Calling Next with 1 trump + off ace, an EV analysis
Posted: Thu May 02, 2019 4:33 pm
Ok so we're going to create an EV model here (EV = expected value). EV models are not mathematical proofs but rather mathematical arguments that can often help lead us to correct play. An EV model is only as good as the assumptions that underpin it, and those assumptions are often guesswork (hopefully born from expert intuition) but that's not necessarily a bad thing as far as leading us to the correct/incorrect play.
For example, if all our assumptions are conservative, I.E. they err on the side of caution, the side that actually goes against our argument, then we have a very strong, compelling EV argument. IOW, the most compelling EV model would be one that is based on unrealistic assumptions, as long as those assumptions are conservative, meaning unrealistic in the direction that actually goes against the claim an EV model is trying to establish.
In contrast, watch out for EV models purporting to establish a correct strategy in euchre, poker, or any card game when all their assumptions are highly favorable to their argument. Any EV model can spit out the answers you want if you tweak the assumptions enough. That's why EV models can't PROVE anything but they can be excellent teaching tools, and they can sometimes make very powerful arguments.
Ok so now I'm going to make a simple EV model that suggests that calling Next with 1 trump + an off ace is correct. Here we go, and remember the best way to attack my model is to attack its assumptions.
The score is 0-0, we're in seat 1 and the dealer just turned down the
We have
And we call next.
Assumptions: If we call next with this weak holding we will get euchred 65% of the time, we will make a point 30% of the time, and we will score 2 points 5% of the time.
The big key assumption of the model: If we pass seat 2 will always call something. This is the weakest part of the model. It is not a conservative assumption, but nevertheless a necessary one. The model simply gets too complex, overloaded with too many other permutations to account for if we allow Seat 2 to pass. The good news is the assumption is a decent approximation of reality vs many good opponents.
Given that you're passing a hand that blocks nothing, a hand with no bowers, and assuming that seat 2 is good enough to know how to call to hit his partner's range those times he's marginal which won't be that often given your holding (I.E. making good reverse next calls when the cards warrant it), Seat 2 will not be passing very often. Odds are Seat 2 has a hand to call with when you pass such a weak holding.
I would love to know my passing % after seat 1 passes with a hand that blocks nothing. I bet it approximates 0% when you factor out euchre hands I would pass and other hands I would pass when I have all suits blocked, and seat 2 passing those type of hands doesn't hurt the model becuz those are basically aggressive passes that have more in common with calling then a legit pass, in the mathematical sense.
Ok more assumptions: If we pass, Seat 2 will complete a loner sweep 5% of the time, get euchred 10% of the time, score 1 point 65% of the time, and score 2 points 20% of the time.
OK time to do the EV calcs!
First step, do the math on your next call: (.65 x -2) + (.30) + (.05 x 2) = -.9
I.E. every time you make this call you lose .9 points in a vacuum on average.
2nd step do the math on a 2 seat call should you pass instead:
(.05 x 4) + (.10 x -2) + (.65) + (2 x .20) = 1.05
I.E. every time you pass you your team losses 1.05 points on average.
Since 1.05 is greater than .9, we are better off calling Next with this holding than passing.
Ok, what I want from you guys:
Plug your own numbers into this model, your own best guesses, see what conclusion it churns out and post it in this thread. I don't care if it reaches a different conclusion than mine. There are no wrong answers here. Or I should say there are no provable wrong answers here.
Some follow up questions. Are there any assumptions in my model that you vehemently disagree with? Or are the numbers acceptable enough that you are now convinced that calling next with 1 trump and an off ace is the right move? Let me know what you guys think.
For example, if all our assumptions are conservative, I.E. they err on the side of caution, the side that actually goes against our argument, then we have a very strong, compelling EV argument. IOW, the most compelling EV model would be one that is based on unrealistic assumptions, as long as those assumptions are conservative, meaning unrealistic in the direction that actually goes against the claim an EV model is trying to establish.
In contrast, watch out for EV models purporting to establish a correct strategy in euchre, poker, or any card game when all their assumptions are highly favorable to their argument. Any EV model can spit out the answers you want if you tweak the assumptions enough. That's why EV models can't PROVE anything but they can be excellent teaching tools, and they can sometimes make very powerful arguments.
Ok so now I'm going to make a simple EV model that suggests that calling Next with 1 trump + an off ace is correct. Here we go, and remember the best way to attack my model is to attack its assumptions.
The score is 0-0, we're in seat 1 and the dealer just turned down the
We have
And we call next.
Assumptions: If we call next with this weak holding we will get euchred 65% of the time, we will make a point 30% of the time, and we will score 2 points 5% of the time.
The big key assumption of the model: If we pass seat 2 will always call something. This is the weakest part of the model. It is not a conservative assumption, but nevertheless a necessary one. The model simply gets too complex, overloaded with too many other permutations to account for if we allow Seat 2 to pass. The good news is the assumption is a decent approximation of reality vs many good opponents.
Given that you're passing a hand that blocks nothing, a hand with no bowers, and assuming that seat 2 is good enough to know how to call to hit his partner's range those times he's marginal which won't be that often given your holding (I.E. making good reverse next calls when the cards warrant it), Seat 2 will not be passing very often. Odds are Seat 2 has a hand to call with when you pass such a weak holding.
I would love to know my passing % after seat 1 passes with a hand that blocks nothing. I bet it approximates 0% when you factor out euchre hands I would pass and other hands I would pass when I have all suits blocked, and seat 2 passing those type of hands doesn't hurt the model becuz those are basically aggressive passes that have more in common with calling then a legit pass, in the mathematical sense.
Ok more assumptions: If we pass, Seat 2 will complete a loner sweep 5% of the time, get euchred 10% of the time, score 1 point 65% of the time, and score 2 points 20% of the time.
OK time to do the EV calcs!
First step, do the math on your next call: (.65 x -2) + (.30) + (.05 x 2) = -.9
I.E. every time you make this call you lose .9 points in a vacuum on average.
2nd step do the math on a 2 seat call should you pass instead:
(.05 x 4) + (.10 x -2) + (.65) + (2 x .20) = 1.05
I.E. every time you pass you your team losses 1.05 points on average.
Since 1.05 is greater than .9, we are better off calling Next with this holding than passing.
Ok, what I want from you guys:
Plug your own numbers into this model, your own best guesses, see what conclusion it churns out and post it in this thread. I don't care if it reaches a different conclusion than mine. There are no wrong answers here. Or I should say there are no provable wrong answers here.
Some follow up questions. Are there any assumptions in my model that you vehemently disagree with? Or are the numbers acceptable enough that you are now convinced that calling next with 1 trump and an off ace is the right move? Let me know what you guys think.