Every reg deviates from the solver. Cbets more often because "the pool overfolds". Calls the river less often because "the pool doesn't bluff". Probes the turn more often because "after check-check nobody has anything". Some of these deviations make money. The rest only seem to.
Comparing frequencies with the solver without the pool cannot see this difference. It sees that you cbet 95% instead of the solver's 71% and writes it down as a leak. But if your pool folds to the cbet often enough, it is not a leak but an exploit. And if it doesn't fold, it is a leak, no matter that "everybody plays it this way".
There is only one way to tell one from the other: play both strategies, yours and the solver's, against the same pool — the regs at your tables, from your own database. For this we are building the Play like solver tab in the LeakPrice account. Below is how it calculates, what it will show and what a trial run on a live database showed.
How it is calculated
- The pool. The regs at your tables, from your own database. Their frequencies are locked in the solver wherever there are enough hands, and the hands behind each bet are taken by showdowns.
- Your line. Your frequencies by hand category from the flop to showdown: air, gutshot, weak, middle and top pair, overpair, straight and flush draws, two pair and better. Not one overall cbet frequency, but "how often you bet top pair and how often air".
- The solver's line in your seat against the same pool.
- The difference — how much more or less than you the solver would win in each situation and on each street.
Why the usual "solver vs solver" is not enough, we covered in the article "Solver vs the pool". If you lock frequencies alone without hand composition, the exploit comes out inflated. If you ignore that the opponent adjusts, the solver finds "check-raise your whole range", which never happens at the table. So there is no "best response to the pool" here: two ready strategies are compared, yours and the solver's, against the same pool.
What the tab will show
Where the solver beats you. Only what is proven: the whole price interval is below zero. Everything else is marked "needs more hands". Click a spot and the breakdown opens: your frequencies next to the solver's by hand category and one line on what to change. For example:
River, you are in position facing a bet of up to 50% pot. Raise two pair and better more often: you raise 42%, the solver 71%.
Where you beat the solver. Spots where your game against this pool earns more than the solver's. This is an exploit: keep it, don't fix it to match the solver.
How the pool plays against the solver. Frequencies of the regs at your tables next to the solver's in the same situations. You can see where the pool makes mistakes and in which direction. For example, the BB facing a delayed 33% cbet on the turn folds 51% vs the solver's 39% (an example from the article author's trial run).
By flop. On each flop — the three most expensive spots and the hands where you differ from the solver. A single flop is an example, not a verdict: the "better" or "worse" conclusion is made for the spot as a whole.
A month later — "before → after". On new hands, next to each spot you will see whether your minus has narrowed and whether the exploits remain.
A trial run on a live database
An NL50 player, SRP BTN vs BB, against the pool of NL50 regs. This is a trial run on 60 flops, and the interval method has changed since, so below are only directions for the spot, without a price.
Where the player beats the solver.
- BTN cbet. They bet almost always: on J♠9♥3♦ — 98% vs the solver's 71%. A comparison without the pool would call this a leak. But against this pool a frequent cbet beats the solver for the spot as a whole. J♠9♥3♦ is the exception: there it loses.
- BB turn probe after a checked flop. They bet 49–66%, the solver 24–41%. In the common LeakPrice pool of regs (NL200–NL5000) regs fold to the probe 6 points more often than the solver, and against the NL50 pool the frequent probe also turned out to be better.
- BB defense against a 33% cbet. On Q♠7♥6♥ the pool cbets 59%, the solver 47%. The player folds 16% vs the solver's 30% and check-raises 18% vs 12%, and for the spot this beats the solver. In a similar spot (CO vs BB), with aggressive regs locked, the solver also defends wider, though there is little money in it.
Where the player loses.
- BTN folding to a check-raise after a 33% cbet. On dry flops they fold 65–73% vs the solver's 40–44% and fold top pair 27–41% of the time, while the solver never folds it. Meanwhile the pool check-raises only 1–4 points more often than the solver, so there is no reason to fold that often.
- Folding to a small river bet. 64% vs the solver's 32%.
All three deviations that a comparison without the pool would write down as leaks make money against their pool. And they lose not where they are aggressive, but where they fold.
Why there is no single big number
We could write: "you lose X bb/100, play like the solver and get +X". We don't. Over the whole database such a number looks small, and you have to work spot by spot anyway: each has a specific direction and specific hands, what to do more often and what less.
The price of each spot is an interval over 30 resamples of your database. If the interval crosses zero, the spot does not go into "proven". On a small database there are fewer proven spots, and that is more honest than guessing.
What will be calculated
Seven spots: SRP BTN, SB and CO vs BB; 3bet pots BB and SB vs BTN, BB vs SB and BTN vs CO. For the article's author they cover 83% of hands with a flop in spots that have solver trees. In each spot — the 15 most frequent flops of your database, and next to the price it says what share of the spot's hands they cover. Where the solver has not been run, we make nothing up.
Where it will be
Play like solver will be part of the LeakPrice subscription. It already includes a full database review and a pool review once a month, plus the account: the solver's preflop and flop, progress on leaks.
- NL2–50 — $89 a month
- NL100–200 — $149 a month
- NL400+ — $249 a month
Orders and payment are in the account: leakprice.com/cabinet. The express review is free, in the Telegram bot: t.me/leakprice_test_bot.