I've written this software that allows you to train in various situations. (It's for No-limit Holde'em only.) It's online software, meaning you connect to the server and train against other people. Currently, the following situations are available:

2-player, 2-bet flops:

• CO open vs. BB call
• MP open vs. BTN call
• BTN isoraise vs. UTG limp/call

3-player, 2-bet flops:

• UTG open vs. MP and BTN calls
• MP open vs. BTN and BB calls

2-player, 3-bet flops:

• BB 3bet vs. CO call
• BTN 3bet vs. MP call

3-player, 3-bet flops:

• BB 3bet vs. BTN and SB calls
• BTN 3bet vs. UTG and MP calls

Other (more experimental) situations:

• MP faces BTN on a board of K 8 3
• Heads-up, preflop

Later, you will be able to define your own training situations. That means you can figure out what part of your game has the most potential to improve your winrate, and train that.

The features of this software are:

• You are told your range and your opponent's range at the start of the situation.
• Each player is dealt a hand out of their range.
• You play out the hand from that situation.
• After the hand, you get to see the other player's cards. This means that you can quickly adjust to your opponent's play.
• You swap positions after each hand. This means you're learning about proper strategy for both players at the same time.
• Your statistics are recorded. This means you can see how you're doing in each position, compared to the EV of that position.

The software is a bit green, but that just means it lacks some of the bells and whistles you're used to in online poker.

Hmm, I guess I should say a little bit about why I think this software is useful. I'll keep it brief: most people (myself included) don't think enough about their opponent's range. They rely too much on "conventional wisdom", which is like a long, complicated set of rules and exceptions to those rules, detailing how you should play in every situation. Range is a concept that is somewhat neglected in that conventional wisdom. Also, this software presents another under-represented concept: taking into consideration your own perceived range.

In summary: this is good software for training range-based thinking; for training play against regulars; for learning what kinds of exploitable tendencies regulars have; and for training proper adjustments to your opponents.

I think it's really useful software. And it's free. Download the client at https://dl.dropbox.com/u/173372/client.zip.

This poker software is great! I had fun playing a HU scenarios 100 bb's deep with another player and, at the same time discussing our plays. I actually learned a few things in such a short session that probably would have took hours of trial and error in these particular spots. Discussing plays was probably the best part I liked about this software, sharing thoughts and ideas is so great for improving my game. I can see this software being used as a coaching tool also. It needs more players however so I encourage people to check it out and idle in the lobbies if they like it to help it grow a bit more. Overall great software =D

Awesome, I'll try it out and tell you what I think


EDIT: Oh I didn't realize it was like an online software. That's interesting. I guess I have to wait til you get on or something since I'm the only one online right now

Have you thought about making a version that was basically you vs. a computer? I'm not sure how you could program that.. I'm sure you could easily program it for calling a bet vs. a range on a particular board.. You would just calculate the pot odds on the call, and what the computer's total equity is vs the Hero's range on a certain board.. I'm not exactly positive on how you would calculate it for betting.. I think you can bet with 35% total equity (fold equity + hand equity) if your bet will work at least 35% of the time.. You'd basically be creating a new PokerStove that will allow you to play the hands in the ranges specified.. and the computer would actually bet, call, or check depending on his equity against the Hero's range on a specific street.. Example: The computer has a 25% range preflop. Hero calls with only a 10% range, but it's filled with suited connectors and pairs. The flop is Jd 6c 3s The villain (computer) will bet a lot on this flop since he has a lot of fold equity (I found the formula for FE. Collin Moshman responded to someone's question.. It's (Hand Combos they call you with / Hand Combos in their current range) x 100.. We call. Turn is a Tc.. The PC will either check or bet depending on how much equity he has in his range against our slim calling range. You'd take our 10% range and just subtract hands that fold the flop.. It would be a huge project, but would possibly revolutionize poker because everyone would have access (I don't mean freeware/purchase wise) to a tool that allows you to practice repetitively, and you would get to see the hand you played against.. I think it would be epic..

but that's if we could learn from a computer that basically plays perfectly. Would it teach us that well? I think probably, but it would be pushing every possible edge, and there are clearly players that don't come close to that.. So we would be adjusting to a PC that plays perfectly based on ranges. (The game currently assigns hero a range, and we are only dealt hands in that range correct? If so, I believe that means the PC would actually be able to play GTO vs us when we are HU since the ranges are exact, and we will never play a hand outside of the ranges assigned to us.)

