When to block/donate

[raydog]

When I first learned of donating [calling dealer up from 1st seat with a weak hand, simply to keep their partnership from potentially scoring 4 pts.] I thought it was ridiculous, and simply due to an emotional bias: we remember those times we were winning 9-6 and were beaten by a loner on the opponents' deal! Even if it was rare, and statistically normal.

But I decided to do a quantitative analysis, and here is what I found.

I used a euchre-playing program of my own design (not perfect, but quite good, and I think satisfactory for the purpose), as well as a program which determines the odds of winning from a given starting score, given the dealing partnership [N/S], first hand dealt to E/W (defenders) and turn card [all other cards randomized], and the odds of the 6 possible outcomes from random future games [4, 2 or 1 pt. scored, by dealing partnership or defenders].

I have found that, with randomly dealt cards, the distribution of outcomes is as follows:
4 pts. to dealing partnership [N/S]: 6.5%
2 pts. to N/S: 14.5%
1 pt. to N/S: 45.7%
1 pt. to non-dealing partnership [E/W]: 18.6%
2 pts. to E/W: 12.2%
4 pts. to E/W: 2.5%

I then established 3 sets of hands dealt to 1st seat, for the initial deal:
1) worst hand:
turn card:
2) average hand:
turn card:
3) below average hand:
turn card:

Next, I looked at 2 possible approaches to playing the hand:
A) normally (i.e., 1st seat passes with a weak hand);
B) 1st seat calls dealer up, effectively donating to prevent the dealing partnership [N/S] from scoring 4 pts.

And finally, I examined 3 different scenarios of initial score:
I: E/W winning 9-6
II: E/W winning 9-7
III: E/W winning 8-3

3 hands X 2 ways of playing X 3 initial scores = 18 separate cases.

Score 9-6, in favor of E/W:
[Just as a point of reference, before any cards are dealt, E/W has an 81.6% chance of winning the game, being ahead 9-6 and not having the deal.]

With the worst hand dealt to E/W, I found the distribution of outcomes to be as follows [note: random cards for all subsequent hands, if necessary]
4 pts. to N/S: 22.09%
2 pts. to N/S: 18.6%
1 pt. to N/S: 52.9%
1 pt. to E/W: 1.9%
2 pts. to E/W: 4.5% [mostly from euchring N/S]
4 pts. to E/W: .01%

If 1st seat doesn't donate (passes), the odds of their partnership winning are 63.7%
If 1st seat does donate, the odds of their partnership winning are 74.1%
Thus, clearly better to donate in this scenario.

With the average hand dealt to E/W, I found the distribution of outcomes to be as follows:
4 pts. to N/S: 8.5%
2 pts. to N/S: 13.2%
1 pt. to N/S: 54.7%
1 pt. to E/W: 7.4%
2 pts. to E/W: 15.7% [mostly from euchring N/S]
4 pts. to E/W: .5%

If 1st seat passes, the odds of their partnership winning are 78.4%
If 1st seat does donate, their odds of winning the game are 77.7%
Thus, pretty much a wash (the difference isn't significant). Not ostensibly better to donate in this case.

With the below average hand dealt to E/W, I found the distribution of outcomes to be as follows:
4 pts. to N/S: 12.6%
2 pts. to N/S: 13.9%
1 pt. to N/S: 62.4%
1 pt. to E/W: 1.8%
2 pts. to E/W: 9.3% [mostly from euchring N/S]
4 pts. to E/W: 0%

If 1st seat passes, the odds of their partnership winning are 72.8%
If 1st seat does donate, their odds of winning the game are 75.9%
Thus, slightly better to donate under this scenario (though still quite close).

Score 9-7, in favor of E/W:
Worst hand dealt to E/W:
If 1st seat doesn't donate (passes), the odds of their partnership winning are 57.4%
If 1st seat does donate, the odds of their partnership winning are 68.2%
Thus, clearly better to donate in this scenario.


Average hand dealt to E/W:
If 1st seat passes, the odds of their partnership winning are 72.3%
If 1st seat does donate, their odds of winning the game are 72.6%

Thus, pretty much a wash (the difference isn't significant). Not ostensibly better to donate in this case.

Below average hand dealt to E/W:
If 1st seat passes, the odds of their partnership winning are 65.9%
If 1st seat does donate, their odds of winning the game are 70.4%Thus, clearly better to donate in this scenario.


Same results as for the score 9-6; not surprising.

Score 8-3, in favor of E/W:
Baseline chance of winning, all random cards: 89.3%

Worst hand dealt to E/W:
If 1st seat doesn't donate (passes), the odds of their partnership winning are 83.1%
If 1st seat does donate, the odds of their partnership winning are 84.1%

Thus, slightly better to donate in this scenario.


Average hand dealt to E/W:
If 1st seat passes, the odds of their partnership winning are 88.1%
If 1st seat does donate, their odds of winning the game are 86.2%
Thus, somewhat better NOT to donate in this case.

Below average hand dealt to E/W:
If 1st seat passes, the odds of their partnership winning are 85.8%
If 1st seat does donate, their odds of winning the game are 84.7%
Thus, slightly better NOT to donate, but perhaps insignificant.


