I've released version 2 (alpha!) of my range-based Texas Hold'em training software, Range vs. Range.

There are two training modes, Regular Training and Advanced Training. The features of Regular Training are:

• Online play against real people, with your results tracked
• Short, mid and deep-stacked play, 6-max training, heads-up training, and limit hold’em
• Train full hands or specific flop situations (starting play part way through a hand) like 3bet pots, defending the blinds, or wet boards
• Track your results for specific situations, compared to the rest of the population (users of Range vs. Range, playing that situation)
• See which side of a situation you are better at. For example, do you out-perform when defending the big blind from a button steal, but under-perform when you are the button and the big blind defends (compared to the rest of the population)?

Well, that’s just the basic stuff. Then there’s Advanced Training:

• All players' fold, check or call, bet or raise ranges are recorded throughout the hand. And these are their true ranges, because you are playing against real people who are free to play however they choose.
• Because Advanced Training keeps track of all ranges as they narrow through the hand, the pot is awarded proportionally at showdown. If your showdown range has 60% equity against your opponent’s showdown range, you will get 60% of the pot at showdown, and your opponent will get 40%. This cuts out most of the variance from regular poker, so your results from Advanced Training very quickly reflect the quality of your play.
• After the hand, you can review your opponents’ ranges (visually, there are screenshots below or externally) for every action they took and every action they didn’t take (for example, you can see their fold range even though they actually called in the hand). You can see exactly how they played the hand, with their entire range.
• After each hand, there’s in-depth quantitative analysis of all players’ play, so you can see how you exploited your opponent, got exploited, or could have exploited your opponent or been exploited. The following analysis is provided:
  • the fold equity of each bet made
  • for each bet made, the profitability of that bet as a bluff
  • for any bet that isn’t an immediately profitable bluff, the EV required to make it a profitable semibluff
  • for any bet or raise on the river, which hands (and what percentage) were bluffs, value bets, and medium-strength hands (which generally shouldn’t be bet)
  • for any call on the river, which hands would have been better as folds, and how much EV was lost by calling
  • for any fold on the river, which hands would have been better as calls, and how much EV was lost by folding
  • hands that were bet or raised as bluffs on the end, that would have had showdown value as checks or calls

With all of this analysis for every bet made in the hand, it’s very easy and quick for you to adjust your strategy towards stronger, more well-rounded play. Because you are playing your full range against your opponents’ full range, you cut out the misinformation and biases you normally get from variance in poker.

By quickly adjusting to your opponent’s play, and by playing against someone who is quickly adjusting to your play, you can quickly optimize your strategy towards optimal play, and start to focus on the small adjustments that can exploit and defend weaknesses in a near-optimal environment.

The website is at http://rangevsrange.com/.

If you want to try it, I strongly suggest you play Regular Training until you totally understand how that works, what situations are all about, and what the statistics/results mean. Then read the web page about Advanced Training and the Advanced Training Example before diving in. Like the name says: this is pretty advanced stuff. When I get around to writing the tutorial, it'll be here. There's also a FAQ here.

The very first version of this software went online about eight months ago, and since then it has rarely had 2 or more players online. So if lots of people start playing, I can't guarantee that the server will cope, because I haven't tested it under load. Probably only about 6,000 hands have been played on the server total, and most of them weren't with the latest version, so there may still be bugs. Just let me know if you find any, I'm usually pretty quick to fix them.

In fact, this thread should probably be called "Range vs. Range Version 2 Alpha Released".

Also, if you have any questions about how it works, or suggestions for improvements, I'm all ears. You can contact me here, PM, Gmail or Skype. I've thought about this topic a lot (as you'd expect) in the last couple of years but I'm still open to new ideas or different ways of thinking about things.

Here are some screenshots from Advanced Training.

The range chooser. In this case on the button, facing a cbet from a middle position raiser:


The review window, from a LHE hand I played against myself:


The analysis window, from the same LHE hand:

To add a bit more context, let me explain that Range vs. Range is about improving your strategy against strong opponents. It's about beating regulars, improving your winrate, and moving up in stakes, beating players who are more experienced than you. It's the first time you've been able to get objective feedback on your play. Every other training method (coaches, Leak Buster, videos) is about learning to play more like the other winning players, not objectively learning to play better.

You can probably tell I'm passionate about the subject. I'll try not to go on about it too much.

