Obviously I'm not surprised by this finding. Glad your data backs up my claim. I've tested hands like this vs Pass-Call Next (even tho that might sound crazy given we have no next cards), and calling won at a sample large enough to reach a 95% confidence interval. This hand type has never been tested against pass-pass until now. I know calling had a negative expected outcome, and I predicted pass-pass would have a higher negative expected outcome, but the numbers are closer than what I would've thought. Either way, I feel comfortable saying calling is best. Again even in the worst light--a statistical tie--ties go to the caller.raydog wrote: ↑Thu Jan 13, 2022 7:19 pmQuestion #10:
Your team is up 4-2. You're in the 1 seat. The dealer upcard is the
You have , , , ,
Tested on 10,000 hands,
bid: EV = -0.66 (564 / 3,720 / 5,716) 564 sweeps; 57% euchre rate
pass: EV = -0.74 (2,700 loner calls by S2/S4, 23% success rate; 750 sweeps; 1,180 total calls by S3)
While I think there is a reasonable argument to be made, based on these results, that bidding is the better option, there are a slew of assumptions that need to be made about how well my program mimics the reality of how all these hands would be bid and played. So I'd be more confident concluding that bidding is not a bad decision - it's not going to cause any significant EV hit - but passing is not necessarily a terrible option.
As far as some assumptions with this holding. When S1 calls spades we are always leading a spade. If we pass-pass, and the S2 calls red, I would always lead the red non-trump card (an argument could be made for leading the turned down suit through the maker tho given how dirty it is trying to force an overtrump). If S2 calls clubs, I would lead the KD (again an argument could be made for leading the turned down suit through the maker trying to create an overtrump situation). If it gets to my P in the 2nd rd, I would always lead trump if I have it. If he calls clubs I would lead the KD. If the action gets to S4 in the 2nd rd: when he calls red I'm leading the non-trump red card. If he calls clubs, I'm leading the KD.
Worthy of note, I've talked about this awhile back in other threads but my claims have never been verified.
Claim 1: Same situation BUT if we have:
Pass-pass is now better than calling spades in R1 becuz we have good defense in the 2nd round given that we block reverse next calls.
Claim 2: If we have:
Passing with the intention of calling hearts in the 2nd rd and probably leading the QS hoping the KS is buried is better than calling spades in the 1st rd.
The crux of this situation is S1 has no defense and nowhere to go in the 2nd rd. When that's not the case S1 should no longer make this very marginal call. Am I correct? Can you test claim 1 and 2?
I think we've covered most of these possible hands from the dealer spot but if you discover new ones please let us know!raydog wrote: ↑Thu Jan 13, 2022 7:19 pmI am personally going to revisit a whole subsets of hands which my program currently passes and try to identify very clear cases where bidding, while generating a negative EV, is still favorable to passing. And this will be a hand I bid. My program already does this in many cases, but I'm finding more and more "holes" in my algorithm.
I'll throw one out there for you to test. It's not a good hand but I'm curious. Score is 0-0, you're the dealer.
Upcard:
You hold:
Discard the 9S. If this hand is +EV--meaning it's better than passing--I'm gonna crap my pants. And just in case having that AH makes passing the better play test this hand too (same KH upcard):
BTW as far as other spots "where bidding, while generating a negative EV, is still favorable to passing" I'm predicting you'll find a lot of them in the 2 seat 1st rd.
I recently tested this 2 seat hand:
Upcard:
We have:
Calling beat out passing at a 95% CI.
If S1 led a non-ace spade I always played off with the 9H. If S1 led the AS, I would trump in and then lead the AD on 2nd street unless my P threw off a diamond on the first lead suggesting they have another diamond. In that case I'm torn between leading the 9H or leading my last trump. With no more voids to work with, I think leading trump is best but don't know. If S1 led a diamond and I took that trick with my AD and my P threw off a spade, I would lead the 9h on 2nd street hoping my P is void. If on that diamond lead my P threw off a heart meaning he probably has another heart I honestly don't know the best lead on 2nd street, I would actually lead trump, but leading the 9H could be best. If my P threw off an Ace on my diamond trick I would always lead trump on 2nd street as my P is probably loaded.
And whenever my P takes the first trick they will always lead trump if they have it. If they only have 1 trump and use it on the first lead, then I would have them lead the AH if they have it or the cleanest fresh suit to lower the probability of me getting overtrumped; unless they had a signal to lead something else, E.G. if S1 leads a non-ace spade, and I throw off the 9H and my P takes the first trick with a trump and has no more trump and no off fresh AH to lead, then they should lead a non-ace heart if they have one as I am likely void in that suit now. If S4 has no hearts and no trump, then obviously leading diamonds is their only choice.
Jesus, that's a lot of assumptions behind a relatively simple marginal call. I'm exhausted after writing all that, probably worse for the reader lol. Just shows how complex a euchre hand can be.
As far as a more marginal 2S hand to test how bout this one:
Upcard
We have
Eric Zalas was actually myopically focused on +EO (expected outcome). It somehow never occurred to him that a -EO call could be +EV overall if passing had a worse -EO--a critically important concept in euchre. It actually makes his entire work on euchre farcical yet I would still recommend his work to any expert becuz the hand samples and simulations Zalas did are interesting and important. You just have to ignore Zalas' flawed conclusions.
I've thought of this problem before too. It's almost like we need two simulators. In one simulator everyone makes the correct play answering the question "what is the best play in a very tough game). In the other simulator you have 1 person making the correct play while everyone else plays like an amateur answering the question "what is the best play in a typical euchre game you'll be in on an app and usually in real life". But honestly I don't think this theoretical problem is that big of a deal and for the sake of parsimony I think we should ignore it. Plus it's not that hard to make adjustments to your findings. For example, say this 2 seat call was very slightly -EV overall:raydog wrote: ↑Thu Jan 13, 2022 7:19 pmThe downside is that, if most players don't bid this way then my program will veer farther and farther from the actual subset of hands left in play late in R2 (and by "actual" I mean what one will encounter when playing with live players). Hopefully that won't undermine the results that come from making the "correct" bid.
Upcard
We have
We could still deduce that this hand is probably a call with an amateur P who is not calling with R+1+0, L+1+A, 2 non-bower trump + 2 aces, even 3 low trump- three suited-no off ace hands, and of course those marginal 2 non-bower trump + an off ace hands that are actually +EV calls.