null hypothesis to zero and click the "test" button. After doing all that this site give me my P value, which right now is P = 0.0471. Which suggests I've already reached a 95% confidence level. WTF?! Am I doing something wrong here or is it really the case that I have reached that level?
irishwolf wrote:
Fri Mar 12, 2021 5:38 pm
This is what Eric did not do in his book in his test results. He got results but against what? The problem with euchre, due to inadequate shuffling, the standard deviate is very high (4.00 to 6.00). Plus/minus one standard deviate is 68% of the population, two is 95%. Euchre does follow what is called a normal distribution curve. Two shuffles or over and under never gets the cards randomized.

So chew on that some!
Since Eric was using a computer simulation, I assume the problem with non-random shuffling was eliminated. When you say "he got results but against what?" is that where the null hypothesis comes in? When I set the null hypothesis to zero that effectively means my null hypothesis is the EV of calling and passing are equal and obviously I'm hoping my data will reject that hypothesis in one direction or another showing us which strategy is +EV. If it is true that my 70 hand sample has already reached a 95% confidence interval (which I don't understand how that could be the case), then I would reject the null hypothesis and accept the alternative hypothesis that the EV of passing and calling are NOT equal, as passing has an EV of -.4286 per hand, therefore calling with QhTh9hAsKs from 3S-R1 is the best play.

Again, let me know if I did anything wrong.

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Sat Mar 13, 2021 9:11 pm

Been outside working all day, just catching up. We did our hands differently. But that is okay. Here is how I did it:

This is how I did my data. Dealt a set of 25 hands. Then played it both ways, first by S3 ordering the KH. The recorded the results, +1, +2, or -2 (a euchre) for each hand. When done, I just added up the points for the 25 hands. Then I played each hand Passing. So if the Dealer made KH then I recorded results (-1, -2, or +2 (if dealer got euchred). Opponents getting points was always recorded as a negative number. If S4 passed, then it was up to S1 to Make or Pass. NOW THE HARD PART COMES ON 2R. If he did then points are recorded (+1, +2, +4 or -2 (euchred)). S1 only made trump next or crossed the suit with a biddable hand, no wild guesses (same for S2, S3 but S4 had to call). If S1 passed then likewise for S2, or S3 or back to S4 (STD). For all 25 hands I did this.

At the end I now have two sets total points for 25 hands. I did this all the way to 300 hands (12 sets of total points played both ways). I summed those 12 sets and divided by 300 to get a EV for playing both ways. That is my mean, EV. Total points summed in this manner.

So what is different is that the doing your method, the cost of Passing is greater. If S3 passes, and would have made a sweep but passed and S1 got euchred, that is -4. In my data, that would have been only -2. I think your data will show a higher negative EV of S3 passing. Probably more accurate, mine is more conservative. I think the number of -4s in your data shows those euchres, lol.

Here is the other thing, S3 hand, any MAKING NEXT FOR S3, bombs big time as S3 has no next. In my data, I only had S1, S2 calling or crossing suit with a biddable hand. So you know that could increase results one way or the other. The real hard decision making is when S3 & S4 passes those are estimates, only way to do it. But that is life.

1) I deal out the hands to get the EO of calling vs the EO of passing which then gives me the EV of passing which is all entered in my excel spreadsheet. EG: I deal out a hand and calling gets 2 points, then I play the same hand and passing leads to the dealer passing and my P calling and getting euchred in the second round (-2), so the EV of passing for that one hand is -4. Another example: calling gets euchred (-2). If I pass the dealer gets 1 point which is -1 pts for my team. EV of passing for this hand = 1. I.E. passing does 1 pt better than calling for that hand.

Hey, you are on the right track, Keep going. 70 is too small a number. That calculator is just giving you the difference in means as being significant based on your sample size. I don't see that as swinging the other direction. 1.77 std dev. is based on your possible range of -4 to +4. Disregard that I gave you a different SD as it based on total points, different data. Not comparable.

Again, keep going. You are doing fine. In tests and statistics, there can be different approaches to testing.

Irishwolf

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Sun Mar 14, 2021 11:09 am

i had more time to review. Yes, that is correct even with just 70 hands, the P value of 0.047 does suggest you have reached 95.53 confidence level that Ordering is better than Passing. And just play around with and insert sample size of 300, for example. Now the P value moves to 0.0001, the CI narrows and you now get 99.09% confidence that the Null Ho is rejected (i.e. there is a statistical difference between the two means of Passing vs Ordering). The reason you reach the 95% confidence with just 70 samples is because the two Means are so different between Passing and Ordering. I love statistics!

Even if your mean dropped from -0.42 to -0.20 for 300 hands, you will get a 95% confidence level the two means are significantly different, Ordering is better than Passing. I predict just on 70 hands, that will NOT happen. As sample size increases, the CI tightens and the confident level increases.

If done correctly, better than a unpredictable Simulator, lol.

BUT KEEP GOING! I LOVE IT!

WES SAID, "And I enter the mean of my sample which again is the same as the EV of passing. So far that number is -.4286. Then I enter the standard deviation. So far that's 1.7740. I enter the sample size (70), and then I set the null hypothesis to zero and click the "test" button. After doing all that this site give me my P value, which right now is P = 0.0471. Which suggests I've already reached a 95% confidence level. WTF?! Am I doing something wrong here or is it really the case that I have reached that level?"

~Irishwolf

Richardb02
Posts: 748
Joined: Mon Dec 10, 2018 8:57 pm
Location: Florida

Unread post by Richardb02 » Sun Mar 14, 2021 3:47 pm

irishwolf wrote:
Fri Mar 12, 2021 5:38 pm
Okay Wes, this is what I did not tell you. I or you or anyone, does not need 450 hands. And this is where Eric failed in his book. What I do is do my hands in sets of 25 hands. 25 is not significant, you could do in sets of of 20. I do this then I test for randomness. I like 25 as it fits my notebook. So my sets are in 25 hands to each position + stock. And I deal in 1's to meet "randomness".

Okay, stay with me. So for randomness with 18 cards, you can test now test to see if the Right, Left or any off suit ace meets the expected frequency. So for example, the JH should fall approximately 27.8% to S1 or S2 or S4. For example in a hand it was 7 of 25 times you would expect it to be to any position, and 16.7% to the Stock. Believe me it will vary, like 5 to 9 but over the course of six sets of 25 hands it should be close to EV. You can flip a coin, heads or tails, it will never be exactly 50:50 but should approach the expect frequency.

When done, you now have two sets of data, a compared comparison between Ordering vs Passing. So for 300 hands I have 12 sets (24 data points matched) of results of Order vs Passing of scoring a point. You can now do a Statistical Test called a Paired Comparison Student's t-test. You are testing that there is no statistical difference, generally at the 95 level of significance. But it will tell you what the results is.

And guess what, I don't need 300 hands. I can do the test with 200 hands (16 data points). It is a powerful test. Plus I can slice and dice and do many things with the results (points, euchres, etc). In science or testing anything. How you organize or format your test data is very important. What do I do with the results? But I am anal or OCD, just my nature!

Without the t- test, and you just do hands and now have raw data results. It becomes opinion if the data is significant or not. So I like doing my tests like this.

This is what Eric did not do in his book in his test results. He got results but against what? The problem with euchre, due to inadequate shuffling, the standard deviate is very high (4.00 to 6.00). Plus/minus one standard deviate is 68% of the population, two is 95%. Euchre does follow what is called a normal distribution curve. Two shuffles or over and under never gets the cards randomized.

So chew on that some!
Thank you IrishWolf. I have always appreciated your analysis. Even more so since you explained, in detail, how you have great confidence with fewer test hands. You even enabled my detail-avoiding,non-OCD, non-statistically trained mind to wrap itself around your major concepts. Please accept this as my highest compliment!

Richardb02
Posts: 748
Joined: Mon Dec 10, 2018 8:57 pm
Location: Florida

Unread post by Richardb02 » Sun Mar 14, 2021 3:57 pm

This thread has 3 hands in it. This is the hand being discussed, in the past 12 or so posts:
(Card_K-H) Up Card
(Card_9-H) (Card_10-H) (Card_Q-H) (Card_A-S) (Card_K-S) Seat 3 hand
Score 6 to 6. Assume all equally skilled players.

The simple question is should your order or pass?