For a bit more context on how this came about: the first seed of it came when I saw Dwan stack Cates (and deep) in the durrrr challenge after calling a 4bet and flopping some ridiculous low full house when Cates held pocked Kings, and I thought How can you practice those situations? How can you know what's going to extract the most value there? My first thought was that you couldn't know, but I kept thinking. I was also reading Mathematics of Poker, and those who have read it know that it's all about ranges. Then somehow I came up with the idea of defining range-based situations.

Well, that was the starting point, then I got to coding. Now I think I've created something pretty useful. And something pretty revolutionary, too. And of course, there's a lot more to do...

Just a thought: if you log on and there's no one to play with, you can open the software a second time, create a second account, and play against yourself. Just to see what it's all about, and have a look at the features. That's how I've done most of my testing.

can i use this software on my mac?

No. I'm sorry about this, but I do this in my spare time and there's just too much other stuff for me to polish with it first.

Of course, there are some virtualization options, as a work-around.

Just a couple of questions.

How do you see which hands you had selected in a "bet block" so far? You can only see the percentage, but not the hands.

Is it possible to see the combo count as well, not just the percentage?

You can't see at the end what hands you had the whole time?

You can't modify your preflop range, you can only use predefined situations?

Hand analysis windows, it would be nice if the hand history was visible there as well.

How about profit for each hand on the river for your value hands?

->
On the turn, when Okhropir raises to 181 chips:
Okhropir risks 181 chips to win 81 chips. zombi folds 56.36%.
A bluff here makes an immediate loss of 33.33 chips.
(because 56.36% * 81 - 43.64% * 181 = -33.33)
For a semibluff to be profitable (compared to folding):
Okhropir needs an EV of 76.38 chips (or 18.95% from a pot of 403 chips, if the bet is called).
(because 33.33 / 43.64% = 76.38)
<-

Ok, so at this point, how do I find out which hands were good pushes for this situation, and which hands were not?

It only says that a bluff makes an immediate loss, but not which hands were those.

How do I see the ranges both players had at this point?

what is the advantage to using this over something like StoxEV?

You can't see the full list written down anywhere. You can only see it graphically. Apart from the obvious color-coding, you can see which combos are in which ranges by mousing over the cards you're interested in. For example, mouse over AKs to see which AKs are in your bet range, your check range, your fold range, and your unassigned range.
No, sorry. I knew people would ask for this, but no one has explained to me why this is worth having (apart from just percentages).
You mean your one true hand, that was actually dealt to you but you never shown to you? No. If you get to see that at the end of the hand, it makes it too easy to be results oriented – giving that hand more weight in your consideration of what you should have done, than the ones you didn't have (but just as easily could have!)
Yes. This leads into such a complicated conversation... but I'll try to keep it simple. When you play at the tables, the proper thought process is this: what do I have, what range do I expect my opponent to have, and what range do I
expect my opponent to expect me to have? Then what's the best move? The subtle point there is: it doesn't matter what your range truly is.

So when people want to put their true ranges into RvR, I say they're a bit misguided. Unless for some reason, you're used to playing against people who know your true ranges, but you're not adjusting, and you want to learn how to play that situation better. More likely, your perceived range is more like a standard range, and so these standard situations are what you need to practice.
There's just not enough room in the window, I guess. If you want to see both at once, you can. You can bring up a separate hand history window from the lobby. So you can put them side-by-side if you want to.
That can probably be done. But I'm not sure what you do with that information. I mean, I guess it'd be neat to know how profitable various hands are, but isn't it enough just to know that you played them right?

Thinking about it some more, perhaps there are situations where you want to compare how you played a hand, or a part of your range, across two hands, two different lines, and see which way was more profitable. This is something I'll think about some more.

To answer your second question first: you can review the ranges in the review tab. See the screenshots in this thread.

The other thing is something RvR doesn't do, at least not yet: when you get all in before the river, with cards to come, RvR doesn't do showdown analysis. It only does it on the river (all in or not). That's a deficiency of the current version. My only excuse is that it's hard: to figure out which hands have equity to call, and which hands were better shoves than folds (good semibluffs), in a range-vs-range all-in situation with cards to come.

(I'll probably get to it, but I don't expect anyone to care about what's yet to come.)

I kind of just want to lol at this, but I admit I've kind of been waiting for someone to ask. I look at CREV and think "I can't believe people paid money for that". Now I guess I just sound harsh.