CONCLUSIONS:
Donating clearly can be advantageous, but it is perhaps overrated. With a truly mediocre hand, donating does appear to increase the partnership's chances of winning the game, especially when winning 9-6 (or 9-7) and the opponents are dealing. The the advantage quickly attenuates as the strength of the hand increases, or the distance from the final 10 points increases. To claim that 1st seat should donate (with the score 9-6 in their favor) whenever they don't have a loner blocked - even if a bower is turned - seems to be too extreme. The ultimate goal is to win the game, not to prevent a loner.

A caveat about my analysis: I accept that my tables of point distribution may be inaccurate. I would be happy to re-run this analysis with numbers you feel are more appropriate. As for the "odds of winning", that is simple math [calculated by a program, and based on the given point distribution - irrefutable].

I welcome your comments.

[irishwolf]

Excellent stuff Raydog.

I need more time in digesting your analysis but off hand, I think you are right on the money.

I have toyed with this Donating under various situation myself, several ways, hands and scores. I got very similar results to what you are showing. But I did some analysis with 4 of a suit, 1 trump, one stopper, two stoppers, no stoppers etc. Also with what the dealer had, (Ace, Ace doubleton, KIng doubleton). All impacts the outcome.

With the Jack up, any score is the real DANGER ZONE. Jack is not up, wasting points is my analysis. My analysis says, I can donate with being accurate (that is opposition successfully makes a loner) 1 in 4 tries, makes a sweep 1 in 4 tries, and only 1 point twice if I do not donate (4 + 2 + 1 + 1 = 8) vs I donate all four times (2 + 2 + 2 + 2 = 8) that is the BREAK EVEN POINT.

I can beat those odds! I do not have to get it right 100% of the time. And I will donate at most any score if I think that Jack up will result in a loner. Opponent makes a loner, and your side does not, you will lose the game 90% of the time. Oh and by the way that is with the rule you do not have to have a trump to Donate! lol


So I do not have to be on the money but once in 4 donates, as long as only one sweep. If I cannot do better than that, I should not Donate. It is a different way of looking at it.

I also think there is a big difference between 9 to 6 and 9 to 7. Reason is that donating at 9 to 7, you put the opponents to 9 to 9. You have a greater chance to lose than at 9 to 6.

Coming back to the Jack up vs any other card, twice the chance of a loner (at least with the Jack up). But what if you have One stopper, an Ace or Ace + KIng guarded off suit. Those hands make a difference. But so does it matter that a worst hand of 4 of suit, with or without a trump. One trump means nothing, or even worst in that it takes it out of circulation such that your partner cannot have it.

I myself do not like donating, even at 9 to 6/7 unless that Jack is up!

Lastly, did you actually when you donated, do the outcome of opponents coming back to win the game as you at least gave them one free point? I see as 25 to 30% opponents will come back at 9 to 8/9 and win.

But good stuff, Raydog! Good Job!

IRISH

[irishwolf]

I have had more time to look at your test(s). I have questions:

1) You have 18 tests and how many Hands were run for each of those tests. I could not find that in your post?

2) You posted or combined results as it looks like for the various score combinations, boiled it down to posting results. It would be better to look at head to head Donating vs Not Donating. And the most significant is doing a t-test for significance, head to head for the two ways or at least 9 results. Score differences of 8 to 3 and 9 to 6/7 is apples and oranges, IMO.

3) To me the most significant test here is your of Jack up with the below average hand vs the one all 9s&10s. That average hand vs a King up is apples and oranges vs the Jack up.

4) Even with the Jack up, below average hand you have TWO potential stoppers (if played correctly as queen doubleton has value in defense) with a Green singleton ace and queen doubleton. Even against a Jack up, my testing resulted in successful loners averaging 10%. But close to your testing. You remove that Ace in your average hand and below average hand and the loners will Double from my testing. My point is what Eldest has as a hand makes a big difference and your Worst hand does show that.

All that said, you still did some good work Raydog.

IRISH


You said: "Next, I looked at 2 possible approaches to playing the hand:
A) normally (i.e., 1st seat passes with a weak hand);
B) 1st seat calls dealer up, effectively donating to prevent the dealing partnership [N/S] from scoring 4 pts.

And finally, I examined 3 different scenarios of initial score:
I: E/W winning 9-6
II: E/W winning 9-7
III: E/W winning 8-3

3 hands X 2 ways of playing X 3 initial scores = 18 separate cases."

[LeftyK]

Irish said " Opponent makes a loner, and your side does not, you will lose the game 90% of the time. Oh and by the way" -- not true according to a chart I use that says at any juncture in a game when you go down four points the odds of losing are 80%......not quite 90

[jspectre]

Yeah, unless you donate only in situations where it's statistically correct to do so, you can look at donating as a losing move in order to prevent the worst case scenario, and the viability of such an approach depends on the degree to which you can afford to take that potential loss in order to stop a loner attempt. Therefore, the threat of the loner must either be reasonable or your lead is so great that it can rarely ever hurt to shut the door on the opponents deal. On a game by game basis I find this pretty easy to navigate, even without knowing the real stats, but on a site like Nidink it's much more complicated, because it may only take one loner in a 5-6 game run to end your whole tournament.

