LeakLab reads your PLO database and shows where you lose more than strong regulars do, in the same spots, over and over.
Not one more page of tracker statistics.
Playing under different screen names, or in more than one room?
LeakLab merges the whole history into one profile.
The sample is big enough for the deviation to count as significant.
You keep firing out of position too often.
The problem is not the number 70.8% itself.
The problem is that in the same situations winning regulars play considerably more carefully.
LeakLab does not stop at the deviation: it names the spot, the street, and the direction your strategy should move in.
Tell us which rooms and which stakes you play. If you have several accounts or screen names, LeakLab merges them into one profile. The more hands, the sharper the analysis.
We do not look at the average of the whole field. Your numbers are compared against players who play those stakes and those formats regularly. That makes it far easier to see where your strategy really differs from the strong pool.
Not every bad number is a problem. Sometimes a few pots in a row simply did not get there. So LeakLab weighs the sample size and the statistical error. When there is not enough data we write exactly that: it is too early to call it. No invented leaks for the sake of a good-looking report.
Dozens of small deviations are never equally important. LeakLab works out which mistakes are potentially costing you the most and puts them in order. You work on the most expensive leak first. Then you move to the next one.
In H2N or PT4 you can see:
And that is where the questions start.
LeakLab was built to answer exactly those questions.
Shows statistics.
Shows statistics → compares them against regulars → checks how reliable the conclusion is → finds the deviation → explains the problem.
Hand over your database from every room and get a review back: where your game parts company with the strong pool, which street the money starts leaving on (Preflop → Flop → Turn → River), and what to fix first. Every conclusion carries its sample size — when the data is thin LeakLab says so instead of dressing variance up as a leak.
A spot matrix, 13×13 ranges and frequencies for every four-card combination. Set where you sit, what you are facing, how deep you are and which format, then read the strategy for that exact spot. No separate subscription: the solver opens in the same window as your review.
Spot after spot: open, defend versus open, versus 3-bet, squeezes, 4-bets, limp pots. The session deals a hand, you answer, and the trainer checks that answer against the same base the solver reads. Misses collect in a list of their own, and you can drill them again on the spot.
CoinPoker and PokerOK take no HUD and hand out no history. Upload screenshots of the table instead — LeakLab reads the screen names, the bets street by street and the cards shown down, and builds a card on every opponent: hands, positions, verdicts, and search by name.
The whole preflop for PLO: the spot matrix, 13×13 ranges and action frequencies down to the individual four-card combination. No separate solver subscription — it opens in the same window your analysis lives in.
You pick:
and you get the strategy for that exact spot.
Without hunting for one hand among thousands of combinations by hand.
You open the range.
You pick the first two cards.
Then the exact four-card combination.
And you see the action frequencies:
First you see every main game situation.
For example:
You pick the spot you want. The matrix of the first two cards opens.
You pick a cell, and LeakLab shows the two cards that are left.
That way you get from a whole range down to one concrete four-card PLO hand, quickly.
A matrix of spots: positions × situations, every spot at once, with the frequencies on each. Click a spot → a 13×13 grid over the first two cards, with the suits built into the grid itself as triangles: bottom left single suited, top right double suited, pairs down the diagonal. Click a cell → a second 13×13 grid, this time over the two remaining cards, where the triangles mean bad suited / good suited. The screen replaces itself, and you walk back out through the breadcrumbs.
Rooms PS · GG · MTT · HU, stakes Low · Med · Med+ · High, stacks 30–200bb. The Easy RFI / Smart Range switch changes the data, not the drawing: Easy is the narrower, simplified range. Per-hand detail is live on PS · Low · 100bb; on every other preset the spot says plainly that it is still being prepared.
The solver shows no EV in bb: it is not in the data, and we are not going to draw a pretty number.
The solver shows how a spot is played. The trainer finds out whether you play it that way with the clock running.
How it works:
A mistake here costs nothing. The same mistake at the table costs the pot.
The score is a self-diagnosis tool: we cannot see whether the solver is open in the next tab. What it does count is the hands the server actually dealt and the time it actually allowed.
Your move, and the clock is running.
You answered: call.
How the spot is actually played:
A miss. The hand goes to "My mistakes" and comes back until you answer it correctly twice.
CoinPoker and PokerOK give you no statistics at all: no HUD to attach, no hand histories to export. The player across the table is a nickname and nothing else.
LeakLab reads what you do have — the screenshots of the table. Drop a batch in and get notes on everyone who sat in the hand with you.
The notes stay inside LeakLab: it is your database, it is not sent to the room and not fed into a client.
Hold'em, PLO4 and PLO5 — the game type is read from the table itself, and the card count follows it.
In poker it is very easy to mistake variance for a pattern.
Imagine: over your last 500 hands one line has run you a big loss.
It may be a leak.
It may be a handful of big pots that did not get there.
So LeakLab does not look at the result alone. We count how many times the situation came up, and how wide the statistical error around it can be.
If there is not enough data for a confident conclusion, the service says so plainly.
There is evidence of a problem → we show the leak.
There is not enough evidence yet → we do not invent a problem.
A database review is paid separately — one buy-in of your limit.
Prices are set in US dollars; the charge is in roubles at the Central Bank of Russia’s rate on the payment day.
You work on that situation for a few weeks.
Then you leave a new request, this time for the fresh database.
LeakLab compares the result:
And it shows whether your strategy is moving in the right direction.
That turns working on your game from:
into:
LeakLab earns its keep once the database is big and reading HUD statistics is no longer enough.
It is for you if:
The more the better. Some metrics settle on a fairly small sample; others need thousands of repetitions of one specific situation. So LeakLab does not set a single magic number for the whole database. Whether there is enough data for a conclusion is judged separately for every metric.
Nothing bad. LeakLab will say that there is not enough data yet, and will not sell a random deviation to you as a real leak. As more hands come in, the conclusion may become statistically significant.
Yes. Databases from different rooms and the screen names of one player can be merged into a single profile. That is especially useful when your sample in each individual room is fairly small.
Against the pool of regulars, not against every player at the table. The job is to show where your strategy differs from the players who consistently play that format.
No. They are different tools. A tracker is very good at storing hands, running a HUD and letting you read statistics yourself. LeakLab is for a different job: finding the most likely systematic leaks in a big database and showing you where to start.
Not only. The solver is one part of the platform. The idea behind LeakLab is to connect the analysis of your real game to the work that follows from it. First you find the leak. Then you open the matching GTO spot.
If that data is not in the underlying model, then no. We would rather show no number at all than produce a pretty one that nothing stands behind.
Yes. On Pro you can have new hands analysed again and watch how the metrics found earlier have moved.