To answer your question: as I understand it, CREV lets you figure out your EV, after making assumptions about your opponent's ranges. It takes ages to calculate the EV of a spot, it's only for a single hand in your range, and it's dependent on your assumptions.

Conversely, RvR allows you to play out a hand quite quickly (at least if you actually know your own strategy), allows you to play your whole range at the same time, and you're playing against a real opponent, so no assumptions are necessary.

Now I'm saying all this without having ever used CREV (and I don't plan to), so if I'm wrong, readers please correct me. (I guess the bit I'm most likely to be wrong about is the single-hand / full-range difference. I guess I can see how CREV might allow you to evaluate the EV of all hands in your range from a single point, in theory.)

Because the percentages you are displaying are relative percentages of the total range at that point.

Say, I added 20% of the range in the (value) bet range, and I now want to add 10% more for bluffs. How many hands is this? That 10% could be around 10 unsuited hands (if we started from a full range), or just one pair (if we started from a range of 60 combos). If you displayed combos, one could easily see now many hands is that.

Also, Janda does his counting in combos in his videos.


Ok, that's even better.


The ranges you put as default are overly complicated I think. You have ranges with hands with weights 9,7,5,1 or 4,3,2,1 for various hands, and I wanted to start with something really simple.




That information would be pure gold. If you could know which of your mistakes were the biggest in the terms of money. You do have all the information, so, for at least the river, or the last street played, you could show all the mistakes in the money terms, sorted by the largest to smallest.

Well I think it already does that. It tells you your mistakes, and how bad they were. It tells you when you bet a medium-strength hand, and how much equity it had against the calling range. It tells you when you fold a hand that you should have called, and how much EV you lost by calling. It tells you when you call with a hand that you should have folded, and how much EV you lost by folding. It tells you when you bluff-raise a hand that had equity to call, and how much equity it had against the betting range.

The only thing it doesn't do, which I thought was what you were asking, is tell you how much equity your (correct) value bets had against your opponent's calling range. Because I don't think that's something you need quantified. It tells you that you were correct to bet (or raise); I figure that's enough.

Via PM:
Sure, that's the psychological, self-control aspect of poker. If you sit down and figure out what your strategy should be, when you go to the tables maybe you'll tilt yourself into playing differently, and worse.

The way I see it, that's not my problem, that's not a problem for my software to solve. People can work on their strategy on RvR, and obviously it's up to them to take that to the tables. Yes, I've considered that you could take hands from Regular Training and then construct the player's strategy for use with the Advanced Training analysis. But that would be so slow, and variance would set in. As it stands, RvR takes a lot of the variance out of training.

When I've been playing on RvR, with Advanced Training, I've found myself thinking much more in terms of my ranges. Especially when I've played the very specific situations. Then when I take it to the Regular Training tables, I'm still thinking mostly about how I'm playing my range. So I think this problem, isn't too much of a problem.
You're not the first person to say this. I don't think it's a serious flaw. If you're really paranoid, just play against one of your poker buddies, don't play against randoms. Then again, randoms don't know who you are anyway. Let's say you show someone your entire strategy on RvR; OK, how does someone take advantage of that? They still don't know your screen name, or your real name.

Even if someone knew your strategy from RvR, well RvR is about improving your default game, your unexploitable game, you non-exploitive game. The strategy you work on, on RvR, it doesn't really matter that much if people know it. If they meet you at the tables, you're going to be adjusting from your default strategy anyway, and they're not going to know how. The only info that maybe they have on you, is that there's no point trying to exploit you. Maybe.
Not really. On RvR, you want your opponents to be as strong as possible. But... I get a lot out of playing against weaker opponents. It's good to experience exploiting weaker regs. That's what RvR is all about: beating the regulars. And it's fun. Playing stronger people is better for your strategy, but playing weaker people shows how well you're doing, when you beat them soundly.

But, what stops you from logging all the data, finding the most robust strategy and them botting it?

Nothing.

Look, if you want to find reasons not to trust me, you can. Like why haven't I put my name behind this? Why is it free? How do we know it doesn't contain a virus that installs a key-logger or network sniffer, that I'll use to steal all your money at the tables?

I'm sure there's more. But what it comes down to is, you have no guarantees. You never do, with any software, with any person. You just have to be cautious, be careful, and think it through. Personally, I never trust anyone 100%.

