Well, I ran though an advanced scenario. Took quite a while.

Some (mostly) usability stuff:

"Hands xxx called that were weak enough to fold:

3h Ah, with equity 0.00%, EV-diff -4.00 chips
Ad 3d, with equity 0.00%, EV-diff -4.00 chips
Ac 3c, with equity 0.00%, EV-diff -4.00 chip
"
That's from the final review. I suppose the equities are directly against villain's betting range? All the wordings like "medium-strength", "weak", "value hands" are against the strategy villain happened to choose?

Keeping up with suits is pretty difficult (when maybe calling QsJh but folding the rest, raising with KhTs but folding the rest)... also finding where some unassigned hands are lurking. More color coding would help, also a way to toggle between showing only fold, call, raise, ua ranges. A color key would be good, also for the review window. Showing the ranges side by side for each review step might also be useful, now they pop back and forth between the players.

When I was finished with the scenario, I got the analysis window and a new hand popped up. What I wanted to do is back up a bit and fine tune some bluff hands into one player's range to see how it would work. Couldn't be done.

Might also be interesting if you could display something like "Ah9s has 35% equity against villain's betting range" on the fly, when it's my turn to pick hands.

Stuff like "xxx folded 65% of his range on turn, which could be exploited with the pot and bet sizes" might be helpful to some. Then again, it only takes a calculator for someone who might be using this. This might work for people who are starting out to think in ranges without a theory or math background... actually I could see this more as a teaching tool than "learn with friends" tool... if said friend chooses ranges badly, that distorts the lines I should pick if I understood this right.

I'm really not sure what I learned from this. Made me think about my ranges for a scenario, but I could do that before. Maybe I'm still missing something.

Disagree. I must say I haven't given this too much thought before your comment "only perceived range matters", so I could be wrong. I'll give a short example of how I see this, feel free to fill the gaps.

If my perceived range on the river is such that I have 90% nuts, 10% bluffs and I bet all in (1 PSB), villain's optimal play is to fold all his hands but the nuts. If my actual range is 10% nuts, 10% 2nd nuts and 80% bluff, well, I bet all my bluffs (and nuts, hoping for a missclick). There's actually no need to bet the second nuts, as a missclick is likely way more infrequent than villain actually having the nuts.

It's pretty common for BB to call all his range in a FL steal situation.

Don't want to sound snarky, but with the current user amounts, it'll take quite a while for the statistics to become meaningful. It's not a matter of playing a scenario 100 times... it's a matter of playing a scenario with different runs of the board a gazillion times. But I see your point.

Hey I was online for awhile tonight under 'tbb'.

I really like the software and the concept but it could definitely use some improvements.

I mentioned some of this stuff earlier when we were online today but just thought I'd reiterate just in case it gets forgotten.

-Display number of combos next to the percentages for the different allocations ranges, as well as a total running number of combos somewhere.

-Selecting hands and allocating them needs to be a bit easier imo so that the process is quicker. This would help get more people to try it out and stick with it. Not sure what to tell you but it should be more graphical when dealing with all the different suits. I'm thinking like a drag and drop type functionality where you can visually see the hands in each box for check/call/raise.

As it stands, it does take some getting used to but it is still a great piece of software. I do appreciate it. My only qualms are that it could be prettier and that there is rarely people online to practice with. Perhaps you should organize a group that goes online at certain times.

I think you are, yes... Yes. The idea of Range vs. Range is that you're always playing a realistic situation. You're always playing a hand of poker (it's just that with Advanced Training you don't know what your hand is). You're always playing against real people, who are doing their best to beat you. And in so doing, you can play your best poker, and when the software identifies your mistakes, you have real feedback that you can incorporate into your strategy. The software won't tell you what the GTO play is; it will only tell you how you could have played better against this Villain. Over time, this lets both players polish their strategies towards GTO play, in a way that is uniquely appropriate to each player.
If you want to call QsJh: 1) Click "QJo"; 2) Click "None" (bottom right) to deselect all suit combos; 3) Under "Select offsuit:", click "SH" to re-select Spade-Heart suit combos.