Well, I'm online now at least. Finding someone else to play against is probably the major challenge I'm facing in making this software. One thing I have done is, you can subscribe to email notifications, so that when someone logs on, you get an email (so if you're available, you know there's a game available too).

Yes, and I tried a little, but I concluded that it's not a good idea because:

• The computer player would have to be adaptive. If you are just playing against a static "quite good" strategy, you could be training yourself to play well against that particular strategy, but also training yourself to have massive exploitable weaknesses, because your opponent is not noticing and adapting to those weaknesses to train you out of them. Because Range vs. Range is human-vs-human, even though we're not always playing against geniuses, we are always having our (obvious) weaknesses exploited.
• The computer doesn't have any ideas. The computer can't explain why it chose a particular action. You can't have a strategy discussion with the computer, but you can with your human opponent on RvR. I can't over-emphasise this: Range vs. Range is a really good place for strategy discussion.
• It's really hard. It has to vary its bet size; it has to slowplay sometimes; it has to turn made hands into bluffs sometimes; it has to checkraise sometimes. But it has to balance all of these actions so that none are exploitable. No one has come close to making a decent no-limit hold'em AI as far as I know, and if I tried (solo) there's no way I'd come even as close as those who have previously tried. But that's okay, because I don't want to anyway, for the reasons above.

Don't you see? That's what Range vs. Range is now. A tool that allows you to practice a situation repetitively. Any situation. Until you perfect your strategy.

But it doesn't allow you to practice mindlessly. It doesn't let you practice "I just called with 77 in the big blind". It forces you to train your whole strategy - your whole range, in consideration of your opponent's range. And it goes one step further: every second hand you are in your opponent's shoes, learning how he is (or should be) playing his range against yours.

Yes, that's how it works. Hero is dealt a hand out of hero's range; Villain is dealt a hand out of villain's range; Hero and villain both know both hero's and villain's ranges. But you just can't make a computer use this information to play GTO. Or at least, I haven't seen anyone explain how. I have enough trouble trying to figure out how to play near-optimal myself, in a specific situation, let alone generalising and codifying that understanding so that a computer can do it in all situations!

You know, it would be interesting (if frustrating) to train against a GTO opponent. You'd lose money, and the challenge would be to lose as little as possible. (And if you were ever up, you'd know it was purely variance.)

But you'd lose one of the benefits of Range vs. Range there too. The role of GTO play in the strategy of someone like you or me is only as a starting point against a good player. But where we want to be is playing "near-optimally", meaning that it's mostly GTO, but we're deviating in small ways that aren't particularly noticeable, or exploitable, but that are exploiting the observed weakness in our opponent.

Firstly congrats on creating this fantastic software. My mate and I played some today and we found it incredibly beneficial both for study/discussion purposes and super fun to boot.

Some feature requests:
- Please create a Limit holdem mode with fixed betting. (we are LHE players. it works as is but need to manually adjust the turn and river bets)

- most important in my view: please allow ranges and situations to be created by the user. The current ranges are much too tight for us LHE players

- minor: change the bet/call/ fold button colour to blue or something so its easier to know when it your turn. Since they are small I sometimes don't notice when they have popped up.

oh one more feature request:

- display equity of our hand vs villains range on the flop (and perhaps our preflop equity too). obviously this becomes meaningless once any bets go in, but this would still be invaluable training imo.

There needs to be a 1 player mode for this.

I saw this software months ago and the problem back then is the same problem now.. noone uses it! So I couldnt say either way if this software is good or not as I have never actually been able to use it.

The 2 player mode is a great idea, but who is hanging around on the off chance that someone else decided to use this software.

At least if there was a 1 player mode people would be able to mess around with that until someone else comes online and that way you would build a userbase.

find a buddy to play with you!

So in order to use this I have make arrangements with people and pre-arange a time with a friend?

This software has been around for months. Its not catching on. The reason is people just dont really do that, hence when you logon there are 0 users.

This will continue to be an issue until the software catches on. In the mean time, I have created an option where you can receive a notification email when someone logs on.

This way (as long as people don't disconnect immediately after logging on!), people will be able to get together to play.

I'm glad to hear you found the software useful and fun! Thanks for your feedback.