Made two accounts to try this out. I couldn't figure out how to change the ranges in the basic situations. So tried the advanced simulation. Every time I made a chance in one player's range, the other players range got messed up ('you're not done yet') so I actually never saw a flop.
Also, in limit, you could prefill the bet/raise amount. I'd actually prefer if it was a .50/1 game so any EV values come out directly in BB... but what ever.

Found this in the log when shutting down.

FWIW, making all the statements about CREV and then saying the quoted part doesn't really make much sense. You should know your competition better... at least as much as to try it out a bit. I'm not saying you're doing exactly the same stuff as CREV, but certainly would seem that much of the same concepts apply.

Hmm, if I was competing with them I'd have to agree with you.

By the way, I've revised my opinion of CREV a bit. I agreed that it can be useful, here. My current opinion is that it can be useful when trying to find an exact optimal solution in theoretical situations, because you can compare strategies.

Range vs. Range is more practical: useful for improving your strategy because the feedback you get is based explicitly on your strategy and what mistakes you're making.

You can't change the ranges in any situation. All you can do is choose a different situation.

Let me explain, because I hear this a lot.

First, understand this (I mean, if you don't agree, don't bother reading on): Your range has no influence on the profitability or correctness of a given play or strategy, for any given hand.

Instead, all that matters is your hand, your opponent's range, and your perceived range. (Potentially also: expected future mistakes.) Perceived range means what your opponent thinks your range is, which usually isn't precisely what your range really is.

Okay, so if you agree with that, here's my argument for inflexible, fixed situations. To train a realistic situation – for example, to get better at being preflop aggressor out of position in a 3bet pot – you need that situation to have a realistic perceived range for you, and a realistic range for your opponent.

Almost all strong players at reasonable stakes have similar perceived ranges. Other players don't know your range. Your perceived range comes primarily from the other players in the game. That's the reason why nitty play works at the micros: people don't give you credit because they don't realise that you aren't just another fish. But then at mid-stakes, people don't play so nitty because people catch on and know how to exploit nitty play, so they loosen up – they end up playing a closer strategy to everyone else, which is closer to optimal play. In both cases, micro and mid-stakes, you should expect people to put you on a fairly standard range.

If they knew your range, and your range was significantly out of the ordinary, wouldn't you adjust it anyway? So I don't think it's ever safe to assume that your opponent knows your range, when your range is significantly out of the ordinary. (And if they know your true range but you don't realise it... well, you're being out-played and you should probably go back to a less exploitable range... if only you realised.)

One last thing: remind yourself whether you agree with my premise. Because if you don't, that's where this discussion should start.

Yes. I haven't fixed this because it doesn't cause any problems, and I have more significant things to work on yet. For the interested, (this is the software forum) it happens because I'm not tearing down the various elements of the UI in the right order.

The latter part (allready constructed ranges messed up) was about the time when I tried to construct my bet/call/fold ranges for a situation. I was wondering if it was because I was logged on twice from the same pc.

Well, I read anyway.

When you give the ranges in the starting situations, I assume those are the perceived ones? Well, since I don't really agree with the FL scenario BB defense range, whether it's perceived or actual, it doesn't make sense to me to drill that out. The CO vs BTN 3! makes more sense to me now that I checked, so I'll be giving this another shot.

BTW, I think it would make sense to build a very tight range situation, i.e. UG raises, UG+1 3!, just to keep the ranges very manageable while getting used to the SW.

No. I test the software all the time in exactly this way. I can't tell what happened to you. I guess that you misunderstood the software, but it's possible that there is a bug. If so, grab a screenshot and post it here, or send it to me at rangevsrange on Skype or at Gmail.
Wait, do you agree or disagree? It really is important.
I worked out the LHE ranges with a LHE pro. In general I'm happy to change the ranges if they're not realistic, because I'm not an expert. But in the case of the fixed-limit situations, you'd have to make a strong case to convince me to change them (but feel free to make that argument. I will certainly listen, and it will be an interesting conversation).
If you think the situation is unrealistic, of course it makes no sense to train it. I agree. I'm always willing to discuss how realistic the situations are, and to adjust them on the basis of consensus.

But please realise, that a lot of the value will be lost if everyone can create their own ranges and train that. Firstly, this means that statistics can not be kept, and so you can't compare your results to those of the average player. Second, doing it this way allows for a significant amount of data to eventually show what's wrong with the ranges anyway. For example, we might eventually see that 22 is not worth a call in the BB. Or find that the bottom of our MP opening range is quite profitable (averaged over all players) and so decide to increase that range.