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sun Mar 14, 2021 4:44 pm

irishwolf wrote:
Sun Mar 14, 2021 11:09 am
i had more time to review. Yes, that is correct even with just 70 hands, the P value of 0.047 does suggest you have reached 95.53 confidence level that Ordering is better than Passing. And just play around with and insert sample size of 300, for example. Now the P value moves to 0.0001, the CI narrows and you now get 99.09% confidence that the Null Ho is rejected (i.e. there is a statistical difference between the two means of Passing vs Ordering). The reason you reach the 95% confidence with just 70 samples is because the two Means are so different between Passing and Ordering. I love statistics!
It's funny, I was thinking about this hand as I was trying to get to sleep last night, and I had the same kind've realization and was planning on making a similar post today. My sample is right, and my method is correct. I have reached a 95% confidence interval already. In fact if you go back to my data points, I actually reached it in 60 hands! And this makes sense "because the two means are so different" but here's another more intuitive way to explain what's going on here: The number of hands it takes to reach a 95% confidence interval is totally dependent on the nature of the hand we are testing. Let's take an extreme, absurd example to illustrate this. Say we wanted to test what was better between going alone in hearts or passing with this hand:

(Card_J-H) (Card_J-D) (Card_A-H) (Card_10-H) (Card_9-H)

Obviously we would reach that 95% confidence interval in no time showing us going alone in hearts is better than passing.

But if we were testing a more marginal hand like say L+1+an off ace when you don't block much from the dealer spot, now it would take way more hands, probably hundreds to reach that 95% interval.

Now back to your hand from the 3S-R1:

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_A-S) (Card_K-S)

The reason I reached a 95% confidence interval in just 60 hands is becuz this spot isn't as marginal as it looks. Calling really is THAT much better than passing and it's actually easy to explain why beyond the numbers. The problem with passing is you have SUCH little help for your P in the 2nd round, and if your P is a strong aggressive player who plays good defense from S1-RD--always calling something veering towards Next when he doesn't block reverse next--you're just setting your P up to fail. Passing your hand from 3S-R1 creates too many -4s and -3s and we know exactly why that is the case. We only really have help for our P in the 2nd round if he calls spades, but most of the time he's calling Next!

Another thing about your hand, is calling gets a 2 pt march more than most would think. All it really takes is for S1 to have the Right or the effective boss trump and your team has a great chance at a march. Just need the dealer or S2 to not have 2 trump which can easily happen as there are only 2 unaccounted trump left beyond the upcard KH.

Also the reason why this hand is an intuitive mind-f**k if one doesn't do this sample is becuz when you pass this hand the majority of time you do better or the same as calling! 32/70 (45.71%) of the time passing does better than calling, and 15/70 (21.43%) of the time passing and calling perform the same. IOW 67.14% of the time passing does better or the same as calling and YET calling is the clear better play!!! This is the perfect recipe for creating a stubborn confirmation bias! Two-thirds of the time one feels good about passing. That's very hard for the human brain to escape without doing a sample.
irishwolf wrote:
Sun Mar 14, 2021 11:09 am
Even if your mean dropped from -0.42 to -0.20 for 300 hands, you will get a 95% confidence level the two means are significantly different, Ordering is better than Passing. I predict just on 70 hands, that will NOT happen. As sample size increases, the CI tightens and the confident level increases.

If done correctly, better than a unpredictable Simulator, lol.

BUT KEEP GOING! I LOVE IT!
I'm not gonna keep going with this hand. But that's not out of laziness! I really think it's THAT over. Calling is clearly better. It isn't JUST that I reached a 95% confidence interval so fast and that we know my method is correct but also becuz we can both easily explain what is going on. The fact that our pass causes our aggressive P to get euchred so much in the 2nd round becuz S3 has such little help, and the fact that if S3 calls his team will get a 2 pt march more than most would predict. And another underrated fact: having no gap in S3's suited king (AsKs) means his non-fresh spade lead is guaranteed to force out enemy trump or walk. This helps S3's call get a lot of 1 pt scores and 2 pt marches.

The only critique of this sample is that I still haven't tested if calling does better than passing assuming an amateur S1. An amateur S1 isn't calling in the 2nd round unless they have a real hand, so they're not gonna run into as many euchres after S3 passes in the 1st rd. So that could change things. But honestly I don't really think it will matter. Becuz S1 passing in the 2nd rd is costly in itself (becuz passing in euchre is inherently costly), I would predict that calling still easily beats out passing even with an amateur P in S1.

Ok now it's time for me to eat crow. You were right and I was wrong. And I was wrong to claim you were dishonest. In my defense tho I felt you were acting way too sure and certain about something when I assumed you hadn't even reached a 95% confidence level, and I totally underestimated the real power of a kitchen table sample. But now I can see why you were so sure of yourself! So I am sorry about that and ashamed about it, and I hope I don't make that mistake again becuz it's a terrible mistake to throw someone's integrity under the bus without the proper level of evidence. In a sick ironic twist I was guilty of being more certain than I should've been, the exact crime I was incorrectly accusing you of committing.

Ok, so I'm going to stop doing this sample, but I am NOT EVEN CLOSE to done. Now that I understand the power of a kitchen table sample, and now that I am supremely confident in my method, partly becuz you endorsed it and also becuz I have a better understanding of what's going on with these numbers, I'm gonna do a ton of these samples. There are all kinds of close spots than can effectively be resolved now. This is extremely exciting to me.

Here's the first hand I will sample, it's a variation of your hand, S3-R1 vs KH:

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_A-S) (Card_9-S)

I want to demonstrate that calling beats out passing with that hand, so we can say it doesn't matter what suited ace we have.

Then the next hand I want to sample in this spot is this:

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_A-D) (Card_K-D)

Notice this hand actually hits a Next call decently. That could change things.

Down the road I wannad do this hand from 3S-R1 too:

(Card_A-H) (Card_10-H) (Card_9-H) (Card_A-S) (Card_K-S)

Is this hand baggable becuz we at least have two aces to help our P out in the 2nd round? I wanna sample this and find out.

Edit: I was wrong about reaching a 95% confidence interval in 59 hands, it was 60 hands. So I changed the number.
Last edited by Wes (aka the legend) on Sun Mar 14, 2021 6:00 pm, edited 1 time in total.

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sun Mar 14, 2021 4:49 pm

Richardb02 wrote:
Sun Mar 14, 2021 3:57 pm
This thread has 3 hands in it. This is the hand being discussed, in the past 12 or so posts:
(Card_K-H) Up Card
(Card_9-H) (Card_10-H) (Card_Q-H) (Card_A-S) (Card_K-S) Seat 3 hand
Score 6 to 6. Assume all equally skilled players.

The simple question is should your order or pass?
Calling is better and it's not really close. Anybody who passes here sux at euchre! :P

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sun Mar 14, 2021 8:12 pm

Wes (aka the legend) wrote:
Sun Mar 14, 2021 4:44 pm
There are all kinds of close spots than can effectively be resolved now. This is extremely exciting to me.

Here's the first hand I will sample, it's a variation of your hand, S3-R1 vs KH:

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_A-S) (Card_9-S)

I want to demonstrate that calling beats out passing with that hand, so we can say it doesn't matter what suited ace we have.

Then the next hand I want to sample in this spot is this:

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_A-D) (Card_K-D)

Notice this hand actually hits a Next call decently. That could change things.

Down the road I wannad do this hand from 3S-R1 too:

(Card_A-H) (Card_10-H) (Card_9-H) (Card_A-S) (Card_K-S)

Is this hand baggable becuz we at least have two aces to help our P out in the 2nd round? I wanna sample this and find out.
I almost forgot what thread I'm in. Obviously the first hand I gotta test is S1-R1 vs QH upcard:

(Card_A-H) (Card_K-H) (Card_9-H) (Card_9-D) (Card_9-S)

So many samples I wanna do, so little time. Goddamn. If I broke up with my gf I'd have more time...... :|

Richardb02
Posts: 748
Joined: Mon Dec 10, 2018 8:57 pm
Location: Florida

Unread post by Richardb02 » Sun Mar 14, 2021 9:41 pm

Please humor me, let’s simplify, summarize & verify:
____________________Richard________Wes___Irish______Tbolt
Result of Ordering_____-.13_________-.1____ -.1_____Pass as a stop-loss

Result of passing_______NR_______-.5886____-.53?________NR
*NR= No result or no reply.

I deduced Irish’s Result of Passing from, 0.40 to 0.46 EV of Ordering vs. Passing and then using -.43 as the average and netting out the -.53. Did I reflect your numbers correctly Irish?

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Sun Mar 14, 2021 10:47 pm

This is to Wes. Wow, your words has elevated your status in my eyes, big time. It takes a real person to express your feeling in such a manner. Okay, I feel vindicated.

But now I think I have helped to unleash some hidden statistical weapons that were not there before the controversial hand. First was the combinatorics you started using, and now some knowledge on statically testing hands and theories. Oh will it make you a more powerful euchre player. Now I am scared. lol

It far more powerful than any simulator because you never know how good the programmer is/was and the difficulty of setting those hands. I know because I argued or discussed with Sword, Freddy, the author of his book & simulator. I pointed out issues on hands he did ES. His simulator was useful but with some issues.

