3/12 Fri Got euchred #2

Ask questions, discuss and debate your strategies, euchre polls and more
Post Reply
User avatar
Dlan
Site Admin
Posts: 672
Joined: Thu Apr 19, 2018 10:08 pm
Location: Ohio

3/12 Fri Got euchred #2

Unread post by Dlan » Sun Mar 13, 2022 12:16 pm




raydog
Posts: 260
Joined: Thu Sep 16, 2021 6:56 pm

Unread post by raydog » Mon Mar 14, 2022 11:48 am

My gut instinct, when I saw this hand, was that S3 should have declared, R1. 2-suited, with 3 trump higher than the turn card.

I plugged this into my simulator, and was surprised to see that it was better to pass (which is what my simulator does). By a large margin. Even if I replace the QC with the AC, still better to pass (smaller, but still significant margin). Which is why I have a lot to learn as a euchre player (and why I trust my program - not absolutely, but it plays better than I do).

I also see that S4 (dealer) should have called this hand. EV = +0.08 if they call, EV = -0.35 if they pass. My simulator has its flaws, but that's a very significant difference, so I trust it.

I went ahead and looked at S1 calling S, R2. This was a toss-up: virtually the same points if pass or call. BUT, only 22% of hands made it this far, and as my simulator would have eliminated the actual hand [would have had S4 calling, R1] this result is suspect. The subset of hands making it to R2 is determined by how my simulator plays, not how actual players play (as Irish has poignantly and correctly stated), so this point remains moot.

But I do think there are 2 solid take-aways for those improving their game:
1) don't call from S3, R1, in this position (if perhaps you were so inclined, as I was);
2) do call from S4 - you'll do better in the long run. (according to my simulator - WHICH IS NOT THE DEFINITIVE ANSWER - it's best to at least call with partner with any R + 1 trump)

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

Unread post by Wes (aka the legend) » Mon Mar 14, 2022 7:36 pm

raydog wrote:
Mon Mar 14, 2022 11:48 am
My gut instinct, when I saw this hand, was that S3 should have declared, R1. 2-suited, with 3 trump higher than the turn card.

I plugged this into my simulator, and was surprised to see that it was better to pass (which is what my simulator does). By a large margin.
Yep, it is very counterintuitive how tight one should be in the 3rd seat and it's not just the fact that we're in the worst seat in the game without the lead and a guy behind us, S4, just created a void and can now overtrump us. In the past to help explain how tight we should be in this spot I would evoke what I call "the parlay problem" which goes something like this:

1) If S3 knew the dealer would call if he passed he would never make this call and risk a euchre.

2) If S3 knew his P had a 1+ pt call should the dealer pass, he would never make this call and risk a euchre.

IOW in order for S3 to be happy with this call a parlay has to happen. Both conditions 1) and 2) must NOT be in place (the dealer has to have a passing hand AND S1 has to have passing hand or a call that would end up getting euchred). Anyone who's dumb enough to sports bet (like me) knows parlays are really hard to hit.

The key implication/conclusion: To overcome this parlay problem S3 needs a REALLY good hand, often a much better hand than one would think. As an aside experts intuitively get the parlay problem which is why paradoxically an expert will often pass more from S3 than an amateur assuming that expert can trust his P in round 2.

Of course the above way of thinking is really just a heuristic helping one wrap their brain around the strange situation of S3-R1 hands. While this way of thinking is useful in helping one make the right decision, a simulation is much better.

Irishwolf even pointed out a hand where this mental model would clearly fail:

Upcard: (Card_K-H)

S3: (Card_Q-H) (Card_10-H) (Card_9-H) (Card_A-S) (Card_K-S)

Clearly looking at this hand through the lens of the parlay problem would lead S3 to pass this ostensibly mediocre hand and yet Irishwolf's sample showed that passing this hand is a mistake. I did a sample of my own reaching a 95%+ CI and got the same result as Irishwolf. The problem with this hand came in the 2nd round. It barely helps a Next call, same story should S1 call clubs. It really only supports a S1 spade call. So S3 passing this hand ends up burning too many points in the 2nd round.

But this dynamic changes when S3 holds the Left in his hand. When that's the case the "parlay problem heuristic" works. When S3 holds the Left that obviously means he'll have the Right on a Next call, and that makes all the difference.

Back to the actual hand. Notice that should S3 pass:

(Card_J-H) (Card_K-D) (Card_Q-D) (Card_Q-C) (Card_9-C)

He actually can help his P's S1-R2 calls in two different suits. 1) he has the Right bower for S1's next calls. 2) Should S1 jump the fence in clubs S3 will have 2 trump + a void. I count that as significant help. That's a lot of support for the 2nd round. Just bad luck that S1 called spades.
raydog wrote:
Mon Mar 14, 2022 11:48 am
Even if I replace the QC with the AC, still better to pass (smaller, but still significant margin). Which is why I have a lot to learn as a euchre player (and why I trust my program - not absolutely, but it plays better than I do).
I agree that's a good pass....in theory. Like if Ed/Wolf and really anyone from the Monday/Friday games are my P, I'm passing that hand unless we're in a closeout spot, I.E. tied 9-9 or up 9-X where X = 7 or less. Keep in mind in closeout spots the math changes becuz we now effectively only get 1 point for euchring the enemy so we should loosen up a little.

The big monkey wrench in these simulator results of course is how well S1 plays in the 2nd round. If S1 is the typical amateur who plays poorly in the 2nd round, my belief is that L+2+suited ace hands become a call from R1-S3. Not a call we wanna make, but a call we gotta make. The same could even possibly be true for weaker L+2 hands like the one in the OP. IDK, I suspect that even with an amateur P we'll be better off passing the OP hand, but once you give S3 that suited ace to go along with his L+2 now I can't pass that with an amateur P but it's a super easy pass for me when I trust my partner.

