I think a lot of people would save time by having the option of just checking a list of what to do with their preflop hands. At least for opening this should be pretty straightforward, and I think doing something conditional for betsizes would work pretty well for the big blind.

Players would simply not see hands where either one folds at their first decision point and the hands they do see would usually be presented only after getting 3bet or getting to a flop.

At a guess I’m thinking this might speed up the game from anywhere around 10% to 1000%*.

* I take a conservative attitude towards my estimation abilities.

Well, it certainly isn’t up to me.

Personally I think 99% of the utility is in just being able to specify one thing for every hand. But sure, it would be neat to specify proportion of action with a given starting hand.


I thought about it a bit more and it seems like around 50% of total hands won’t come up for player review at all merely by automating the first two decision points. So playing 1k hands 500 of the hands will take approximately 0 seconds.

I think a feature of this method is that too tight play in the bb (vv common) is going to bleed someone out quickly, giving them quick feedback on the ranges involved.

As there’s typically 42 hour time limits it doesn’t seem like much of a bother to just edit the preferences when needed. Sure, there’ll be some hands autoplayed before you can change them, but not that many.

If you want to adapt every 10-20 hands I guess it’s a problem, but then you probably shouldn’t do something like this at all.

Personally I’d consider it a feature if I get to play with the exact same preflop strategy for thousands of hands and letting a computer doing the work. I am much more interested in seeing how different strategies turn out and focusing on the post flop play than fine tuning preflop strategy.

The reason why I haven’t started building this functionality yet is because I haven’t been able to decide how detailed it would need to be, so I really appreciate the ideas everyone have shared.

As I said, I think you get a big part of the benefit with a basic implementation. For instance just optional autofolding would remove a lot of hands from view.

I think specifying the ranges in the same format as pokerstove uses for preflop notation is a good way of handling input. It lets us freeride on the GUI and most people are familiar with pokerstove anyway. Solving the issue of mixed strategies should be fairly workable.

Re: opponent specific ranges I think that can be kludged by specifying the ranges for groups of tables and mostly doing invites.

I personally think the ability to define probabilities for different actions is a must.

Imo that’s much more important for real money play. Since this site is mostly about practice I’d be happy to take a hit in win-rate in order to increase the number of teachable moments per hour of play.

I also think the very act of formulating a strict preflop strategy, and seeing it played out, is more instructive than just grinding. It should help focusing on ranges and the implications of those. At any rate it is something you can’t get merely by playing a lot of cash hu. And I think a project like this should try to capture benefits not available through just regular poker.

As long as unspecified cases are passed along to the user I don’t think it can hurt much to just trying something out.

But the real problem is how to enter rules when someone has already raised.

Imo there’s really only a few reasonable open sizes limp, min-raise, 2.5, 3 and 3.5. Irl limps and big raises doesn’t really happen except against bad players (+ the very occasional high stakes guy like durrr). And most people will only use one raise-size and fold the rest.

The stack-sizes are something I did not consider, but I see two simplifications right away.

1) Bucket the effective stack sizes, and just fill out the buckets you think you will use frequently. There’s not much point in making differences between 4bb and 5bb if your main focus is 100bb and 200bb. You lose some finessing, but I really doubt anyone in actual play distinguishes between 100bb and 120bb. In fact I don’t see all that much adaption at the first two decision points between 100bb and 200bb.

Unless people want to practice shortstacking I think even something like 4 buckets could cover a lot of ground. It still leaves a lot of possible unspecified cases, but the important thing is to cover a lot of cases, not all.

Using this approach the button can cover something like 95% of his hands by specifying 4*3 (for those who limps) or 4*2 (for those who raise or fold) ranges. The big blind then get to specify 8*2, 8*3, 12*2 or 12*3 respones (well, some of those are against open-folds).

2) Even simpler is to just reset the stacksizes at the start of every hand. If you play a 100bb table *every* hand is played 100bb deep.

For real money this doesn’t make much sense, but for practice purposes I think it’s great. It allows you to specify what you want to practice and it saves the cognitive burden of checking the stack sizes at the start of every hand (it’s small, but that stuff piles up). It also still allows complete freedom to focus on whatever stacksize you want. Frankly I’d want this feature even if there were no automation at all.

This approach means that the small blind only has to specify 2 or 3 strategies and that the bb only has to respond to less than 25% of the scenarios from bucketing. it simplifies the game by quite a lot.

If this sounds like a good plan to start with, I can prioritize it very high on my list of things to build. So, what do you all think?

I think any implementation is better than no implementation, and feedback from play should clarify what the issues are. Money isn’t at stake so problems are no biggy.

My wishlist first implemention is one were the stacks reset after every hand, the input is in some sort of pokerstoveese and where you can specify strategies for a few number of sizes and leave the rest blank. To me, things like mixing hand use , opponent specificness and so on can come later on.

Great. Thanks for chipping in with your thoughts everyone.

We all agree that automating any of the repetitive routine plays is better than automating nothing. Therefore I will start working on it and begin with the situations and conditions that are the simplest to define and implement.

I also agree with everyone that having an option to auto-fold the absolute worst hands from the BB against any size of raise (including the min-raise), will help to speed up the game, so I will get that done. And regarding john’s concerns that this could lead to timing tells if those same hands will be auto-checked when the SB only limps, this will actually not be an issue since folding in BB and checking are logically completely different situations and the same rules cannot be applied to both (in other words even if you always fold a really weak hand from BB it does not mean you should check it 100% of the time when SB limps). So the user will need to create separate rules against a raise and against a limp, if he wishes to automate both situations. (Timing tells can still occur, but at least it will a choice made by the user if he only defines rules for some of the hands but not all of them.)

While I may not be able to convince everyone how important it is to mix up your plays and define probabilities for raising and limping with certain hands, I personally would have no use for "conditional moves" if probabilistic rules were not allowed, and therefore I will add that feature from the beginning.

And since several people have expressed they prefer having fixed stack sizes that are reset after every hand, I will be adding that option hopefully soon too.

I’ll let everyone know once I’ve made some progress with this and we can start testing it out.

One last thing I forgot to mention earlier that would make the "perfect" implementation even more complicated, is varying bet sizes. For example if someone wanted to raise with AA to 2.5xBB half of the time and 3xBB the other half. However, I think this may be the least important of all the features mentioned so far, so we can probably ignore that for now and just use a single fixed raise size for all hands. The size of the raise will of course be user defined, and can be different for SB and BB.

I think the auto-plays are working fine, but before I update the site menus and add the links to the auto-play settings page, I’d appreciate if you guys give it an initial try first. You can set up the plays in the following link: http://www.correspondencepoker.com/automatedplays.php

It doesn’t have nearly all the features mentioned (for example Pokerstove type of input format is missing), but it should let you get started. (The page layout may also look a bit weird if your browser or settings are not identical to mine, but I’ll work on fixing all that later.)