If you want to find where unassigned hands are lurking, click "Go", and it will select all unassigned hands. You can then choose to fold, call or bet with all of them with one click.
Some good ideas here. I originally had the toggling of fold, call, raise, UA ranges, but I changed it to this so that you could see them all at once. I'll think about whether the two ways of doing things can be combined in a way that doesn't make it more complicated.
That's right. Your opponent's ranges are secret until after the hand, and after the hand you can no longer play it out against him. Even if I coded it so you could replay some of a hand, it wouldn't be the same because you'd know his ranges (at that point in the hand) better than you did the first time.
But you don't know Villain's betting range. Would you want to? One of the features of RvR Advanced Training in my opinion is that you get to see exactly how good your hand reading is, because ranges are known at showdown. (Think about it: how else can you ever know if your hand reading is good, when playing at the tables? If you give Villain too narrow a range you'll one day see him prove you wrong, but if you always give Villain to wide a range, you'll never have any hard evidence to the contrary.)
The software does this. It tells you your fold equity, how profitable a bluff was, and how much equity you needed to semibluff, for each bet.
Yes. If your friend is weak, you can get better at playing against weak opponents. If your friend is strong, you can get better at playing against strong opponents. Ideal is to have both weak friends and strong friends. If you can only choose one, strong is better. But if you only have weak friends, you can still learn a lot, like how to optimally exploit them.
You haven't really made much of an argument here. I know what you're saying: if your perceived range is 90% nuts, your strategy should be to bluff everything you can. My point is, if you hold the nuts, all you need to know is that your perceived range is 90% nuts; if you hold the 2nd nuts, all you need to know is that your perceived range is 90% nuts; if you have a weak hand, all you need to know is that your range is 90% nuts. That's all you need to know to decide how to play. You don't need to know that your range is actually 10%/10%/80%. It's not relevant to how you will play any particular hand. All that matters is the hand you hold, your perceived range, and your opponent's range.
But is it reasonable to assume that BB is doing that, when he might not be? Isn't it better to assume that BB is folding a reasonable number of hands? He'll punish himself by calling with 72o, we don't have to do it for him. If I'm going to train one way or the other, I'd rather train for him folding some hands, because that makes my play less exploitable, and still wins money from people who play too tight, or too loose.
Snarky is fine

How about you organize a group that goes online at certain times, and I'll work on those features you want?

I meant call with all of his continuing range. As in "never 3! OOP when facing a steal".

I'm pretty sure it starts oscillating between maximally exploitive plays, not GTO.

Yeah I figured out how to do things, but didn't figure how to visualise my different ranges (practice and familiarity with the color coding might help). I didn't notice the unnassigned hands were selected (and I believe LazyAce missed that too).

If this is a game, no. If this is a study tool, trying to learn about GTO/exploitation, yes.

I believe it told LazyAce that he shouldn't have folded AK at showdown analysis, yes. I meant on the earlier streets, too.

As I said I may need to think this over. Perceived range in a scenario (at the starting point) is the starting range that you gave?
If there's a hand in my actual range that's not included in the starting range, I can't train playing that. Or if there are hands in the starting range that I wouldn't play like that, I don't need to train playing them.

Couldn't edit any more. I meant as an option.

Only if both people choose to play that way. If either one chooses to play closer to optimal and make small adjustments, then they will win: they will capitalize much more on their opponent's largely exploitable play, and lose only minimally to their opponent's minor exploitation of their play.

Yes, but I don't expect this will happen. Let me know if it does. The ranges are defined in the situations with weights, and hands that people might play but normally don't are given low weights, but not removed entirely. I think you'll find that if you play a hand (as standard, against unknowns or people who know your game well), it'll be in. Again, let me know if any aren't.

But if you want to train the specific exploitative range that you use against a particular fish, without him knowing that you use that range in that situation, you can't do that on RvR. RvR is more for training near-optimal play. Like I said on the FAQ: you're already good at taking fish's money. Range vs. Range is about playing better against better players.

IMO, it doesn't matter if there is a hand or two, or three missing from your actual range in a RVR situation. From what I can tell, the point of the software isn't for you to train how to specifically play a specific hand on a specific board texture. The benefit of this software is to practice balancing your play, to think more in terms of your overall range instead of the cards in your hand, recognize where you might be exploited and where you can exploit your villains.

Maybe this shouldn't be in the SW thread but somewhere in theory. Anyhow, some more terms should be clarified. When you say "chooses to play closer to optimal" do you mean closer to GTO, or closer to optimal against villains (non-perfect) strategy?

When villain bluffs too little, it's not optimal to play GTO against him.

I'm confused... how does that work out, when the calculations (you should have folded this, you shouldn't have bet that) are based on villain's actual play, not GTO play?

I (and most other online FL regs I've met) tend to never 3! against a steal OOP in BB.

Yeah I mostly agree... but if a part of my range (that I'm used to playing with) is missing, it would affect how I can play a specific hand that is included.

If the flop is 234ss or QJTss and I never have sets or straights there, it'll make a big difference on how I can play my (semi)bluffs like flush draws in the spot.

OK. When you open BTN and BB calls, what do you put him on (assuming unknown, but with some indication he's a decent player)?

When you open BTN and BB 3bets, what do you put him on (assuming the same)?

If you're interested, you can see the thread where Rapala and I discussed the current Limit ranges starting here.

If you can do the same, I can create no-3bet limit situations.

Overexploiting people is dangerous. Underexploiting people is safe. When you see that your opponent is exploitable, it's best not to try to maximally exploit them. Think about that for a second and see if you agree.

If you overexploit someone, you're very vulnerable to being exploited yourself. When you underexploit someone, you don't make as much profit from their non-optimal play as you could. But the really important point, the thing that makes underexploting the right thing to do, is that at the point that you realise that your opponent is exploitable, so does he realise that you now know he's exploitable. So 9 times out of 10, he's going to adjust towards optimal play, or even metagame you and adjust to exploit the way he thinks you're going to adjust to him. That's why overexploiting people is so dangerous.

(BTW I just made up the words overexploit and underexploit, but I think I could have equally said overadjust and underadjust. It's almost the same, except that underexploit is more what you attempt to achieve, while underadjust is what you do to make it happen.)