Tangent: Here's a fun simulator exercise. Assume S1 is a strong player. Score is 0-0. Is there ANY S3 L+2+suited ace hand worth calling? Take the very best hand:

Upcard: (Card_9-D)

S3: (Card_J-H) (Card_A-D) (Card_K-D) (Card_A-C) (Card_K-C)

What's the best play for S3? Pass? Call? Go alone? For the record I would go alone with this hand.

What about this hand:

Upcard: (Card_9-D)

S3: (Card_J-H) (Card_A-D) (Card_K-D) (Card_A-C) (Card_A-S)

Pass? Call? Go alone? The problem with going alone with this hand is S3 only has 1 void and S3 starts off in a squeeze on 1st street. Any time S2 leads a black card S3 is gonna hate life. Thus it is harder to dodge that first lead with this holding.
raydog wrote:
Mon Mar 14, 2022 11:48 am
I also see that S4 (dealer) should have called this hand. EV = +0.08 if they call, EV = -0.35 if they pass. My simulator has its flaws, but that's a very significant difference, so I trust it.
I honestly don't know how people are still making this mistake in the Monday/Friday games. 1) Every expert I know calls with this hand, that's not the strongest argument ever--it's even a bit circular--but it's still evidence no one should ignore. 2) Eric Zalas' work shows this is a call. 3) Your work shows this is a call. 4) The recommendation on this site--I.E. the best euchre site on the internet--says this is a call. Like what else are people looking for. You're not gonna get a stronger case than this. I even have another bullet point to throw at people. Since at least 2012--I'm not familiar with the people before that--No one in my weekly euchre tournament who passes R+1+0 from the dealer spot has ever won the Player of the Year/Euchre championship. In my tournament most people pass that hand. Not surprising as probably 95% of the euchre player pool on this planet pass that hand. Only a small subset call that hand and the champions are coming from that small subset.

An interesting contrast between your work and Eric Zalas' work. Eric's simulation showed that a S4 call with R+1+0 is a losing play, I.E. it had a negative expected outcome (going off memory it was -.28), but it was easy to surmise that passing would have a worse negative expected outcome, therefore calling R+1+0 is +EV overall, a dynamic that somehow eluded Eric. Your work reaches the same conclusion but it tells a different story. Your simulator suggests that R+1+0 is a pure winning play, with a positive expected outcome of +.08. What's going on here? I think I can explain the disparity. I believe the "players" in your simulator play better than the "players" in Eric's simulator. You've been constantly tweaking trying to make the players better. I bet Eric hasn't. Specifically when one calls with R+1+0 they often have to play off in spots hoping their P takes a trick. I bet the dealer in your simulator is playing off in certain spots, but the dealer in Eric's simulator is just trumping in whenever they can take a trick. That's just my guess so basically I'm betting your result is more accurate becuz you have a better simulator than Eric's.

All that said, there is still a S4 R+1+0 grey area spot left to be resolved. What should we do with R+1+0 when we have all suits blocked? My general recommendation is to pass. I know Don disagrees with that and I know Wolf has disagreed with some of my passes from some of these hand types. To establish proof of concept we'd have to test the weakest hand from this configuration. If calling beats out passing from that hand then we can basically throw the whole idea out. E.G. a hand like this should be tested:

Upcard: (Card_9-H)

Dealer holds: (Card_J-H) (Card_9-D) (Card_J-S) (Card_9-S) (Card_J-C)

If passing beats out calling and thus we make it past the proof of concept stage then we would next want to get greedy for the sake of seductive simplicity and hypothesize that it is best to pass with ALL R+1+0 hands when we have all suits blocked. To test this hypothesis we would simply have to test the very best "R+1+0 all suits blocked" hand. If passing STILL beats out calling then it's game over for calling, but if calling wins then the simplicity we all crave and love goes out the window. It gets messy from that point. For this purpose we'd wanna test a hand like this:

Upcard: (Card_A-H)

Dealer holds: (Card_J-H) (Card_9-D) (Card_J-S) (Card_9-S) (Card_J-C)

And in case it matters change all 9s to kings and test this hand:

Upcard: (Card_A-H)

Dealer holds: (Card_J-H) (Card_K-D) (Card_J-S) (Card_K-S) (Card_J-C)

Like I said before, if passing beats out calling from those hands then its game over for calling with R+1 when we have all suits blocked*.

*:The exception being when our R+1 = both bowers. That's a totally different hand to me and thus not relevant to this discussion.
raydog wrote:
Mon Mar 14, 2022 11:48 am
I went ahead and looked at S1 calling S, R2. This was a toss-up: virtually the same points if pass or call. BUT, only 22% of hands made it this far, and as my simulator would have eliminated the actual hand [would have had S4 calling, R1] this result is suspect. The subset of hands making it to R2 is determined by how my simulator plays, not how actual players play (as Irish has poignantly and correctly stated), so this point remains moot.
I also call spades in that spot, leading the Right, followed by the Ace. I'm glad your simulator indicates I'm not burning points. That's a spot I've always wondered about becuz the argument for passing is plausible to me. We do have 2 out of 3 suits blocked if we pass as we fully block reverse next calls and calling isn't that attractive as we are 4 suited which really hurts the value of our little TS trump. My argument for calling isn't very sophisticated. It basically goes like this: R+1+A is just too strong of a hand to pass when I don't block all suits period. This is definitely a grey area spot for me and your results back that up. Give S1 a void and I bet the numbers swing away from a statistical tie to a clear spade call. For example, change S1's hand to this:

Upcard (Card_10-D) turned down.

S1: (Card_J-S) (Card_10-S) (Card_A-C) (Card_Q-D) (Card_9-D)

Now I feel strongly that's a good spade call.

Post Reply