So now, Wes, this is like a new toy, to explore and utilize like no other. Now you should never ponder on any hand which is the best direction to take. Even though there are variables you will not have 100% assurance because what players do or not do is more unpredictable than the cards themselves and has to be factored in. The cards themselves have few secrets, and are so predictable (probable that is). However players are unpredictable and probability is more difficult. It's what makes euchre so interesting.

Players are like the old song by Conway Twitty, Don't Call Him a Cowboy Until You See Him Ride (or Don't call him a euchre player until you see him play).

If you hand that you want me to test, I could help. And thanks again for you honesty.

And I was thinking about Don. I am glad he did not lock us out again, but was most like getting ready to. Thoughts crossed my mind, him thinking, WHAT DO I DO ABOUT THESE TWO, LOL? But it's okay Don, thanks for being patient with us. More patience than I might have had in your shoes. So we will approach hands or discussions a little differently, I think.

Irishwolf

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Mon Mar 15, 2021 5:09 am

irishwolf wrote:
Sun Mar 14, 2021 10:47 pm
This is to Wes. Wow, your words has elevated your status in my eyes, big time. It takes a real person to express your feeling in such a manner. Okay, I feel vindicated.
Thx man. I still feel disappointed in my behavior, but there's nothing I can do. I can't go back and change the past. I f**ked up. But I'll never be disappointed in being proven wrong. I love that. To me that's exciting. Being right is boring. When I'm right my reality doesn't change. But when I realize I'm wrong about something, then I can make improvements. I do try to challenge my own assumptions, and prove myself wrong, but that only gets one so far. The process of improvement gets sped up big time when there's others out there that can challenge me and prove me wrong.
irishwolf wrote:
Sun Mar 14, 2021 10:47 pm
But now I think I have helped to unleash some hidden statistical weapons that were not there before the controversial hand. First was the combinatorics you started using, and now some knowledge on statically testing hands and theories. Oh will it make you a more powerful euchre player. Now I am scared. lol
I've told Edward many times, yes we're really good players but it's those grey areas that haunt me, and without the math to attack those grey areas our growth as players is inherently limited. And now I'm very excited becuz now I have the full tool kit to figure things out and make some breakthroughs in the grey areas--as you mentioned, the combinatorics math and knowing how to properly test hands statistically. On the latter part I never realized it was that easy becuz I never realized how fast one can reach a 95% confidence interval at the kitchen table. I erroneously thought it would take tens of thousands of hands, perhaps 100s of thousands, maybe even millions. I didn't know, I just falsely assumed a kitchen table sample was for all practical purposes undoable becuz it would take essentially forever. Then I come across your hand, QhTh9hAsKs, and in 60 hands I cross the 95% threshold. I can do around 20 hands per hour. That's only 3 hours of work to reach a very strong conclusion! Meaning many of these so-called grey area spots can be solved in a matter of hours! That's a lot less than forever! Even in the worst case scenario, I bet the longest it will take me to solve a spot is a month, and that's assuming a lazy pace of doing only 10-20 samples a day. A month is nothing to me.
irishwolf wrote:
Sun Mar 14, 2021 10:47 pm
It far more powerful than any simulator because you never know how good the programmer is/was and the difficulty of setting those hands. I know because I argued or discussed with Sword, Freddy, the author of his book & simulator. I pointed out issues on hands he did ES. His simulator was useful but with some issues.
It would be nice tho to have a good simulator to save time grinding away at the kitchen table, but I don't mind the grind. Put on some good music in the background and it's almost meditative.
irishwolf wrote:
Sun Mar 14, 2021 10:47 pm
So now, Wes, this is like a new toy, to explore and utilize like no other. Now you should never ponder on any hand which is the best direction to take. Even though there are variables you will not have 100% assurance because what players do or not do is more unpredictable than the cards themselves and has to be factored in. The cards themselves have few secrets, and are so predictable (probable that is). However players are unpredictable and probability is more difficult. It's what makes euchre so interesting.
I agree to an extent and that's why I am so excited but I do see some limitations to a kitchen table sample approach. Firstly, the good news is I think we can effectively solve all first round decisions using kitchen table samples. That's HUGE. That's an amazing fact to me. BUT solving 2nd rd decisions seems unfeasible to me. For example, let's say I wanted to figure out if a certain super marginal diamond Next call was better than passing assuming we don't block reverse Next. Well to get to the 2nd round with a distribution that makes sense I'd have to give the dealer a heart upcard, deal out the cards, and scrap everything and reshuffle every time someone had a legit heart call. That seems too onerous to me. That's where we really need a good simulator I think.

Another issue that may come up is it's not necessarily clear how to interpret the data. For example, lets say I test out a certain donating hand, and my kitchen table sample shows that donating has an EV of -.2 suggesting we should NOT donate with that hand. But is it that simple? It is as simple as "It's -EV therefore don't do it". Perhaps it's worth paying a small cost to donate as a way of managing some variance, I.E. stopping some 4 pt loners from happening. Basically like insurance. Same idea for marginal Next calls. I don't know the answer to that question. I don't even know if the question itself is valid.
irishwolf wrote:
Sun Mar 14, 2021 10:47 pm
Players are like the old song by Conway Twitty, Don't Call Him a Cowboy Until You See Him Ride (or Don't call him a euchre player until you see him play).
I feel like the variation in the player pool won't hinder us on this front. There's basically two types of players. 1) strong aggressive players that play the situation which make up 1-5% of the population, and 2) players that just play their 5 cards trying not to get euchred which make up 95-99% of the euchre population. To the extent that it's necessary it would be easy to adjust for player styles in our kitchen table simulation.
irishwolf wrote:
Sun Mar 14, 2021 10:47 pm
If you hand that you want me to test, I could help. And thanks again for you honesty.
Right now I'm working on the hand in the OP:

(Card_A-H) (Card_K-H) (Card_9-H) (Card_9-S) (Card_9-D)

Vs a (Card_Q-H) upcard, S1-R1.

Edward was the one who played that hand. I was his P. Every time Edward passes that type of hand it puts me on tilt. I've been trying to convince Edward to call in this spot for a long time now. So I'm real fired up about this one. And it's not really cuz I wanna be right or I want him to be wrong. It's really about the fact that Edward is the closest I'll ever get in real life to having the kind've chemistry I strongly desire in a euchre partner. And when Edward makes plays I vehemently disagree with it f**ks up that chemistry. Ideally I want our partnership to have zero leaks so every time we play another team we're truly playing at our best, not unnecessarily burning points in any spot. Basically I want me and Edward to be the best f**king partnership in the motherf**king world!!! But we got quite a few spots to clean up first, not just spots where he's wrong, but spots I'm wrong in too.

Anyways, I've done 30 hands so far. Here's my data set (each number represents the EV of calling):
-1,-1,3,3,-1,-1,3,-1,-1,3,0,-1,0,-1,0,-1,-1,0,3,1,3,0,-1,0,3,-1,-1,0,1,-1
My standard deviation is: 1.6220463873634
My mean is: .30
My P-value is: .3194

IOW so far the EV of calling is +.3, but obviously I have a ways to go before I reach that 95% confidence interval. But so far I like the chances of calling being correct. Funny thing about that sample: 70% of the time passing does better or the same as calling, and YET calling is still coming out ahead of passing! Another counterintuitive mindf**k spot. It's those 3s that are messing things up for the pass strategy. I.E. those times S1 passes up a heart call that would've scored a point, and then it passes around and S1 calls Next and gets euchred.
irishwolf wrote:
Sun Mar 14, 2021 10:47 pm
And I was thinking about Don. I am glad he did not lock us out again, but was most like getting ready to. Thoughts crossed my mind, him thinking, WHAT DO I DO ABOUT THESE TWO, LOL? But it's okay Don, thanks for being patient with us. More patience than I might have had in your shoes. So we will approach hands or discussions a little differently, I think.

Irishwolf
Lol I was thinking the same thing. I was like "man I hope he doesn't lock this thread too." It goes without saying I'm eternally glad he didn't. And I'm also glad you never gave up on that discussion.

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Mon Mar 15, 2021 12:28 pm

I would be interested in the results you get for the Hand AH KH 9D 9D 9S, WITH QH up.

i some work on it and got a euchre of 43% +/-3%. Sweep 8 to 11% and 1 pointers which gives -86 - ( + 18 + 49) = (EV) -.19 compared to the -.10 you projected. S1 ordering up the QH. I also played with S1 always leading the KH to the first trick (it makes a difference). You know there is variability as to trusting tricks to S3 which must happen to score a point.

However, what the points would be when S1 passes (a night mare) be extremely difficult and variable. You gave a -.58 and I think closer to -.40 to -.45. A range because not only the variability of the cards but how players might play the hand. Not so sure you can zero in on a correct number other than it is a greater negative number less favorable for S1 passing. Actually, does it really matter that much? I am pretty much done with that hand.