[raydog]

This was an exercise in testing intuition with something more quantifiable, and it does seem that donating is a viable and useful option:
1) especially when Right bower is turned (as Irishwolf noted) - I will further explore threshold hands to try and pinpoint what sort of potential 1st seat needs in their hand to make donating NOT the right decision (and, as I mentioned earlier, I suspect it is something less than an absolute stopper);
2) possibly still when A turned (again, I will try to determine the threshold);
3) the "one loner in 4 tries + 1 sweep in 4 tries" [what I expect dealer to get if I DON'T donate] is a simple and elegant shortcut, IF you have the experience to intuitively assess these odds, or a great, unbiased memory of thousands of previous played hands. Again, I'm just trying to quantify, as if I wanted to code a computer algorithm and tell the AI exactly how to play. Which is what I am trying to do (as a side hobby).
4) the statistic that 90% (or 80%) of games are lost when the opponents score a loner seems somewhat off-topic to me. It's a way of saying "getting great cards increases your chances of winning". Donating comes with a cost - a rather steep cost - so it is critical to look at the score when determining a strategy for attenuating the opponents' advantage when it looks like they have good odds of making a loner. Still a big advantage to be dealt a potential loner, but the prize is winning the game, not the hand. So the defense has options.

I played 100,000 random hands for each scenario of "cards dealt to 1st seat + turn card", in order to create the tables of expected point outcomes (4, 2 or 1 pt., for each team).

I didn't look at what dealer has because 1st seat cannot know this; they must make a decision based solely on the information available to them [Irishwolf mentioned toying with the dealers hand when exploring scenarios].

Finally, 9-6 is indeed different from 9-7. Significantly (statistically) better odds of winning the game with the former score. But the point is that, at both scores, donating under the "worst hand" scenario noticeably increases the odds of winning the game. That is all the info needed to make a decision.

[irishwolf]

Lefty, I never sweat the Peanut gallery!

"Oh and by the way" -- not true according to a chart I use that says at any juncture in a game when you go down four points the odds of losing are 80%......not quite 90%"

IRISH

[irishwolf]

And Lefty says, ... "Oh and by the way" -- not true according to a chart I use that says at any juncture in a game when you go down four points the odds of losing are 80%......not quite 90."

So Lefty if you have a chart, then who developed that chart? How many games/hands and at what scores Lefty? If you have that chart, why do you not Post it?

The only chart I know that was published, was by FRED, aka, SWORD. I know him well and his chart and it was not developed about scoring loners, winning & losing directly related to Loners. It was about winning & losing based on where you stood vs your opponents score. But that is not what you said, is it?

I know of no of player or simulator that has tested & developed such a chart about Winning and Losing based on scoring a Loner? So Lefty, put up or shut up!

And frankly, you have no idea what I based my comment on! I do not Post unless I have tested. And by the way, I based my comments on 700 games, 7000+ hands Mr. Lefty!

Frankly, I think you are B-u-l-l S-h-i-t-t-i-n-g! You should not mislead this Forum.

IRISH

[Dlan]

The only chart I know that was published, was by FRED, aka, SWORD. I know him well and his chart and it was not developed about scoring loners, winning & losing directly related to Loners. It was about winning & losing based on where you stood vs your opponents score. But that is not what you said, is it?

Perhaps you could post that chart as there are many that haven't seen it.

[irishwolf]

I can do that Dlan. I will send it to you in a private email attachment. Make sure you give Fred Benjamin (Sword) his due. He also wrote that euchre book, Euchre Strategies, 2007. As far as I know, it not copyrighted. I will send it shortly and let you post it as a Tab or however you think is best for this forum. I think the date it was done would be the same as his book, 2007 or 2008 from ES long ago. I think since it was published as open access on ES it okay to put it here, just give him credit. Its a chart of 10 by 10 so it has 100 cells, each cell 100 hands, 10,000 in total as I understand it.

It based on Probably of Wining/Losing based on various scores. His chart says nothing about one side scoring a loner. Thus, it is only indirectly related to one side scoring a loner and the does not score a loner. , the team scoring a loner will win approximately 90% +/-4%. For example, comparing it to his book, if the score was 9 to 5, I just scored a loner, then probability of winning is 90%. If our team just scored a loner and score is 8 to 5, probability of winning is 88%. Oh but if score is 5 to 1, probability of winning is 798%. As I said, for all scores and situations in euchre, the team that scores a loner and the other team does not, will win 90%. Only indirectly can you related that to Fred's Chart. What chart and how Lefty twisted his information, if he did, I have no clue. I standby my data, 700 games I followed, statistically a game is 101 hands/game. So that is over 7000 hands per one player, ME!

Nothing in euchre is exact or certain, (except the Right Bower is good for one trick) it's all about statistically probable. Of course a team can come back from to 9 to 1 but its not statistically probable.

Irish

[irishwolf]

Dlan, did you get the Chart? It could be commonly used by all, so what about a tab WINNING PROBABILITIES, as a suggestion.

IRISH

[Dlan]

View the chart by clicking link below

viewtopic.php?f=8&t=745

[raydog]

I have tested 5 different scenarios, all with turned. I start with the hand dealt to 1st seat, and compare passing to donating (calling dealer up).