I'm happy to implement LHE. Of course it should be easy for me to do: the client won't need to be changed; the server just has to give different options to the user. It would be fun to see how range-based thinking applies to LHE. Man, edges are going to be so small!

But implementing it will be the easy part for me. I need your help in creating the situations. If you give me the ranges, pot size, positions, etc., I can easily input the situations into the server.

Give me the ranges, and I'll create the situations. But when defining a situation, it's best not to put hands only in the always-in or always-out categories. If you give me ranges, specify which hands are:

• Always in. I think an example of this would be medium pairs calling an opening raise. You're never folding these, but never 3-betting them.
• Usually in. KQs on the button calling a bet would be an example of this. Sometimes, people will raise. Mostly, people will call. (Correct me if I'm wrong of course, I'm no LHE expert.)
• Sometimes in. Speculative hands. Speculative opens, speculative calls, speculative 3bets, hands so weak you'd usually fold to a 3bet when out of position (does that happen?).
• Never in (of course).

I will change the action buttons' colours. I think I can probably do that today. I've been trying to figure out how to make it more obvious that it's your turn to act, and I think colour is a good idea. I'm going to go with red=fold, yellow=call, green=raise, but of course I'm open to opinions.

I can see how this is more relevant to LHE than NLHE. I will be doing this, but only in the medium-to-long term. (It's part of a broader effort to create a range editor, range calculator and equity indicator.) In the mean time, you're going to have to make do with PokerStove (or other similar software).

Thanks again for your feedback!

I've done the colour thing. Also made those action buttons a little bigger.

Thanks for implementing my feature requests so quickly.

Here are a few situations to start with. (yeh I know these ranges look nuts to a nl player . btw cold calling other than sometimes in the SB is a big no no in LHE, 3 betting is standard. Also, raising and then folding to a 3bet never happens in LHE)

Situation 1:

CO opens: 22+, A2s/A4o, K4s/K7o, Q7s/Q9o, J7s/J9o, T7s/T9o, 96s, 86s, 75s, 65s

BTN 3 bets (always): 44+, A2s/A5o, K6s/KTo, Q8s/QTo, J8s/JTo, T8s, 97s, 86s, 76s

Btn 3bets (sometimes): 22-33, A2o-A4o, K4s-K5s, K8o-K9o, Q6s-Q7s, Q9o, J7s,J9o,T9o, 96s, 85s, 75s, 65s

CO calls his entire range
----------
Situation 2:

Btn opens always: 22, A2+, K6o/K2s, Q8o/Q4s, J8o/J6s, T9o/T6s, 96s, 86s, 75s, 65s, 54s

Btn opens sometimes: K6o, Q3s/Q6o, J7o/J5s, T8o, 98o, 87o, 64s,43s

Bb calls always: 22, A2, K2, Q2s/Q5o, J2s/ J7o, T2s/T6o, 92s/ 96o, 82s/86o, 72s/75o, 62s/65o, 52s,43s, 32s

(no need for sometimes call)

Situation 3:

CO opens, btn 3bets (same ranges as above).

Now SB caps always: 77+, AJo/ATs, KJs, KQo, QJs,JTs
Sometimes: 66, ATo, KJo/ KTs, QJo

Rapala, you forgot to take the 4bet hands out of CO's calling-3bet range for Situation 1, and BB's 3bet range out of his flatting range for Situation 2. Can you do that? Your guess will be waaay better than mine, evidently. (Note that you may want to move some of the nearly-premium ones from "always" where they are now to "sometimes" because they'll sometimes just flat the 3bet - perhaps?)

Also one more important thing, because again I value your opinion on this higher than mine: how much more often should you see an "always" hand than a "sometimes" hand. I.e. could I assume something like "sometimes" = 50% of "always"?

(Also, two questions out of curiosity:

• Why does SB cap 77 vs. a CO open and BTN raise? What's wrong with flatting?
• What's wrong with the BTN flatting a bet, to make it such a big no-no?)

It turns out I don't even know the rules of limit hold'em. In this situation:

• BB bets flop
• CO raises
• BTN 3bets
• BB caps
• CO folds to the 4bet

It's now on BTN, facing a 4bet. BB capped the betting. But now it's heads-up. Does this mean that the cap can be removed, and BTN can reraise?

Or is it only for streets that start-off heads-up?