As far as S1 having a hand like AH KH 9H AS 9S at S1 vs S3 (order vs pass). Now that is much more interesting decision to me. No next and diamonds and clubs will be loaded somewhere (S2,S3 or S4). S3 now has two aces to assist with any call from S1. Lots to think about. I do not know for sure, but I am going to find out as projects on my list?
Any discussion on that, start a new thread.

~Irishwolf

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Tue Mar 16, 2021 6:23 am

(Card_A-H) (Card_K-H) (Card_9-H) (Card_9-S) (Card_9-D)

Vs a (Card_Q-H)

Update: Now at 70 hands. Here's my data set (each number represents the EV of calling):
-1,-1,3,3,-1,-1,3,-1,-1,3,0,-1,0,-1,0,-1,-1,0,3,1,3,0,-1,0,3,-1,-1,0,1,-1,3,0,0,0,0,-1,-1,-1,-1,0,0,3,3,0,0,0,-1,-1,1,0,0,1,-1,-1,3,0,-1,0,-1,-1,3,-1,3,0,-4,1,3,-1,0,3
My standard deviation is: 1.6054
My mean is: .2714
My P-value is: .1617

The EV of calling is now at .2714. Still have a ways to go to get to a 95% confidence level.

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Tue Mar 16, 2021 10:36 am

When you get to about 135 hands you will be at 95%. Your mean might change but not more than +/- 3% when you get there.

So we did our hands a little differently, I summed each 25 set and got -.19 on ordering before doing anything if S1 passed.

If I add your mean vs Ho = 0 (-.27 + -.19 = -.46 if S1 passed). I think comparable results doing this. I think I can do that. I mentioned that I thought the real number was lower than - .58. It will be interesting or me to see. I did not want to spend more time as I know there is a statistically significant difference between ordering vs passing. (Too many other projects to do.)

~Irishwolf

User avatar
LeftyK
Posts: 261
Joined: Wed Jun 10, 2020 9:45 am
Location: North Carolina

Unread post by LeftyK » Tue Mar 16, 2021 7:25 pm

You guys know about that euchre sim that's called "Euchre Challenge and Teacher" by Fred Benjamin? It does 1000 hand sims. You can install it on VDI and reload the VDI once the trial period ends. You guys are making this way to difficult :) - now I'm giving away trade secrets :) but It's buried in this thread.

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Thu Mar 18, 2021 4:32 am

irishwolf wrote:
Mon Mar 15, 2021 12:28 pm
I would be interested in the results you get for the Hand AH KH 9D 9D 9S, WITH QH up.

i some work on it and got a euchre of 43% +/-3%. Sweep 8 to 11% and 1 pointers which gives -86 - ( + 18 + 49) = (EV) -.19 compared to the -.10 you projected. S1 ordering up the QH. I also played with S1 always leading the KH to the first trick (it makes a difference). You know there is variability as to trusting tricks to S3 which must happen to score a point.
My sample is now at 100 hands. My euchre rate for calling is 41%, sweep rate is 12%, and 1 pt rate is 47%. And my EO of calling is -.11. From this configuration I always lead the AH whether I call or my opponents call in the first round as that's how I would actually play it. I'm not worried about the "variability as to trusting tricks to S3" as I just play the hand the way I would normally play it, throwing off in spots I normally throw off. Here's my updated numbers:
-1,-1,3,3,-1,-1,3,-1,-1,3,0,-1,0,-1,0,-1,-1,0,3,1,3,0,-1,0,3,-1,-1,0,1,-1,3,0,0,0,0,-1,-1,-1,-1,0,0,3,3,0,0,0,-1,-1,1,0,0,1,-1,-1,3,0,-1,0,-1,-1,3,-1,3,0,-4,1,3,-1,0,3,-1,0,0,0,3,3,-1,4,0,0,0,-1,0,1,0,-1,-1,3,-1,0,-1,0,0,0,-1,1,3,-1,-4,-1
My standard deviation is: 1.6134
My mean is: .23
My P-value is: .1571

The EV of calling is now at .23. Still have a ways to go to get to a 95% confidence level.

Note: If you notice there are two -4 data points in my sample. You may be wondering wtf happened there? Well in both hands a call got euchred in the first round (-2), but if I had passed S4 goes alone with JdQhThAsAc and now I don't lead trump, and my P had the Right in both cases so a loner gets euchred (+2). This hand is controversial because most people aren't going alone with it (I would tho). Had S4 just called with this hand I go back to the "lead AH" strategy and he makes a point in both cases instead of getting euchred, so instead of calling being -4 for both hands it would end up being -1. That's a 6 pt swing in favor of the pass strategy, making my result more conservative than reality in most cases.

Here's what my numbers look like if I change both those -4s into -1s:

My standard deviation is: 1.5062
My mean is: .29
My P-value is: .0571

So calling would be +.29 EV and I'd be almost at the 95% confidence interval.
irishwolf wrote:
Mon Mar 15, 2021 12:28 pm
However, what the points would be when S1 passes (a night mare) be extremely difficult and variable.
It's actually not that hard to do. If S2 or S4 or S3 calls in the first round, I'm always leading the AH and taking it from there. If it gets passed around, I'm calling Next and always leading my one trump (9D). The rest plays itself. A little more work would have to be done if we had S1 passing in the 2nd rd instead of calling Next but it's still manageable as I'm still gonna play every spot as if I'm in that spot.

One could argue that me playing every spot as myself is a potential weakness of the sample as that certainly doesn't reflect reality but in this case they would be wrong. The more aggressive the game is the better the pass strategy would do as the value of S1 bagging goes up in the first round. So me playing every spot as myself actually gives me a more conservative result, a result that's biased towards the passing strategy. To still get an EV that says calling is the best play anyways really tells us something here.
irishwolf wrote:
Mon Mar 15, 2021 12:28 pm
You gave a -.58 and I think closer to -.40 to -.45. A range because not only the variability of the cards but how players might play the hand. Not so sure you can zero in on a correct number other than it is a greater negative number less favorable for S1 passing. Actually, does it really matter that much? I am pretty much done with that hand.
It's looking like my estimates for calling (euchre rate 40%, 1 pt rate 50%, 2 pt rate 10%, E0 = -.10) are pretty close to dead on. But my estimates for the cost of passing are off (My EO for passing is -.34 or -.4 if S4 doesn't go alone in those 2 hands), but as you noted, not off enough to change what looks like the inevitable conclusion so far: calling is better than passing with this hand from S1-R1. Still gotta keep doing this sample. Not gonna quit til I reach that 95% confidence interval.
irishwolf wrote:
Mon Mar 15, 2021 12:28 pm
As far as S1 having a hand like AH KH 9H AS 9S at S1 vs S3 (order vs pass). Now that is much more interesting decision to me. No next and diamonds and clubs will be loaded somewhere (S2,S3 or S4). S3 now has two aces to assist with any call from S1. Lots to think about. I do not know for sure, but I am going to find out as projects on my list?
Any discussion on that, start a new thread.

~Irishwolf
I'm wondering if we should just have a general thread for hands we test + perhaps other misc math/combinatorics topics.

User avatar
Dlan
Site Admin
Posts: 672
Joined: Thu Apr 19, 2018 10:08 pm
Location: Ohio

Unread post by Dlan » Thu Mar 18, 2021 10:39 am

When talking about different hands, please start a new thread. If need be, posting a link back to an early thread is always possible.

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sun Mar 21, 2021 2:28 am

Update on the S1-R1 (Card_A-H) (Card_K-H) (Card_9-H) (Card_9-S) (Card_9-D)

vs (Card_Q-H) upcard hand:

Sample size: 120
-1,-1,3,3,-1,-1,3,-1,-1,3,0,-1,0,-1,0,-1,-1,0,3,1,3,0,-1,0,3,-1,-1,0,1,-1,3,0,0,0,0,-1,-1,-1,-1,0,0,3,3,0,0,0,-1,-1,1,0,0,1,-1,-1,3,0,-1,0,-1,-1,3,-1,3,0,-4,1,3,-1,0,3,-1,0,0,0,3,3,-1,4,0,0,0,-1,0,1,0,-1,-1,3,-1,0,-1,0,0,0,-1,1,3,-1,-4,-1,-1,0,-1,3,3,-1,0,3,0,-1,-1,0,-1,3,0,3,-1,-1,3,-1
My standard deviation is: 1.6333
My mean is: .2666
My P-value is: .0763

The EV of calling is now at .2666. Almost at a 95% confidence interval.