A)
don't donate: 72.8% chance of winning the game (successful loner by dealing team 13.5% of time)
donate: 75.8% chance of winning the game (euchred 89.2% of time)
DONATE

B)
don't donate: 80.6% chance of winning the game (successful loner by dealing team 6.7% of time)
donate: 79.9% chance of winning the game (euchred 74.0% of time)
DON'T DONATE

C)
don't donate: 83.2% chance of winning game (successful loner by dealing team 4.2% of time)
donate: 84.8% chance of winning game (euchred 56.1% of time)
DONATE

D)
don't donate: 86.6% chance of winning game (successful loner by dealing team 1.4% of time)
donate: 84.1% chance of winning game (euchred 58.6% of time)
DON'T DONATE

E)
don't donate: 86.5% chance of winning game (successful loner by dealing team 1.9% of time)
donate: 88.8% chance of winning game (euchred 41.5% of time)
DONATE

Analysis:
I found these results curious. While they probably just confirm the decisions made by expert players, it seems that it's not as simply as saying "I want to prevent dealer from making a loner", as in most of these scenarios the dealing partnership does not actually bid many successful loners (granted that I have chosen hands for 1st seat which give their partnership some ammunition to prevent a successful loner by the opponents - I am trying to find the cusp of donating / not donating).

Does my program play the hands correctly (this is obviously primordial to getting good results)? I think that any errors in play may well benefit the offense as well as the defense, so perhaps cancel each other out. But let's assume that YOU, in 1st seat, are a better player than your opponents. I looked at scenario A and pretended that, when passing R1, dealer (when bidding) gets 11% less 2-pt. results [add them to 1-pt results]. And when donating, you score 1 pt. 11% more frequently. Would this affect your decision to donate or not?

don't donate: win 72.9% of games
donate: win 76.1% of games
Thus, a significant change in game outcomes leads to a very slight change in decision-making criteria. The underlying math of "odds of winning" trumps the possible bias of my euchre-playing algorithm. (In my opinion.)

Another interesting comparison: case C v. case E. In the latter, the stats are better when you don't donate (dealing partnership has many fewer successful loners). BUT, your stats are also better when you donate (many fewer euchres), and on balance it is still better to donate. Hmmm...

In the end, I think having some strength - not necessarily a definitely stopped loner, but a decent chance of stopping it - is sufficient to NOT donate. Up to the player in 1st seat to assess the strength of their hand, but hopefully these tested scenarios can offer some guidelines.
- non-trump aces are important;
- trumps are important (take away from the chance that dealer has them);
- partial block in trump useful (A-9);
- any combination of above has synergies.
________________________________________

*One ancillary question: in scenario E above, when dealer bids alone, should the AH be led? Normally, a lone ace should not be led, but 9H is out of the question, and a trump lead seems futile. Perhaps an exception?

[irishwolf]

Ray,

What was the score? For E. you lead your AH. The dealer has to have 4 trumps to get a loner. You might euchre him as you get two leads of Hs. Even if he has two hearts, he's in trouble.

I would not donate with any of those hands against aggressive players. Why, you have two trumps with a a stopper in those hands except for A) which has two stoppers, one trump. C. D. & E is ridiculous to consider a donate. For example, in sample C., runner has to have AH and 3 trumps or 4 trumps, low odds of that. D. the runner has to have 5 trumps or 4 and a singleton that S1 tosses the wrong ace. In those hands, you have something to run to if the dealer passes. Those are not Junk hands I would donate on. Try 4 of an off suit or all 9s & 10s, singleton queen off suit, three of one suit, two singletons, etc.. Those are my Junk hands for donating as you have nothing to call if the dealer passes as well. Or 4 of one off suit, one trump or none, then a loner is almost inevitable and it could be S2. But to each his own as how people donate.

I just don't agree with those samples you tested for my comfort level and results. If you go to OE on Donates, look at those sample hands of donating or the Posts made by Wes when he donated.

Here is what I think. Suppose the Jack up, any score 9 to 6/7, dealer or his side makes trump all 100 hands. As an example, you have a 15% loner success rate (60 pts), and you attempted loner rate of 45% (30 pts for 30 attempts), 5% S3 has a good hand make a point (5 pts your team), and the remainder 50 hands, 20 of 50 they get 2 pts (20 pts), and the rest, 30 hands only 1 pt (30 pts). In total that is 60 + 30 - 5 + 40 + 30 = 165 pts for NOT donating. And that 15% loner rate would be an estimate based on those hands. If you donated 100 hands = 200 pts. You give away 35 pts, EV = -.35 does not justify donating, IMO.

And you can vary the Loner success rate, make the loners 20%. So add and 15 pts, 180 and it sill does not equal a break even point with 200 pts for 100 donates. Anything less that 15% success rate is not good euchre to donate.

But I don't just donate based on my hand by itself, it's called Intuition. The best players, I played against them, know when is when. But the RUB, you have to stop those damn successful loners. And you can't program that.

