Is This Hero vs Computer Program Useful?

How is the "computer" playing and what is that based off of? Is it only preflop?

Yeah definitely sounds useful. There are similar programs out there, but most I've used are either preflop or post- not both. The hand choices you mentioned sounds novel as well.

Another nice feature would be if you could set a certain amount of EV loss so that the program flags any mistakes that lose more than a certain amount. That way you know to go back and study any spots where you make large blunders.

The computer knows the pot, its hand, position, call and stack sizes and hero image if that’s set up. I then use standard EV analysis to determine its action, such as maximum bet size for +EV for equity < 50%. Of course, computer doesn’t know its exact equity but I have developed a way to put into ranges. I also randomize the computer action so hero cannot easily determine its strategy. Stack sizes are also considered.

Yes, the game is only preflop and I realize that’s a significant limitation and I doubt I can extend that. Another limitation is that the betting sequence is limited to the following: sb fold (very rare), sb bet – bb fold , call or raise. If a raise, sb either folds or calls. The computer fold/call response to a raise is also EV based with a random adjustment.

GWF – Idea is interesting but I’m not sure what you mean by mistakes.

As an example of potential value, suppose a player feels he should be doing better with AT+ hands. So, he has only those hands dealt to him and he plays various strategies and analyzes results to find the most profitable EV one. Of course, the most profitable preflop EV one may not hold up post flop

Useful or not, I’m having fun developing the program. Thanks for your responses.

I think I somewhat misunderstood what you created when I read your first post. I was thinking you were going to program it so your computer "opponent" played like a pre-solved solver strategy or something, so you could play against the solution. So I was thinking it would be cool if it flagged a play you made if it performed very poorly against the solver strategy. I was also assuming you were going to add post flop play.

Anyway it's still impressive what you created. I think the quality of your opponent's strategy is important for the program to be useful, that way you're simulating playing against something realistic.

I don't want to make him too good for then I will never win while testing😀

I was just practicing some simulated hands where I had the solver's prefered play set to display before I acted. So I was literally playing perfect GTO copying the correct answer if my instincts were wrong.

I somehow still lost like 20 out of 25 hands including pretty much every all in. You know you're running bad when you can't even win with all the answers.

Now I was also playing against the solver, but still I understand what you're saying. It's more fun to play against an opponent who doesn't play perfectly 😀

If your program is interactive in a way the same as HU video games are, then my advice would be to include a stack reload option, so that you can feel more pressure as the time passes. Stacks get deeper, numbers are bigger, etc. It's very important. You can then achieve the same emotions during a game like it was small, but REAL money if you get to some rare point with big stacks. And also play for the high score.

But if it's just a training tool that works in some other way, I have no advice then at the moment 😀

Right now, if one of the players busts, a new game can be started. But the idea on having hero add to his stack is very doable.

Thanks for the suggestion