Other stats:

Calling gets euchred: 49/120 = 40.83%
Calling gets 1pt: 58/120 = 48.33%
Calling gets sweep: 13/120 = 10.83%
EO of calling: -.1166
EO of passing: -.3833

Richardb02
Posts: 748
Joined: Mon Dec 10, 2018 8:57 pm
Location: Florida

Unread post by Richardb02 » Sun Mar 21, 2021 8:44 am

Wes (aka the legend) wrote:
Sun Mar 21, 2021 2:28 am
Update on the S1-R1 (Card_A-H) (Card_K-H) (Card_9-H) (Card_9-S) (Card_9-D)

vs (Card_Q-H) upcard hand:

Sample size: 120
-1,-1,3,3,-1,-1,3,-1,-1,3,0,-1,0,-1,0,-1,-1,0,3,1,3,0,-1,0,3,-1,-1,0,1,-1,3,0,0,0,0,-1,-1,-1,-1,0,0,3,3,0,0,0,-1,-1,1,0,0,1,-1,-1,3,0,-1,0,-1,-1,3,-1,3,0,-4,1,3,-1,0,3,-1,0,0,0,3,3,-1,4,0,0,0,-1,0,1,0,-1,-1,3,-1,0,-1,0,0,0,-1,1,3,-1,-4,-1,-1,0,-1,3,3,-1,0,3,0,-1,-1,0,-1,3,0,3,-1,-1,3,-1
My standard deviation is: 1.6333
My mean is: .2666
My P-value is: .0763

The EV of calling is now at .2666. Almost at a 95% confidence interval.

Other stats:

Calling gets euchred: 49/120 = 40.83%
Calling gets 1pt: 58/120 = 48.33%
Calling gets sweep: 13/120 = 10.83%
EO of calling: -.1166
EO of passing: -.3833
I am diligently following this post, but I am missing something. Thank you for keeping this post’s discussion to the OP, it helps immensely!

First, I have a suggestion:
Wes, you use EO, I suggest EVc, EV of calling, as a generic and more easily understood acronym for Expected Value of calling. (BTW, using my BPS analysis, I estimate: Getting euchred 41% of the time [incredible agreement], 59% success breaking down to 49% 1 point and 10% 2 point sweeps [incredible agreement], so EV= .41*-2+.49*1+.10*2=-.13 [incredible agreement vs -.1166]).
Wes recommended EV (relative value of ordering vs passing), I recommend EVr, Estimated Value relative.
Continuing my recommended convention, I recommend EVp (Estimated Value of passing)

So I conclude EVr= EVc(EOcalling)-EVp(EVpassing)=EV(EVrelative)
Therefore: -.1166(EVc) - .3833(EVpassing)=-.4999

But, Wes’ test hands yields EV(EVr) of .2666
I would expect those calculations to correlate. Obviously they don’t correlate. Where am I going wrong?


irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Sun Mar 21, 2021 10:23 am

However, what the points would be when S1 passes (a night mare) be extremely difficult and variable. So I played it differently as the average euchre player, even better players are NOT going to call Next (diamonds) with only the 9D. A worst situation (not reality) than ordering, obviously. I had S1 passing 2R. That is where it becomes more complex in points scored for and against. I only had S2 calling with a reasonable hand, crossing suit or next. Then he passed accordingly and S3 can call what he wants or passes. Bottom line, it is still more favorable for S1 to order. Points will be more Negative calling next than S1 passing is my point which is what most players will do.

WES SAID, "It's actually not that hard to do. If it gets passed around, I'm calling Next (S1) and always leading my one trump (9D). The rest plays itself. A little more work would have to be done if we had S1 passing in the 2nd rd instead of calling Next but it's still manageable as I'm still gonna play every spot as if I'm in that spot."

~Irishwolf

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sun Mar 21, 2021 6:59 pm

Richardb02 wrote:
Sun Mar 21, 2021 8:44 am
So I conclude EVr= EVc(EOcalling)-EVp(EVpassing)=EV(EVrelative)
Therefore: -.1166(EVc) - .3833(EVpassing)=-.4999

But, Wes’ test hands yields EV(EVr) of .2666
I would expect those calculations to correlate. Obviously they don’t correlate. Where am I going wrong?

[/color]
You forgot a negative sign. It's not (-.1166 - .3833). It's (-.1166 - -.3833).

Negative .1166 minus negative .3833 = +.2666

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sun Mar 21, 2021 7:35 pm

irishwolf wrote:
Sun Mar 21, 2021 10:23 am
However, what the points would be when S1 passes (a night mare) be extremely difficult and variable. So I played it differently as the average euchre player, even better players are NOT going to call Next (diamonds) with only the 9D. A worst situation (not reality) than ordering, obviously. I had S1 passing 2R. That is where it becomes more complex in points scored for and against. I only had S2 calling with a reasonable hand, crossing suit or next. Then he passed accordingly and S3 can call what he wants or passes. Bottom line, it is still more favorable for S1 to order. Points will be more Negative calling next than S1 passing is my point which is what most players will do.
I agree with everything above and I love that you tested this out with average players and still got the same result (calling is better than passing). I don't really wanna run that test right now. I got other things I'm working on :)

I know the 2nd round is messy in this instance, but really all we have to do is hammer out what are assumptions are of what an average player will do.

For example, average players play not to get euchred. They need an actual hand to call. Say the dealer turns down a 9C in the 1st round. An average player--whose primary concern is not getting euchred--is not calling this diamond hand from the 2S-2Rd:

(Card_K-D) (Card_10-D) (Card_A-C) (Card_10-S) (Card_9-S)

I'm sure we both agree on that. And once we got these type of assumptions in place on what an average player is calling and passing, the 2nd rd ceases to be messy and becomes automatic. So my contention would be the 2nd rd is relatively easy to orchestrate with expert players and typical average players. I suppose it can get messy with above average good players who get some concepts but still have many holes in their game. Predicting what they'll do can be problematic but this subset isn't large enough to matter imo. I really think all we really need is a way to test how a call/pass decision will perform in a tough game and in a typical average game.

For fun here's my best guesses at what a typical player will call in S1 after the dealer passes:

Calling hands:

R+2 in any suit
R+1+A in Next but probably passing otherwise
L+2+A in any suit
L+2 in any suit?
L+1+2A/3A in Next but probably passing otherwise
3 low trump in Next, but passing otherwise
3 low trump + an off Ace, calling in any suit

Same type of ranges for S2-2nd rd. Just change Next to reverse Next.

PS: Wolf I can't make the Monday night game tomorrow. Can you take my spot?

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Thu Mar 25, 2021 1:55 am

Finally reached the 95% confidence interval so I'm done with this hand.

Update on the S1-R1 (Card_A-H) (Card_K-H) (Card_9-H) (Card_9-S) (Card_9-D)

vs (Card_Q-H) upcard hand:

Sample size: 151
-1,-1,3,3,-1,-1,3,-1,-1,3,0,-1,0,-1,0,-1,-1,0,3,1,3,0,-1,0,3,-1,-1,0,1,-1,3,0,0,0,0,-1,-1,-1,-1,0,0,3,3,0,0,0,-1,-1,1,0,0,1,-1,-1,3,0,-1,0,-1,-1,3,-1,3,0,-4,1,3,-1,0,3,-1,0,0,0,3,3,-1,4,0,0,0,-1,0,1,0,-1,-1,3,-1,0,-1,0,0,0,-1,1,3,-1,-4,-1,-1,0,-1,3,3,-1,0,3,0,-1,-1,0,-1,3,0,3,-1,-1,3,-1,-1,3,0,-1,0,-1,-1,-1,-1,-1,-1,-1,3,3,0,0,0,0,-1,-1,1,-1,3,-1,-1,3,0,3,3,-1,3
My standard deviation is: 1.6377
My mean is: .2781
My P-value is: .0386

The EV of calling is .2781. Calling with this hand is better than passing.

Other stats:

Calling gets euchred: 60/151 = 39.74%
Calling gets 1pt: 77/151 = 50.99%
Calling gets sweep: 14/151 = 9.27%
EO of calling: -.0993
EO of passing: -.3775

The next hand I'm thinking about testing, basically the worst possible 3 trump hand in this spot (becuz I'm not calling vs a Jack upcard unless up 9-6/9-7 or 9-9):

S1-R1

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_9-S) (Card_9-C)

Vs (Card_A-H) upcard.

In the 2nd Rd S1 is passing.

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Thu Mar 25, 2021 2:05 am

Wes (aka the legend) wrote:
Thu Mar 18, 2021 4:32 am
Note: If you notice there are two -4 data points in my sample. You may be wondering wtf happened there? Well in both hands a call got euchred in the first round (-2), but if I had passed S4 goes alone with JdQhThAsAc and now I don't lead trump, and my P had the Right in both cases so a loner gets euchred (+2). This hand is controversial because most people aren't going alone with it (I would tho). Had S4 just called with this hand I go back to the "lead AH" strategy and he makes a point in both cases instead of getting euchred, so instead of calling being -4 for both hands it would end up being -1. That's a 6 pt swing in favor of the pass strategy, making my result more conservative than reality in most cases.