(And yes, when I say "closer to optimal", I mean closer to GTO.)

So my software doesn't say "you should have folded this, you shouldn't have folded that". There's no "should" in it. It simply says "these hands were strong enough to call, those hands were weak enough to fold". This time. It lets you know how your opponent – and yourself – could have been exploited in that hand. It's up to you to guess how your opponent can best be exploited next hand.

Top 70%, which is pretty close to what Rapala gave in post #13.

Once he 3!, I'm not assuming an unknown to be decent. (Not saying that if someone 3!, he's never decent... there are good regs who still play like that, and some like to mix it up, and some may have a good reason to 3! me OOP.)

Anyway, once they do 3!, I'd say 55+(99+ for some),AT+, KJ+,QJ+, JTs, T9s. Add a pip for suits maybe. Mixing it up to "sometimes" ranges isn't that important in limit, imo. Once you look at that range, and realize there's next to none folding equity against anything resembling a showdown value (that would often be stuff like good Kings and better), I'm sure you'll figure out why 3-betting OOP has gone out of fashion Also, where this will have a strong effect is the times he only called. (3! from SB is a different beast as there's now a good chance of getting a full BB of dead money in the pot - but now regs don't have a CC range so there's not the problem of splitting up one's ranges so that they'd be easy to play against).

All this is based on
-villain having memory (in practical terms, the game going on for longer than this hand, or more accurately, longer that it takes for the next steps to materialize)
-villain having the capability to notice he's being exploited
-villain having the capability to counter exploit or move towards GTO.

But given a situation where all that applies, I agree.

Ok I switched 'strong enough to call' to 'shouldn't have folded' due to lack of short term memory... but I don't see much difference (if leaving out the rare bluff raise that comes with my choice of words).

Still confused how RvR helps to gear towards near-GTO play, when the judging 'strong enough to call' and such are based on what this villain happened to do with his range. Like you said, it tells how hero or villain could have been exploited. Unless of course if we assume villain to be good enough to be playing GTO.

Sorry, I mustn't have been clear enough on the distinction I was trying to make. The point is that there's no "should" about it. Range vs. Range doesn't tell you how to play. It just tells you what you could have done better, how you were exploitable, and how your opponent was exploitable.

It doesn't say "You should have done X." Because perhaps your opponent went crazy on the river and played super-exploitably. But that's not to say that the right decision for your river play would have been to adjust to that in advance. You just couldn't know that. (It's a subtle distinction, I know, but in this particular project, I've been forced to be very particular about details.) Instead, the software lets you know what exploitability there was in the hand, and it's up to you to decide how you ought to adjust to that.

It also doesn't say "You should do X next time." Because like I've been saying, adjusting less than completely is probably best.

Objective information is given to you. You decide what to do with it.

That's right, RvR teaches you how to play more optimally, and also teaches you how to exploit opponents in a near-optimal environment. Or perhaps I should say it lets you train playing more optimally, and lets you train exploiting opponents in a near-optimal environment, because like I said above, it doesn't tell you what to do, it just gives you enough information to figure it out for yourself, fairly easily.

RvR lets you train towards GTO play, if both players adjust to each other, and both players choose to underexploit each other.

RvR lets you train exploiting non-GTO play, if your opponent doesn't adjust, or overadjusts.

Both of these are good things, things you will want to train.

The other point worth making, which I think you're neglecting, is that it's not just about "how can I exploit my opponent." It's also about "how am I exploitable." Even if you choose not to defend against that exploitability because you don't think your opponent is counter-adjusting: getting continual feedback on how you are exploitable is very valuable. Because clearly, when you're playing poker, you want to have a good idea of how you are exploitable, if you are exploitable, so you can see when people start trying to exploit you.

great software.

I think a good option would be to allow the hand go to the river every time.

It's kind of frustrating when you spend 20 minutes on building a range on the flop, then the hand is folded.

If it were to go to the river every time, you could analyze street by street in review and see in what spots you're losing money vs villains range.

I've mostly written this feature, which will constitute the next version. It'll be ready in a week or two. It's more than you described though, in two lovely ways:

• It still records result appropriately. If you fold 30% for example, 30% of the pot will go to your opponent straight away, and you'll play on (as a call or raise as appropriate) for the remaining 70% of the pot. So while you can choose to play really tight, and end up with massive equity at showdown, you'll be winning very little because you folded so much of the pot to your opponent earlier. So the statistics and results recorded for you are still accurate. In fact, they'll be more accurate than now, because variance will be much lower.
• You get to play on, even at the river, meaning that you'll usually get multiple showdowns in a hand. Before the river, you'll never fold, so you can play on; On the river, you'll never fold or call, so you can play on. (Exceptions: when you choose to only fold or call with your entire range, or when card removal effects mean that your effective raising range is empty.) This is great, because showdown analysis is where you get the most feedback: it's where the most money is made and lost, and the only true test of your hand-reading ability.

Considering that more hands will end up calling than raising on the river , that doesn't look like an improvement to me. Unless the 'call' branch showdowns are available too.

Also, it would be good to be able to get back to a hand I played some time ago. i.e. hand history.