IRISH

[irishwolf]

Raydog,

If you want to test something of great value, test the following:

Jack (JH) up, then King (KH) up then 9H up, we do score 9 to 6 for the following hands, DONATE vs NOT:

1) 9D 9C 10C 9S 9H (LEAD EITHER 9D OR 9C FIRST)
2) 9C 10C QC KC 9D (LEADING 9D FIRST)
3) AC KS 9S 9D 9H (HERE LEADING AC FIRST, SAVE KS/9S FOR THE LAST)
4) KC TC 9C 9S 9H (LEADING 9C THEN LEAD 9S FIRST - WHICH IS BEST?)


Reason is that there has been a lot of discussion on these types of hands donating vs not. How many you have been testing for hands. And I will look at my data and we can compare.

IRISH

[LeftyK]

And Lefty says, ... "Oh and by the way" -- not true according to a chart I use that says at any juncture in a game when you go down four points the odds of losing are 80%......not quite 90."

So Lefty if you have a chart, then who developed that chart? How many games/hands and at what scores Lefty? If you have that chart, why do you not Post it?

The only chart I know that was published, was by FRED, aka, SWORD. I know him well and his chart and it was not developed about scoring loners, winning & losing directly related to Loners. It was about winning & losing based on where you stood vs your opponents score. But that is not what you said, is it? --- it's exactly what I said. no one cares about what % of games you lose when a loner is scored, just the score itself.

I know of no of player or simulator that has tested & developed such a chart about Winning and Losing based on scoring a Loner? So Lefty, put up or shut up! -- How about you read the internet more and take a medium cup of hot STFU already.....

And frankly, you have no idea what I based my comment on! I do not Post unless I have tested. And by the way, I based my comments on 700 games, 7000+ hands Mr. Lefty!

Frankly, I think you are B-u-l-l S-h-i-t-t-i-n-g! You should not mislead this Forum.

IRISH

here is that same chart albeit from Natty's it's between 75-90% during any one game in a game so I was closer than you were as your statement counts only at 9-5.

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

Oh BTW you just don't get it do you? JustinMe said it best about you..... "I pity you. You're full of yourself and you likely will never learn how to adress others respectfully, remaining stunted forever"

--- Lefty ----___----

Post by justme » Tue Oct 12, 2021 10:09 pm

irishwolf wrote: ↑Tue Oct 12, 2021 8:45 pm
Well Justme, I can't let your comments stand without some push back.

First, your comment, "You have ignored the fact that STD is also a VARIATION, a stupid needless one which has made the game of euchre less enjoyable for many."

You calling Dlan comment, ignored (variant of IGNORANT)

IRISH
ignored means failed to acknowledge - verb
ignorant means lacking knowledge - adjective
neither is a variant of the other

FACTS! >Your pushback, as you call it, is mostly more long-winded twisted BS, not deserving of further reply. I pity you. You're full of yourself and you likely will never learn how to adress others respectfully, remaining stunted forever.

[irishwolf]

You conclusion you draw are not correct, add up all the scores 4 - 0, 5 - 0, 5 - 1, to 9 - 0, 9 -1, 9 - 2, 9 - 3, 9 - 4, 9 - 5, in the chart where dealer's side is 4 or more, 27 of them I believe and you will be right at 90%. But that is not where I got my 90% as I said before. Again, I stand by my comment of 90%. The chart in fact means little to the subject of one side scores a loner and the other does not. You can't just pull numbers out of the hat.

Just more comments from the Peanut Gallery. Post after Post Lefty, it's fact you do not know what you're talking about. Go ask your brother, lol!

[Dlan]

Ok folks, let’s tone this down a little.

The focus of this site should be the discussion of euchre hands.

Discussions regarding personalities are better suited for Twitter and Facebook.

[raydog]

Irish, I'm looking at running the scenarios you proposed, and have a few questions:
1) why lead only 9D or 9C, and not 9S?
2) my program has N calling 20% of the time when JH is turned; in some circle this is anathema. I (obviously) think it is better to call as N with a strong enough hand, obviating the possible loner by dealer but ensuring they don't pass. Thoughts?
3) (not actually a question) I never assumed that dealer would ALWAYS bid a turned bower with the score 6-9 against her team, which seems obvious to me now. So I am incorporating this into my scenarios (correct?)
The answers to these will help me run better scenarios, help me improve my program, and make me a better euchre player!

[raydog]

Irish, I started testing the scenarios you proposed, but also gave some thought to setting up the experiment properly, and the possible bias in the results.

To recap: 1st seat is dealt , turn card is . Dealer behind by a score of 6-9.

First of all, I assumed that dealer MUST bid in the case of being behind 6-9 and having the JH turned. Their partnership can't afford to give up a point, and this is their best shot.

Also, dealer's partner should only call if they have a loner hand, but otherwise pass (knowing dealer is going to call, and so giving partner a shot at going alone). This strategy yields the most loners.

I then played 100,000 hands testing if it was better to lead the 9C or the 9D (and also the 9S - why not?) I believe the theory here is that, outside of 1st seat's hand, there remain 4 diamonds and 4 clubs, but 5 spades. The object is to find a suit only partner is void in. Partner (and indeed 2nd seat) has an equal chance of being void in D or C; what about the chances of dealer being void in those suits?