Here's what my final numbers look like if I adjust these two outlier -4 hands to -1s.

My standard deviation is: 1.5679
My mean is: .3179
My P-value is: .0138

EO of calling: -.0993
EO of passing: -.4172
EV of calling: +.3179

I think it's safe to say that calling has an +EV of around .3 pts in the long run.

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Thu Mar 25, 2021 4:28 am

Not looking good so far for the pass-pass strategy for this hand.

S1-R1

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_9-S) (Card_9-C)

Vs (Card_A-H) upcard.

In the 2nd rd S1 is passing.

Other rules: When S1 calls or S2/S3/S4 call in the first rd, S1 always leads trump. Against a R1-S2/S4 loner S1 always leads the 9S. Against a R2-S2/S4 club/diamond call, S1 always leads the 9S. Against a R2-S2/S4 spade call S1 always leads a club. In the 2nd rd I'm playing S2/S3/S4 as if I was in that seat.

Sample size: 20
3,-1,2,-1,3,-1,3,0,-1,-1,2,2,-1,3,-1,2,-1,0,-1,-1
Standard deviation is: 1.7321
Mean is: .5
P-value is: .2122

EO of calling: -.75
EO of passing: -1.25
EV of calling: +.5

Calling gets euchred: 12/20 = 60%
Calling gets 1pt: 7/20 = 35%
Calling gets sweep: 1/20 = 5%

I know it's just a sample size of 20 but these numbers are pretty staggering. A S1 call is getting euchred 60% of the time and YET it still has a +EV of .5. One might guess that the pass-pass strategy has led to a lot of 2 seat 2nd rd 4 pt loners or something? Nope. There is not one 2nd rd 4 pt loner in this sample. There is a 1st Rd 4pt loner from S4 when he went alone with AhJhJdAsKs and S1's obligatory 9S lead leads to 4 pts. So the pass-pass strategy did suffer a 2 pt deficit vs calling on this one, but that's the only loner so far. Coincidently, S4 had another loner that S1's 9S lead effectively stops by forcing out a trump (JhAhKhAcQc). S4 had the unguarded Left in that hand that would've obviously been pulled on S4's first trump lead.

The big thing that's going on so far is after S1 passes in the 2nd round, S2-S4's team hasn't gotten euchred once. That's not a big surprise tho as it's very hard to euchre a team when S1 has such a bad 2nd rd hand. Where S1 is running bad is so far passing has not led to one first rd S2/S4 euchre. The other key trend one can see just by looking at the numbers in the sample box: When calling is "wrong" it's usually only costing S1 -1 pt, but when pass-pass is "wrong" it's costing multiple pts. Becuz of this dynamic it's going to be really hard for the pass-pass strategy to beat out calling.

Even tho this sample size is small so far, I would venture to guess that most of the above trend is going to continue. I'm feeling very good right now about the hypothesis that one should always call with 3 trump from S1-R1 if they have no where to go in the 2nd rd (except when up 9-8).

Note: My definition of "no where to go" is S1 has 1 trump or less in Next (IOW if I had QhTh9hTd9d I'd pass and call Next) and S1 does not have R+1 in reverse Next (If I had QhTh9hJc9c I'm passing in the 1st Rd with the intention of calling clubs) and S1 does not block Reverse Next for if I blocked reverse Next with a hand like QhTh9hJc9s I would not make this marginal call in the 1st rd.

It goes without saying I'm gonna continue this sample until I reach a 95% confidence interval.

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Thu Mar 25, 2021 1:26 pm

Yes an interesting hand and initial results. I think I will jump in and do some hands myself (with your criteria) to see what happens.

Irishwolf

Not looking good so far for the pass-pass strategy for this hand.

S1-R1

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_9-S) (Card_9-C)

Vs (Card_A-H) upcard.

In the 2nd rd S1 is passing.

Other rules: When S1 calls or S2/S3/S4 call in the first rd, S1 always leads trump. Against a R1-S2/S4 loner S1 always leads the 9S. Against a R2-S2/S4 club/diamond call, S1 always leads the 9S. Against a R2-S2/S4 spade call S1 always leads a club. In the 2nd rd I'm playing S2/S3/S4 as if I was in that seat.

Tbolt65
Posts: 820
Joined: Sun Jun 16, 2019 9:14 pm
Location: Las Vegas

Unread post by Tbolt65 » Thu Mar 25, 2021 1:35 pm

Wes (aka the legend) wrote:
Thu Mar 25, 2021 4:28 am
Not looking good so far for the pass-pass strategy for this hand.

S1-R1

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_9-S) (Card_9-C)

Vs (Card_A-H) upcard.

In the 2nd rd S1 is passing.

Other rules: When S1 calls or S2/S3/S4 call in the first rd, S1 always leads trump. Against a R1-S2/S4 loner S1 always leads the 9S. Against a R2-S2/S4 club/diamond call, S1 always leads the 9S. Against a R2-S2/S4 spade call S1 always leads a club. In the 2nd rd I'm playing S2/S3/S4 as if I was in that seat.

Sample size: 20
3,-1,2,-1,3,-1,3,0,-1,-1,2,2,-1,3,-1,2,-1,0,-1,-1
Standard deviation is: 1.7321
Mean is: .5
P-value is: .2122

EO of calling: -.75
EO of passing: -1.25
EV of calling: +.5

Calling gets euchred: 12/20 = 60%
Calling gets 1pt: 7/20 = 35%
Calling gets sweep: 1/20 = 5%

I know it's just a sample size of 20 but these numbers are pretty staggering. A S1 call is getting euchred 60% of the time and YET it still has a +EV of .5. One might guess that the pass-pass strategy has led to a lot of 2 seat 2nd rd 4 pt loners or something? Nope. There is not one 2nd rd 4 pt loner in this sample. There is a 1st Rd 4pt loner from S4 when he went alone with AhJhJdAsKs and S1's obligatory 9S lead leads to 4 pts. So the pass-pass strategy did suffer a 2 pt deficit vs calling on this one, but that's the only loner so far. Coincidently, S4 had another loner that S1's 9S lead effectively stops by forcing out a trump (JhAhKhAcQc). S4 had the unguarded Left in that hand that would've obviously been pulled on S4's first trump lead.

The big thing that's going on so far is after S1 passes in the 2nd round, S2-S4's team hasn't gotten euchred once. That's not a big surprise tho as it's very hard to euchre a team when S1 has such a bad 2nd rd hand. Where S1 is running bad is so far passing has not led to one first rd S2/S4 euchre. The other key trend one can see just by looking at the numbers in the sample box: When calling is "wrong" it's usually only costing S1 -1 pt, but when pass-pass is "wrong" it's costing multiple pts. Becuz of this dynamic it's going to be really hard for the pass-pass strategy to beat out calling.

Even tho this sample size is small so far, I would venture to guess that most of the above trend is going to continue. I'm feeling very good right now about the hypothesis that one should always call with 3 trump from S1-R1 if they have no where to go in the 2nd rd (except when up 9-8).

Note: My definition of "no where to go" is S1 has 1 trump or less in Next (IOW if I had QhTh9hTd9d I'd pass and call Next) and S1 does not have R+1 in reverse Next (If I had QhTh9hJc9c I'm passing in the 1st Rd with the intention of calling clubs) and S1 does not block Reverse Next for if I blocked reverse Next with a hand like QhTh9hJc9s I would not make this marginal call in the 1st rd.

It goes without saying I'm gonna continue this sample until I reach a 95% confidence interval.

The question I would interject into your thought process here is when or if it is ever good to pass twice aside from the math. What situations could warrant a pass-pass?

Something to ponder and think about.
Tbolt65
Edward

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Thu Mar 25, 2021 7:59 pm

Having had more time to ponder. IMO, I see this as three sets or options:

I. S1 orders the AH (I agree in looking at this, that the negative EV will be around -0.6 (+/-5). Euchre rate about 55%, my estimate!

II. When S4 passes, S1 calls something, with Diamonds being a better call?? But that also will be negative and I don't want to guess at this point (30 to 40% S3 might have next, not considering bagging at S4 unless a stronger hand than H's). I think worth separating out and doing.

III. When S4 passes, S1 always passes. S2, S3 & S4 (STD for kicks). But if S2 calls, I am leading QH. I suspect the points will also be negative but very similar to the other hand of AH KH 9H 9D 9S. Don't even think worth spending the time on that option.

~IRISHWOLF

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Thu Mar 25, 2021 10:35 pm

I can tell you with confidence, S1 calling next is worse than Ordering the AH.

"The question I would interject into your thought process here is when or if it is ever good to pass twice aside from the math. What situations could warrant a pass-pass?"