Or is it only if the 4bet was made while already heads-up?

Or is it any time you're heads up, even if the 4bet wasn't?

Limit hold'em situations are up online. This includes:

• probably awful assumptions about what hands BB would have 3bet vs. BTN, and what hands CO would have 4bet vs. BTN, eagerly awaiting revision
• implementing fixed-limit betting, big bets on later streets, remembering bet counts, capped betting, removing betting cap when heads up

I've only tried them out on my development laptop against myself. Given the whacky (to my NL eyes) ranges like BB defending with 80% of hands, I'm really looking forward to trying these out. I think they're going to be fun.

I think you missed that I said CO calls with his entire range, even AA. This is a common strategy when out of position as btn will c bet almost 100% and CO can check raise his monsters.

I think ‘always’ should be 100% to simplify things and would it be possible randomise the sometimes range between say 20%-50%? If this is a bit complicated then I would prolly say something like 33% as a hard number.

In a vacuum, flatting 77 might have higher ev than capping, however your flatting range will be so narrow that a competent villain can instantly put you on just a few hands and outplay you postflop. My strategy in that spot is to always cap or fold. There are some flatting spots, but we can get to those later.

Same as above. Also we gain initiative in position and fold out the blinds which is hugely valuable. I think the big difference between lhe and nl here is that we don’t have to fear a 4bet we might have to fold to, cos even if we 3bet 42s, we have odds to call the 4bet.

Oops, I forgot about that. (although I have played with a call 100% of my range strategy in the past.)

situation 2: Bb call vs btn raise

Ok so the adjusted range: always: 22-66, A2s-A6s, A2o- A8o, K2s-K8s, K2o-K9o, Q2s – Q8s, Q5o – Q9o, J2s – J8s, J7o – J9o, T2s – T7s, 92s- 96s, 82s-85s, 72s, 75s, 62s-64s, 52s-53s, 43s, 32s

Sometimes: 77,A7s-A9s,A9o,K9s,Q9s,QTo,J9s, JTo, T8s, 97s, 86s, 65s, 54s

sometimes should be ~40%-70% or 60% if you need a hard number.

It varies by site but I think its best to use the most common structure (which Stars uses) which is that 4 bets is the cap even when hu. (with the alternative rules, you could keep reraising in the situation you described, but let not use that format I would think)

Awesome!

btw it doesnt make a difference in actual play, but the 3 way situation is actually SB capping, not BB. BB would have a different range to the one listed.

Ok, so I just have to clarify:

• 74s and 73s are "never"s surrounded by "always"es. I think you meant "72s-75s" not "72s, 75s". Ok?
• You've removed 43s and 32s entirely
• J6o, T9o-T7o, 98o-97o, 87o are now removed

sorry i messed up there. The T9o category was supposed to stay in there.

I've had a look at my database to see how I play there for clarification.

please make:

always call: 83s-85s, 74s-75s, 63s-64s, 53s-54s,43s, T9o -T7o, 98o-97o, 87o

sometimes call: 82s, 72s-73s, 62s, 65s, 52s, 42s, 32s

I know I've muddied the waters a bit now by saying some of the 'sometimes call' is now on the bottom side of the range but I think it still works out ok.

So I have always: 66-22,A6s-A2s,K8s-K2s,Q8s-Q2s,J8s-J2s,T7s-T2s,96s-92s,85s-83s,75s-74s,64s-63s,53s+,43s,A8o-A2o,K9o-K2o,Q9o-Q5o,J9o-J7o,T7o+,97o+,87o

Sometimes: 77,A9s-A7s,K9s,Q9s,J9s,T8s,97s,86s,82s,73s-72s,65s,62s,54s,52s,42s,32s,A9o,QTo,JTo,96o,86o,75 o+,65o

If we have sometimes = 50%, then this is defending 49.2% of the time.

(You were a little ambiguous about 96o-76o, 75o-65o, so I put them in the sometimes pile.)

what makes this any better than equity training in equilab?

It' not the same concept at all.

Tell me what Equilab teaches you, and I'll explain the differences.

have you ever played with equilab?

equilab trains you range vs range or hand vs range, gives you an analysis of turn and river cards and runs the combinatrics of range or hand hitting the flop.

all without being online and need an opponent

Please add NLH HU: BB 3b, BTN calls (custom ranges would be grate)