Here's where the bias comes into play. When coding my program, I needed to decide if dealer, having singleton QC and QD, decides to discard (and subsequently void herself in) diamonds or clubs. I logically decided clubs, since that is the longer suit [from their perspective, having no other information]. Thus, when I ran the two scenarios of leading 9D v. leading 9C, it was slightly better to lead the 9D - dealer was less likely to be void in that suit (because I programmed it that way), so partner was ever so slightly more likely to be able to trump and win that trick, and thus ever so slightly more likely to make the point for their team.

This is probably the logic you use to suggest leading the 9D, am I correct? But I can't actually PROVE it - it depends on an assumption of how dealer will discard. Knowing the proclivities of your opponent in dealer seat is more important here - do they like to void themselves in next suit or green suit, all else being equal.

I didn't pursue the testing because the results simply reflect the inherent bias of my program. But, by the way, leading the 9S gave me the same results as for leading the 9D - no disadvantage. This must be due either to another inherent bias in my program, or the fact that the difference is so small that it is statistically insignificant, even given 100,000 trials. It does not seem to be a logical choice of lead, by a slight margin (considering odds of partner being void in it and both opponents not).

[It would be possible to determine the odds of all possible permutations of who is void in what, given the constraints of the 6 known cards and all other cards being random, but I am not going to go there!]

The raw data:
lead 9D: 74.3% chance of winning the game if donate
lead 9C: 74.1% chance of winning the game if donate
In both cases (unsurprisingly), slightly over 60% chance of winning the game if don't donate.

[irishwolf]

ANSWERS TO YOUR QUESTIONS, RAYDOG: IN RED

Irish, I'm looking at running the scenarios you proposed, and have a few questions:
1) why lead only 9D or 9C, and not 9S? Answer is you have one less card in those two suits, 16.7% chance S3 is short suited compared to 9S lead.

2) my program has N calling 20% of the time when JH is turned; in some circle this is anathema. I (obviously) think it is better to call as N with a strong enough hand, obviating the possible loner by dealer but ensuring they don't pass. Thoughts? I don't understand as we are Donating here, not trying to score a point. Those are two different situations. Of course if S1 is loaded, a sure hand that would win 3 tricks and S1 cannot call next, I might order. All depends who that dealer is. But, again, to your ordering all depends on what trump S1 holds. I have no idea of what that 20% might be?

3) (not actually a question) I never assumed that dealer would ALWAYS bid a turned bower with the score 6-9 against her team, which seems obvious to me now. So I am incorporating this into my scenarios (correct?) Generally, if the dealer has no trump to go along with that JH up, he might pass. Especially, if he has the Left + XH or a euchre hand, he might pass. So yes, it is not 100% he would turn down a bower. But if you have a dead hand, S1 does not know what the Dealer has and might donate anyway. So sure that should be incorporated.

The answers to these will help me run better scenarios, help me improve my program, and make me a better euchre player!

IRISH

[irishwolf]

Hi Raydog, again my comments in Red:

Irish, I started testing the scenarios you proposed, but also gave some thought to setting up the experiment properly, and the possible bias in the results.

To recap: 1st seat is dealt , turn card is . Dealer behind by a score of 6-9.

First of all, I assumed that dealer MUST bid in the case of being behind 6-9 and having the JH turned. Their partnership can't afford to give up a point, and this is their best shot. Still have to consider what the Dealer holds. For example if he held JD AD JS JC AC he should pass. So it is not 100% he makes but those type of hands are few and far between with what S1 holds. Yes, S2 should only order if he has a loner.

Also, dealer's partner should only call if they have a loner hand, but otherwise pass (knowing dealer is going to call, and so giving partner a shot at going alone). This strategy yields the most loners.

I then played 100,000 hands testing if it was better to lead the 9C or the 9D (and also the 9S - why not?) I believe the theory here is that, outside of 1st seat's hand, there remain 4 diamonds and 4 clubs, but 5 spades. YES The object is to find a suit only partner is void in. Partner (and indeed 2nd seat) has an equal chance of being void in D or C; what about the chances of dealer being void in those suits? Just a chance he has to take, but it could be a HIGH trump and the dealer has a weak loner.

Here's where the bias comes into play. When coding my program, I needed to decide if dealer, having singleton QC and QD, Same number of cards unknown, but favor discarding Next as many will lead next, so discard next. decides to discard (and subsequently void herself in) diamonds or clubs. I logically decided clubs, since that is the longer suit [from their perspective, having no other information]. Thus, when I ran the two scenarios of leading 9D v. leading 9C, it was slightly better to lead the 9D - dealer was less likely to be void in that suit (because I programmed it that way), so partner was ever so slightly more likely to be able to trump and win that trick, and thus ever so slightly more likely to make the point for their team. Your program is not a live person, and I would thing Statistically no Significant Difference. Unless you favor that lead of next.

This is probably the logic you use to suggest leading the 9D, am I correct? But I can't actually PROVE it - it depends on an assumption of how dealer will discard. True Knowing the proclivities of your opponent in dealer seat is more important here - do they like to void themselves in next suit or green suit, all else being equal. I think as much as what the dealer thinks S1 will lead.