~IRISHWOLF

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Fri Mar 26, 2021 4:34 am

Tbolt65 wrote:
Thu Mar 25, 2021 1:35 pm
The question I would interject into your thought process here is when or if it is ever good to pass twice aside from the math. What situations could warrant a pass-pass?

Something to ponder and think about.
Tbolt65
Edward
I just figured the "pass-call next" strategy would do worse than "pass-pass". I'm trying to bias my sample in favor of the "passing in the first round" strategy becuz that's the strategy I'm trying to refute. If I'm wrong about that, it still shouldn't matter. IOW, if calling here outperforms "pass-pass" as I predict, it's not like "pass-call next" with no trump, no aces is gonna be a gamechanger. If calling outperforms "pass-pass" it will outperform "pass-call next" as both "pass-pass" and "pass-call next" burn tons of points in the 2nd rd. Pass-call next is no panacea. As said, even if I'm wrong that "pass-pass" will do better than "pass-call next" the difference won't be large enough to matter.

That said in real life EVEN IF "pass-pass" did better than "pass-call Next" I would still choose the latter strategy if I were in that spot and the score wasn't up 9-8, tied 8-8. And I can be in that spot becuz against a Jack upcard I'm not ordering except at 9-9, 9-7/9-6. So if S4 passes that Jack I'm now in that spot which begs the question in this hypothetical: why would I still go against the math and call Next with no trump no aces in that spot?!?! Am I irrational? The answer is I am willing to take a few slight EV losses to protect the integrity of my range. I always want my R2-S1 pass to = I have reverse Next blocked so my P can properly read my hand. This allows my P to go alone much weaker in reverse Next should S2 pass, and should S2 call reverse Next, my P can take more optimal lines to euchre S2 becuz he knows what I hold.

Some examples: Dealer turns down the 9D and I pass from S1-Rd 2 and S2 passes, you're in S3 with:

(Card_A-H) (Card_J-H) (Card_A-S) (Card_J-S) (Card_A-D)

Knowing my range (I have black blocked but I'm not guaranteed to have anything in Next) you have a very easy Spade loner call.

Same goes for this hand in the same situation:

(Card_J-C) (Card_10-S) (Card_9-S) (Card_A-D) (Card_Q-D)

Becuz you know what I have, this apparent marginal spade loner is now a premium, so go for it.

One more example where S2 calls reverse Next. Ok I pass, S2 calls clubs and you have:

(Card_J-C) (Card_A-H) (Card_9-S) (Card_10-D) (Card_Q-D)

I lead the (Card_9-H)

S2 follows suit, you play the AH and S4 follows suit so you take the trick, now knowing my hand you know exactly what to lead next. Lead the Right and we have an instant euchre since you know I have a guaranteed trick in clubs. This is next level chemistry that I bet no euchre team in the world possesses. I don't wanna threaten that chemistry.

So assuming "pass-next" doesn't do as well as "pass-pass" I'd still call Next assuming the difference isn't too large, so you can always perfectly read my hand when I do pass. There's mathematical value in that too, but obviously it's too hard to quantify.

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Fri Mar 26, 2021 4:50 am

irishwolf wrote:
Thu Mar 25, 2021 7:59 pm
Having had more time to ponder. IMO, I see this as three sets or options:

I. S1 orders the AH (I agree in looking at this, that the negative EV will be around -0.6 (+/-5). Euchre rate about 55%, my estimate!

II. When S4 passes, S1 calls something, with Diamonds being a better call?? But that also will be negative and I don't want to guess at this point (30 to 40% S3 might have next, not considering bagging at S4 unless a stronger hand than H's). I think worth separating out and doing.

III. When S4 passes, S1 always passes. S2, S3 & S4 (STD for kicks). But if S2 calls, I am leading QH. I suspect the points will also be negative but very similar to the other hand of AH KH 9H 9D 9S. Don't even think worth spending the time on that option.

~IRISHWOLF
Yeah ideally we'd know for sure which strategy does better between "pass-call Next" and "pass-pass", and then we'd choose to only do the sample with the strategy that favors passing in R1 the most since that's what I'm trying to refute. I don't really wanna do multiple samples on this hand. It's too time consuming plus I simply don't think it's worth my time. My thinking is if "pass-pass" loses to "call" then "pass-call Next with no trump no aces" will lose to "call" also.

As far as III, my general strategy in the 2nd round is don't lead the turned down suit, but I see why you like that idea here vs a 2 seat call. With only 2 hearts left in the wild we are basically putting S2 in a squeeze situation on the first lead. I like that idea. Problem is if I'm in S2 and I don't have a heart, I'm generally playing off in that spot, so the squeeze won't work vs me but it can work vs others. I really like the idea of leading the AH vs a 2S call in the other hand, if we pass-passed. Now S2 can't just play off and if he trumps in he can easily get overtrumped. I'm convinced leading the AH is the best play in that spot vs Me and vs anyone. That said even if the QH lead was better than the 9S lead, I think you would agree with me that it won't be a gamechanger. I.E. if calling outperforms "pass-pass", changing S1's 2nd rd lead strategy from the 9S to the QH won't come close to changing the overall results. So becuz time is limited and we have a lot of hands we both want to sample :) , I won't explore this option.

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Fri Mar 26, 2021 4:57 am

irishwolf wrote:
Thu Mar 25, 2021 10:35 pm
I can tell you with confidence, S1 calling next is worse than Ordering the AH.

~IRISHWOLF
And this is more reason why I don't need to test the "pass-call Next" strategy. I can rely on your work/judgment on that.

If ordering the AH beats out "pass-pass" when we have:

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_9-C) (Card_9-S)

That pretty much ends the pass argument in my book. When you have 3 trump with no where to go in the 2nd rd, you should order unless the upcard is a Jack (except for very specific scores). Still have to finish my sample of course.

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Fri Mar 26, 2021 9:34 am

With this hand (QH 10H 9H 9D 9S - AH up) calling next, compared to a Mean of 0.0 will be in the range of - 1.00 to - 1.07. I give a range because of the variability of randomly distributed cards and the way different players play their cards. Thus, PASS-PASS, here is a really bad.

However, looking and projecting forward so much more can be done. Although, I think I have or strong players SUBJECTIVELY (subjective probability) think we know the range of results being in these situations as to the best strategy of calling and passing. I do not trust humans in predicting probability of results. But there is light at the end of the tunnel with some work to simplify. If do the work with the worst hands at the bottom of the chart like QH 10H 9H 9S 9D with the mean equal to 0 of PASS-PASS then those in mid-range at S1 vs PASS-PASS, followed by hands that are for sure successful.

You now have this range of solid numeral results. This can now be used to predict the results of almost any, or at least similar hands. Not only at S1 but I suggest, even the same hands at different seats like S4 and even perhaps S3. Then you have to "hard code" it in your memory bank. Just reading it will not do! It's called DELIBERATE PRACTICE!

But this requires work and 95% of euchre players don't want to go that far with this game. Okay I am giving away trade secrets but at my age I don't mind because I look at at as, CATCH ME IF YOU CAN! The best euchre for me is playing against the Best.

~IRISHWOLF

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Fri Mar 26, 2021 12:38 pm

Oh, and one more thing on this Euchre Equivalency Scale I am suggesting. Right in he middle of this scale, is to find the hands that are BREAKEVEN POINT. That is the difficult part those hands that on the average, played many times over, you make as many points for your partnership as you lose calling "Pass-Pass".
[And don't plagiarizer my idea or name - Ya'll.]

OK Richard,I know your listen, a new tool like your BPS but not for beginners.

~IRISHWOLF

Tbolt65
Posts: 820
Joined: Sun Jun 16, 2019 9:14 pm
Location: Las Vegas

Unread post by Tbolt65 » Fri Mar 26, 2021 4:27 pm

Wes (aka the legend) wrote:
Fri Mar 26, 2021 4:34 am
Tbolt65 wrote:
Thu Mar 25, 2021 1:35 pm
The question I would interject into your thought process here is when or if it is ever good to pass twice aside from the math. What situations could warrant a pass-pass?

Something to ponder and think about.
Tbolt65
Edward
I just figured the "pass-call next" strategy would do worse than "pass-pass". I'm trying to bias my sample in favor of the "passing in the first round" strategy becuz that's the strategy I'm trying to refute. If I'm wrong about that, it still shouldn't matter. IOW, if calling here outperforms "pass-pass" as I predict, it's not like "pass-call next" with no trump, no aces is gonna be a gamechanger. If calling outperforms "pass-pass" it will outperform "pass-call next" as both "pass-pass" and "pass-call next" burn tons of points in the 2nd rd. Pass-call next is no panacea. As said, even if I'm wrong that "pass-pass" will do better than "pass-call next" the difference won't be large enough to matter.

