What would you do here?

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: What would you do here?

Unread post by Wes (aka the legend) » Tue Dec 21, 2021 6:11 pm

raydog wrote:
Mon Dec 20, 2021 3:17 pm
Sorry, didn't see the updates to this thread until Irish reminded me.

First, I ran the comparison of donate, R1 vs. call next (hearts), R2. No bids the same between the two scenarios, so I just compared overall points scored by both teams over 100,000 hands.
Donate: EV = -1.52
Bid H, R2: EV = -1.11

This was to answer Wes's question. It was not really possible to extract this result from the data I presented because the EV I showed was not always based on all 100,000 hands, just on the hands where bidding differed (when comparing two 2nd round bids, for example, I ignore the results of first round bids because they will likely encompass most of the score and dilute the results, as they are the same for both scenarios).
Makes sense Ray. Good stuff. Based on your data (1.52 - 1.11 = .41), and the sample I've done on this topic, I think it's safe to say donating costs around .4 to .5 pts when compared to other strategies assuming one is donating in the kind've spots I've talked about in the past. This means that donating cannot be justified from an EV perspective, as it seems to me all donates cost something EV-wise. Donating can only be justified as a tool to control variance. And then the question becomes when is it worth taking this -EV hit in the name of controlling variance? An example of one heuristic that I've tried would be "donating IS worth it vs a non-jack whenever we are up by 3 or more." And as mentioned above, by donate I don't mean reflexively donating when we're unguarded but donating in the kind've spots I've talked about in the past. Unfortunately I don't see a way that we could "prove" that making ANY -EV play leads to more wins in the long run. If there is a way to prove that concept only a computer simulation could do it. This is certainly off topic from this thread so no need to respond.
raydog wrote:
Mon Dec 20, 2021 3:17 pm
I then tried the simulation Irish proposed. I had S1 pass R1, bid Clubs R2, then compared leading QH vs. leading 9D. I noticed that when S2 or S4 won the first trick, then led a Spade which S1 trumped with the 10C on trick 2, my program would lead the KD [would actually do this whether the 9D or the QH was initially led]. I stuck with this in the first instance, just to see the results.
QH led: EV = -0.37
9D led: EV = -0.54
[Note that in its current configuration, my program would have led the JC, trick 3, if S1 had an off-suit Ace; it will also lead the JC on trick 4 if it only has the JC and KD remaining, having discarded the QH on some previous trick]

I then tweaked the the program to play exactly as Irish stated: 9D lead, then trump with 10C if Spades* led, trick 2 [if partner is not winning with the Ace, and if an opponent has not already trumped with a card higher than the 10], then lead the JC, trick 3, to set up the KD. Note that things only played out exactly this way about 1/5 of hands:
QH led: EV = -0.38
9D led: EV = -0.49
Good stuff. It goes without saying I'm not surprised by the results.
raydog wrote:
Mon Dec 20, 2021 3:17 pm
But it's not as easy as saying Irish is wrong. There are just so many ways this hand can play out. As I said, it only follows the ideal script Irish laid out infrequently, and when it does play out that way it is apparently not enough to swing the tide.
It doesn't really matter to me whether Wolf is wrong or not. All I care about is what the evidence points too. When it comes to 2nd round situations, I would trust a good computer simulation over any kitchen table simulation due to the true distribution problem. By true distribution problem I mean this: If one were doing this on a kitchen table they couldn't even get to the 2nd round without first giving S4 the JD upcard with S1's hand fixed, and then dealing out the cards hoping everyone has a legit passing hand in the first round. If any player has a calling hand you have the scrap the deal and start over becuz if you get to the 2nd rd any other way your distribution won't be true and thus one's data won't be worth much. This would be a very frustrating and tedious exercise rendering a kitchen table sample highly impractical especially when the Jack is the upcard. One could imagine doing this having to scrap the deal 10 times+ just to get one sample. That's why we especially need computer simulations for 2nd round spots.
raydog wrote:
Mon Dec 20, 2021 3:17 pm
Also, my program will not trump trick 2 with the 10C if S4 leads something other than an A or K of spades (hoping partner can win the trick, as S1's holdings are thin; this explains the * above); but this also happens infrequently.
Your program is doing the right thing (except I think you should play off in that spot if S4 leads the Ks too). Assuming a non-ace of spades is led, trumping in would be very poor euchre play.

You've already explained why but I'll go into more detail. Yes, when you call super thin, you should tend to give your P opportunities to take a trick as much as feasible. I think it's critical to play off on a blank spade here becuz when you are this thin and your team crosses wires that's usually enough to sink your team's ship for a euchre. By "cross wires" I mean whenever you trump a lead that your P was boss in, basically stealing your P's trick. It's also important to point out that when S4 leads a blank spade, you--S1--are not closing the action. It may be unlikely but you can get overtrumped by S2 should you trump in. This possibility is another reason you should never trump in, in this spot. When you call thin like this getting overtrumped often leads to a euchre. No need to expose yourself in this spot.

But there's actually another way to argue for playing off in that spot, and it involves going back to what to lead in the beginning: after you called clubs you had 5 lead choices and 4 of them were toxic. Both club leads would be toxic as the last thing you want to do with this holding is have your team spend 2 trump on one lead. And both diamond leads were toxic. In general leading the turned down suit is very poor play, but in this case it's worse. We're talking about leading the turned down suit after the Jack has been turned down. That's not just your run of the mill poor lead, that's one of the worst possible leads in this game. After the Jack as been turned down not only does that lead almost always gives the enemy an easy trick on first street and thus destroy your teams chances of getting 2 points, this lead can also cause trouble in other crucial ways. Those times S2 has the AD, a diamond lead will now often give S4 a chance to create a void in another suit. This is a terrible occurrence that can easily come back to bite your team later in the hand.

Ok so going back to what I said, on our first lead we had 5 choices, 4 were toxic, so it was easy to choose our only non-toxic lead, the QH. How is this an argument for playing off on a S4 non-ace spade lead on trick 2 after we led the QH. The answer lies in thinking ahead + remembering our "what to lead" puzzle. If we DO trump in on a blank spade lead look at where that leaves us on 3rd street. On 3rd street we'll have two diamonds, and the JC. Notice, if we trump in, we'll have nothing but toxic leads on 3rd street as those leads are still toxic for the same reasons they were toxic on the first lead! Why put ourselves in that bad situation when we don't have to. This is another reason to always play off in that spot vs a non-ace spade lead.



Post Reply