I didn't pursue the testing because the results simply reflect the inherent bias of my program. But, by the way, leading the 9S gave me the same results as for leading the 9D - no disadvantage. That is a surprise, has to be a difference with 100,000 hands as a 16.7% more Spades. But the difference is not going to be that great. I hope that makes sense, maybe 4 or 5% difference is my guess Reason is that the dealer is likely to have a doubleton Spade (AS/XS more than a doubleton Diamond. Got it??. This must be due either to another inherent bias in my program, or the fact that the difference is so small that it is statistically insignificant, even given 100,000 trials. It does not seem to be a logical choice of lead, by a slight margin (considering odds of partner being void in it and both opponents not).

[It would be possible to determine the odds of all possible permutations of who is void in what, given the constraints of the 6 known cards and all other cards being random, but I am not going to go there!] We could if you want 16% void with 5 unknown in S and 23.4% for D's.

The raw data:
lead 9D: 74.3% chance of winning the game if donate
lead 9C: 74.1% chance of winning the game if donate
In both cases (unsurprisingly), slightly over 60% chance of winning the game if don't donate.

Need to put that in terms of % loners, % 1 points for the Dealer and % Euchres by S1/S3. Please!

Reason is that if dealer scores 1 point, there is multitude things that then can occur. So I am confused how the 74.3% translates. See this data, then I have to assume Dealer made loner D lead 25.7% of the time? And with club lead, 25.9% which says better to lead the Diamond for S1/S3. Confirm!

IRISH



X

[irishwolf]

Raydog,

When you post like below, on Donating, should we assume 74.3% S1/S3 wins, that 1) The dealer made one point and on the next deal S1/S3 scored a point to win? OR that 2) S1/S3 Scored a point as S3 ordered? This is confusing unless you post results differently?

Then with not Donating, 60% win could be interpreted several ways that S1/S3 won the game. There are too many variables involved to say not donating, won 60%.

As a suggestion is that we want to know if Donating Stopped a would be loner.

Dealer Loner made (+4 pts):
Dealer Loner attempted (+1 pt):
Dealer was Euchred:
S2 ordered:

Donating we know its -2 except for when S3 has a solid hand (which will always against Jack up about 3% or less if S1 has 1 trump.

The raw data:
lead 9D: 74.3% chance of winning the game if donate
lead 9C: 74.1% chance of winning the game if donate
In both cases (unsurprisingly), slightly over 60% chance of winning the game if don't donate.

IRISH

[raydog]

I follow your analysis and agree with what you are saying, Irish. And apologies if I am sometimes incomplete or confusing in what I am trying to say / show.

A note about my euchre program: it is essentially a "black box", and I have to be very careful interpreting the results it spits out, because I had to make some important decisions when coding it. For example, I don't consider the score when making bidding decisions, because it is an added layer of complexity I have not gotten around to incorporating yet (though I of course see that it is important, especially when analyzing the current topic). If I just let it run it will play this hand [the donating example hand] as it would any other hand. So I need to tweak it in this circumstance to ensure S2 only calls up the turn card if he has a loner, and that dealer always picks up the JH [understood there a few hands where it would be better to pass, as you indicated, but I hope they are so rare as to not affect the result]. These are significant changes which do impact the results, so I have to think carefully about incorporating them.

AS for "raw data", I did indeed skip a step, and am happy to share that additional data with you now. In the case of the 9D being led:

- if S1 passes, the distribution of outcomes is (.255, .128, .549, .007, .061, 0), corresponding to (+4, +2, +1, -1, -2, -4) pts. from the perspective of S2/dealer [negative points means the other team scores, and the game is over]
- if S1 donates, the distribution becomes (0, .949, 0, .051, 0, 0), so essentially euchred 94.9% of the time.

In the case of 9C being led:

- if S1 passes, outcome distribution is (.261, .134, .542, .007, .056, 0)
- if S1 donates, outcome distribution is (0, .954, 0, .046, 0, 0)

What I then do with this data is input it into a second program which simply looks at all the pathways to victory for each team, calculates the probability of each particular pathway, then sums them. So, for example, dealer's team could win with a successful loner, or a euchre + a 2-pt. hand, or a euchre + 2 1-pt. hands, etc. S1 / S3 of course win if they score any point before the other team reaches 4 pts. I use a "random hand" probability distribution for all subsequent hands, and the distributions listed above just for the first hand.

It's probably superfluous to give the results of the case where S1 passes, except to confirm that donating is the far better option. When comparing a 9D lead vs. a 9C lead in the case where S1 donates, the only difference is that second outcome distribution. With a 9C lead, dealer's team euchres just a hair more, which then changes the probabilities of the eventual pathways to victory for dealer's team. Hence, better (for S1 / S3) to lead the 9D. BUT... unlike you I did not assume that S1 tends to lead next when I coded my program, and instead have dealer (as declarer, round 1) discarding the suit where there are more cards outstanding (if choosing between two suits). This gives a slight bias to dealer being void in C rather than D, and could in and of itself explain the small difference in results. As I said earlier, knowing the tendencies of the dealer is probably the most important factor here.

[irishwolf]

Raydog, We are getting there, almost. Thanks for the details, and clarification.
Knowing what occurs when S1 does not donate with the hand in question is critical. The only think it does not give tho, is the attempted loners. That attempted loners and percent made of those attempts is an important number.
The only way to score 2 pts is from S2 to order. However, scoring 1 point could come from two ways, Dealer going alone unsuccessfully, and also when S2 ordering, scoring 1 vs 2 pt sweep. (so still a little mucky to me).