That said in real life EVEN IF "pass-pass" did better than "pass-call Next" I would still choose the latter strategy if I were in that spot and the score wasn't up 9-8, tied 8-8. And I can be in that spot becuz against a Jack upcard I'm not ordering except at 9-9, 9-7/9-6. So if S4 passes that Jack I'm now in that spot which begs the question in this hypothetical: why would I still go against the math and call Next with no trump no aces in that spot?!?! Am I irrational? The answer is I am willing to take a few slight EV losses to protect the integrity of my range. I always want my R2-S1 pass to = I have reverse Next blocked so my P can properly read my hand. This allows my P to go alone much weaker in reverse Next should S2 pass, and should S2 call reverse Next, my P can take more optimal lines to euchre S2 becuz he knows what I hold.

In principle and theory it looks good on paper and I agree with you. I've always said no matter the skill level, if both partners are on the same page they will have success. They will know what their partner expects and what certain things mean and truly a form of communication. That helps their over all play. Now we all know there is much to euchre but if partners understand each other and how they play regardless if they play the same this is just as valuable too. However, against players that pay attention and are catching on to whats going on. This information now can be used against you and you can be exploited. Also even if they don't figure it out, we may be giving up free points where a pass-pass would cut down on that but I suggest doing that against weaker teams more often and stronger teams less often but still do it, pass-pass


Some examples: Dealer turns down the 9D and I pass from S1-Rd 2 and S2 passes, you're in S3 with:

(Card_A-H) (Card_J-H) (Card_A-S) (Card_J-S) (Card_A-D)

Knowing my range (I have black blocked but I'm not guaranteed to have anything in Next) you have a very easy Spade loner call.

Same goes for this hand in the same situation:

(Card_J-C) (Card_10-S) (Card_9-S) (Card_A-D) (Card_Q-D)

Becuz you know what I have, this apparent marginal spade loner is now a premium, so go for it.

One more example where S2 calls reverse Next. Ok I pass, S2 calls clubs and you have:

(Card_J-C) (Card_A-H) (Card_9-S) (Card_10-D) (Card_Q-D)

I lead the (Card_9-H)

S2 follows suit, you play the AH and S4 follows suit so you take the trick, now knowing my hand you know exactly what to lead next. Lead the Right and we have an instant euchre since you know I have a guaranteed trick in clubs. This is next level chemistry that I bet no euchre team in the world possesses. I don't wanna threaten that chemistry.

So assuming "pass-next" doesn't do as well as "pass-pass" I'd still call Next assuming the difference isn't too large, so you can always perfectly read my hand when I do pass. There's mathematical value in that too, but obviously it's too hard to quantify.
Coming to an understanding of what you pass and what I pass how ever infrequent, usually means I have stoppers or at least reverse next stopped majority of the time. Can be a huge benefit and I believe there is that underlying understanding already but I have stated and shown but sometimes you forget and get upset that I do pass with nothing what so ever in next or reverse next. This occurrence is super rare. I feel that it brings a little balance and cuts out both of us always calling something and getting euchred too much. However, if you want to go ahead and have me shore this up so that I'm calling something every single time in seat two as to the pass-pass. And allow for your above assertation to take place. We can have a go at that and see what happens, but I'm telling you this. Against weaker players its going to seem almost magical and therefore confirmation bias will set in. We need to play against better opponents. I maintain that little mistakes and even bleeding however small it is points away at a higher level of play is extremely detrimental and can be the difference maker in a win or loss at these higher level of play.


Tbolt65
Edward

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sat Mar 27, 2021 6:03 am

S1-R1

(Card_Q-H) (Card_10-H) (Card_9-H) (Card_9-S) (Card_9-C)

Vs (Card_A-H) upcard.

In the 2nd rd S1 is passing.

Other rules: When S1 calls or S2/S3/S4 call in the first rd, S1 always leads trump. Against a R1-S2/S4 loner S1 always leads the 9S. Against a R2-S2/S4 club/diamond call, S1 always leads the 9S. Against a R2-S2/S4 spade call S1 always leads a club. In the 2nd rd I'm playing S2/S3/S4 as if I was in that seat.
Breached the 95% confidence interval in just 51 hands.

Sample size: 51
3,-1,2,-1,3,-1,3,0,-1,-1,2,2,-1,3,-1,2,-1,0,-1,-1,-1,3,2,3,-1,3,-1,2,-1,-1,-1,-1,-1,-4,-1,0,2,3,2,3,-1,0,-1,-1,3,-1,2,3,2,-1,2
Standard deviation: 1.8369
Mean: .5294
P-value: .0448

EO of calling: -.5098
EO of passing: -1.0392
EV of calling: +.5294

Calling gets euchred: 27/51 = 52.94%
Calling gets 1pt: 20/51 = 39.22%
Calling gets sweep: 4/51 = 7.84%

Conclusion: So calling clearly dominates the "pass-pass" strategy, so much so that I no longer believe that "pass-pass" will do better than "pass-call Next" with no trump no aces". Consider the last hand we tested (AhKh9h9s9d). In that hand calling beat out "pass-call next with 1 trump and a very dirty ace" by around .3 pts. And then we have this QhTh9h9s9c hand in which calling beats out pass-pass by around .5 pts. I know these hands aren't perfectly analogous, but these results would lead me to now guess that "pass-call next" will beat out "pass-pass". I know this is a pointless academic exercise becuz calling dominates both of these strategies but I actually wanna find out if it really is true that "pass-call Next" with no trump no aces really is better than "pass-pass". So now I'm gonna retest the same QhTh9h9s9c hand but this time I'm calling Next in the 2nd rd instead of passing. All other relevant rules are the same as before.

That said, before I end this post it is worth repeating I think I have firmly established that one should call with 3 trump from S1-R1 if they have no where to go in the 2nd rd. This debate in my mind is effectively over.

What about when we are up 9-8? Is it true we shouldn't call with this hand at that score? Let's find out. I will be utilizing this probability chart:

https://members.tripod.com/borf_books/euchprob.htm

When we call we get euchred 52.94% of the time and we score 1 point or more 47.06% of the time. So calling gives us 47.06% equity.

To break down the equity of pass-pass I first need to find out how often the pass-pass strategy leads to our opponents getting 1 pt, 2 pts, or getting euchred. Looking at my excel spreadsheet here's the breakdown:

Pass-pass leads to our opponents getting 2+ pts 13/51 = 25.49% of the time, 1 pt 32/51 = 62.75% of the time and they get euchred 6/51 = 11.76% of the time. Also keep in mind that if our opponents get 1 pt, then it's our deal at 9-9 with approx 65% equity. Ok here's the EV calc:

(.2549 x 0) + (.6275 x .65) + (11.76 x 1) = 52.55%

52.55% is > 47.06% so pass-pass is better than calling at 9-8.

What about at 8-8? Note: if we get a point then it's 9-8 on our deal and we have approx 72% equity.

Calling: (.5294 x 0) + (.3922 x .72) + (.0784 x 1) = 36.08% equity

For pass-pass note: if they get a pt then it's 8-9 on our deal with 36% equity.

Passing: (.2549 x 0) + (.6275 x .36) + (.0784 x 1) = 30.43%

36.08% is > 30.43 thus even at 8-8, calling is better than passing which implies that only at precisely up 9-8 should we abandon the calling strategy.

Actually the only other score I'm curious about is down 9-8 but I can't do an EV analysis of that score becuz we're not passing at 9 in the 2nd rd and I have no data for "pass-call next" yet.

irishwolf
Posts: 1319
Joined: Tue Apr 24, 2018 9:33 pm

Unread post by irishwolf » Sat Mar 27, 2021 8:58 am

Calling next, I got 31% make a point, 69% get euchred (150 hands).

I can also confirm your results on Pass-Pass. Very similar! I see no need to post the details.

~Irishwolf

Wes (aka the legend)
Posts: 1541
Joined: Wed Jun 13, 2018 3:03 pm

Unread post by Wes (aka the legend) » Sat Mar 27, 2021 3:27 pm

Wes (aka the legend) wrote:
Sat Mar 27, 2021 6:03 am
What about at 8-8? Note: if we get a point then it's 9-8 on our deal and we have approx 72% equity.

Calling: (.5294 x 0) + (.3922 x .72) + (.0784 x 1) = 36.08% equity

For pass-pass note: if they get a pt then it's 8-9 on our deal with 36% equity.

Passing: (.2549 x 0) + (.6275 x .36) + (.0784 x 1) = 30.43%

36.08% is > 30.43 thus even at 8-8, calling is better than passing which implies that only at precisely up 9-8 should we abandon the calling strategy.
Have to fix a mistake. I plugged in the wrong number at 8-8. Ok calling did = 36% equity.

Here's the correct equity of pass-pass: (.2549 x 0) + (.6275 x .36) +(.1176 x 1) = 34.35% equity.

36% is > than 34.35% so calling at 8-8 still beat out pass-pass.

Post Reply