In short: every postflop decision is an edge your opponent hands you — the app measures how much of that edge you capture and gives you a score from 0 to 10.
Plan: concept · the full tool is Pro 🔒
The idea
Most tools give you an average ("you're winning X winrate"). The Exploits tab does it differently: it starts from a truth about the micros — the field plays in predictable ways, and every situation opens an edge to exploit (betting more value, bluffing where they fold, calling where they overbluff, folding where they don't bluff enough). For every edge, the app sorts things into two piles:
- Found: you captured the exploit. ✅
- Left on the table: the edge was there and you didn't take it. 💸
⚠️ Important: "left on the table" does not mean "played against GTO." It means the field gave you an advantage and you didn't cash in on it. This is population exploitation, not solver output.
The score (Exploit Score)
At the top of the tab you'll find your score from 0 to 10, like a report card — the higher it is, the more of those edges you're converting. It's weighted by edge size (leaving a fat value bet on the table counts more than a marginal defense) and comes with a label: Excellent, Good, Fair, Poor or Critical. If you still have few hands in the filter, the app shows "small sample" instead of a fragile score.
Right below, Your next exploit tells you what to do now: the spot that costs you the most in bb (with a pattern — 10 or more mistakes), how many times you let it slip, the right play with the field's number ("Bet as a bluff ~73% pot: the field folds 44% and a bluff needs 31%") and the See these hands button. Then comes Where you leave the most: a single list, in bb, of the spots that cost the most — tap one to open the hands. The spots follow the same order (from the one that costs the most in bb to the least).
The "subjects" (each scenario has its own score)
The postflop tree has 10 scenarios (single-raised pot, 3-bet, 4-bet — as the aggressor or the caller, in position or out of position). Each one becomes a subject in the report card, with its own score. Inside each subject, every exploit comes with a plain-language description of what it is — for example:
- Defense: a line full of bluffs → the right move is to call your bluff-catchers.
- Discipline: a line that barely bluffs → the right move is to fold your bluff-catchers.
- Letting go: a line without enough bluffs → the right move is to give up the marginal hand.
- Value / Bluff / Raise / Check-raise: betting/raising when the edge is open.
And a found × left on the table bar shows how well you're capturing that scenario.
What your exploits are worth (bb left on the table)
In the score card, the app turns your mistakes into bb: each edge you let slip is worth what the field pays in that spot — how often it folds to a bet (bluffs) or, on the river, how often the call wins (defense and discipline). You see the bb left on the table, the bb captured with the right plays, how much you leave per 100 hands and last month's note ("in October you left 0.8 bb per 100 hands — 38% less than in September"); the Your progress chart, month by month, sits at the end of the page, and the spots that cost the most are the Where you leave the most list. Each spot also shows the bb left in it, and in any hand's replay the Engine comment says what each decision was worth ("left ~1.2 bb on the table at the field's price") with the field's number for that spot. On Max, the Mentor uses the same numbers: in the hand comment it explains the why with the field's number, and in the chat and the report it tells you how many bb each spot cost and whether you're leaving less month by month. It's an estimate at the field's price; value, raises and check-raises are not in this count (they have no measured price).
Proven in the field's hands (the green seal)
The edges aren't opinions: each one was proven in the field's hands. When you open a spot, the "Proven in the field's hands" box shows the number behind the edge, for example: "Flop: the field folds 43% to a bet — a bluff needs 30%". The same seal appears under the right play in the hands you left on the table and in the replay. In bluff spots, the seal also tells you the size that earns the most against the field there (on the flop, by board type), when the measurement shows one size beating the others: "Size that earns the most against the field on this flop texture: ~70% pot (it folds 55%)". In any hand's replay, when you bluffed with a size that clearly earns less than that, the Engine comment tells you: "Size: you bet ~33% pot; here ~73% earns more against the field (it folds 56% × 38%) — ~0.4 bb more per bluff in this pot". On a river value bet with a strong hand, the tip is about value size: the field calls almost as often against a big bet, so the overbet usually earns more ("Value size: you bet ~31% pot; with a strong hand on the river, ~166% earns more against the field — it still calls 28%"). It's advice: it doesn't change your score. Only spot names and aggregated numbers — never another player's data.
Field profile and Poker Semi-Pro's own rules
The Field profile box (at the end of the tab, closed — tap Open) sums up how the micro-stakes field plays: how often it folds to a c-bet (and the bluff size that earns the most), what it does after a check and after calling ("attack the check, respect the call"), how often a pair wins calling its river bet, how often it calls each bet size (it barely reacts to size: on the river, with a strong hand the overbet earns ~3× the 1/3 bet; with a pair, betting doesn't pay — check and call), and how it reacts to a river raise and in the 4-bet pot. Each number comes with the advice it supports. It's the same picture for every client — it doesn't depend on your hands.
Where the measurement contradicts the study material, the engine follows the numbers: in some river spots the study says to fold your bluff-catchers, but a pair wins far more than the call needs against the field's bet. In those spots Poker Semi-Pro's own rule applies — call with a pair against a bet of up to 85% pot — and folding the pair counts as a Defense miss (the right play tells you it's our own rule). Against a bigger bet the price eats the edge (the pair wins 38% and the call needs 38%), and the study's rule for that spot applies. In the 4-bet pot, when you called the 4-bet and the villain bets the flop, the study says to fold everything but value, but folding the pair costs ~11 bb more per hand than continuing: our own rule applies there too — don't fold your pair.
For bluffs, defense, discipline and letting go, decisions in spots where the field's hands haven't proven the edge yet (or showed the opposite) stay under observation: they count neither as a hit nor as a miss until the proof shows up. For c-bet bluffs the proof also works by flop type: on a board where the field doesn't fold enough (e.g. two-tone in the 3-bet pot), the bluff stays under observation. Value, raises and check-raises keep counting.
How to use it
- In Diagnosis, open the Exploits tab.
- See your score at the top and, under Your next exploit, click See these hands — the spot that costs you the most in bb.
- Read Your most repeated mistakes (what you did × the right play) and, if you want, filter by flop type — the old Study Tree now lives here (spots and flop types).
- Click a hand to replay it — the street badge (Flop/Turn/River) shows exactly which decision that is (the same hand can have two).
- Take the fix to Training and watch the score climb.
Why this is different from a solver
A GTO solver solves the entire tree and is heavy, technical and in English. Here the goal is to be actionable: point out, in plain player language, how much edge you're leaving on the table against the field and which scenario to study first. It doesn't replace a solver — it helps you fix what matters most, faster.