I one could assume if S2 had 3 or 4 trumps and 4 unknown as S1 has 1 is about 5%. So something does not make sense as Sweeps are 12.8%. Perhaps, your program has S2 assisting with 2 trumps as well? (that is question)

So clarify Loners attempted, and S2 ordering? Can you even get those numbers from your Black Box?

Even so, one could extrapolate, or guess, but the 25.5% successful loner rate is what is valuable. I will research my archives as I have tested similar with S2 being out of the picture.

Loner made: 25.5%
Loners 1 pt: 54.9% (ASSUMES NO ASSISTS BY S2 - BUT THAT IS ERRONEOUS)
Loners attempted: 80.4% (combined)
SWEEPS: 12.8% (BUT S2 COULD ORDER AND ONLY MAKE 1 PT)
S3 orders scores 1 pt: 0.7%
Dealer's euchred: 6.1%

IRISH

You said:

if S1 passes, the distribution of outcomes is (.255, .128, .549, .007, .061, 0), corresponding to (+4, +2, +1, -1, -2, -4) pts. from the perspective of S2/dealer [negative points means the other team scores, and the game is over]
- if S1 donates, the distribution becomes (0, .949, 0, .051, 0, 0), so essentially euchred 94.9% of the time.

In the case of 9C being led:

- if S1 passes, outcome distribution is (.261, .134, .542, .007, .056, 0)
- if S1 donates, outcome distribution is (0, .954, 0, .046, 0, 0)

[Tbolt65]

I have tested 5 different scenarios, all with turned. I start with the hand dealt to 1st seat, and compare passing to donating (calling dealer up).

A)
don't donate: 72.8% chance of winning the game (successful loner by dealing team 13.5% of time)
donate: 75.8% chance of winning the game (euchred 89.2% of time)
DONATE

B)
don't donate: 80.6% chance of winning the game (successful loner by dealing team 6.7% of time)
donate: 79.9% chance of winning the game (euchred 74.0% of time)
DON'T DONATE

C)
don't donate: 83.2% chance of winning game (successful loner by dealing team 4.2% of time)
donate: 84.8% chance of winning game (euchred 56.1% of time)
DONATE

D)
don't donate: 86.6% chance of winning game (successful loner by dealing team 1.4% of time)
donate: 84.1% chance of winning game (euchred 58.6% of time)
DON'T DONATE

E)
don't donate: 86.5% chance of winning game (successful loner by dealing team 1.9% of time)
donate: 88.8% chance of winning game (euchred 41.5% of time)
DONATE

Analysis:
I found these results curious. While they probably just confirm the decisions made by expert players, it seems that it's not as simply as saying "I want to prevent dealer from making a loner", as in most of these scenarios the dealing partnership does not actually bid many successful loners (granted that I have chosen hands for 1st seat which give their partnership some ammunition to prevent a successful loner by the opponents - I am trying to find the cusp of donating / not donating).

Does my program play the hands correctly (this is obviously primordial to getting good results)? I think that any errors in play may well benefit the offense as well as the defense, so perhaps cancel each other out. But let's assume that YOU, in 1st seat, are a better player than your opponents. I looked at scenario A and pretended that, when passing R1, dealer (when bidding) gets 11% less 2-pt. results [add them to 1-pt results]. And when donating, you score 1 pt. 11% more frequently. Would this affect your decision to donate or not?

don't donate: win 72.9% of games
donate: win 76.1% of games
Thus, a significant change in game outcomes leads to a very slight change in decision-making criteria. The underlying math of "odds of winning" trumps the possible bias of my euchre-playing algorithm. (In my opinion.)

Another interesting comparison: case C v. case E. In the latter, the stats are better when you don't donate (dealing partnership has many fewer successful loners). BUT, your stats are also better when you donate (many fewer euchres), and on balance it is still better to donate. Hmmm...

In the end, I think having some strength - not necessarily a definitely stopped loner, but a decent chance of stopping it - is sufficient to NOT donate. Up to the player in 1st seat to assess the strength of their hand, but hopefully these tested scenarios can offer some guidelines.
- non-trump aces are important;
- trumps are important (take away from the chance that dealer has them);
- partial block in trump useful (A-9);
- any combination of above has synergies.
________________________________________

*One ancillary question: in scenario E above, when dealer bids alone, should the AH be led? Normally, a lone ace should not be led, but 9H is out of the question, and a trump lead seems futile. Perhaps an exception?

I am not currently through all the posts in this thread but I would also like to mention perhaps it has been later on but another determining factor is the likelihood your opponents are capable of going alone with various hands. This can allow you to properly pass in some scenarios. This applies to both seat 4 and 2 opponents. Now granted seat 4 will be more favored in going alone by picking up trump and discarding. A seat 2 order can be had as well however small of a chance that may be. But I digress. Assessing a player properly and their ability to go alone plus the deduction about what is possibly in their hand at said 9-6 score is a needed element when considering when to or not to donate in any scenario but especially important in potential game ending ones.


Tbolt65
Edward