poker iq
How the score works.
Poker IQ is a number from 0 to 1,000. It is not an opinion and there is no panel behind it. Five components are calculated from your own hands, each with a fixed maximum, and the sum is multiplied by five. This page is the whole formula, including the parts of your game it cannot see.
the five components
The maximums are not equal, and that is the first thing worth knowing. They add up to 200 raw points, which becomes the 0 to 1,000 you see.
Preflop
max 40How often you enter a pot, how often you raise when you do, and how often you 3-bet.
Starts at 20 points. Each of the three statistics can add up to 10 more, and each loses points for its distance from a reference value. At a six handed table those references are 22 percent VPIP, a PFR to VPIP ratio of 0.75, and 8 percent 3-bet.
Postflop
max 40How often you follow your preflop raise with a bet on the flop.
Starts at 20 points, adds a flat 5, and adds up to 15 more for how close your flop continuation bet percentage sits to 65. You lose half a point per percentage point of distance.
River
max 30How often you win the hands that reach a showdown.
Starts at 15 points and adds up to 15 for how close your won at showdown percentage sits to 50. One point per percentage point of distance.
Leaks
max 30How many open leaks the detectors currently see, and how heavy they are.
30 points minus the weight of your open leaks: 10 for each high, 5 for each medium, 2 for each low. Three open high leaks take this component to zero.
Result
max 60Your winrate over the session, in big blinds per 100 hands.
30 points at break even, 60 points at plus 10 bb/100, 0 points at minus 10 bb/100, a straight line in between, clamped at both ends. This is the only component that looks at money.
Total
max 1,00040 plus 40 plus 30 plus 30 plus 60 is 200. Multiplied by five that is 1,000. A new account starts at 500, which is the middle of the scale and not a judgement.
The reference values shift with table size. At a nine handed table the VPIP anchor moves from 22 percent down to 16, and the 3-bet anchor from 8 percent down toward 6, because eight opponents ahead of you is a different game than five. Heads up moves the other way, hard. What a 3-bet and a continuation bet are is in the glossary.
two brakes
Both exist for the same reason: a score that jumps around on thin evidence is worse than no score.
Damping below 30 hands
A session shorter than 30 hands has its components pulled back toward their own midpoints. The factor is the hand count divided by 30, with a floor of one fifth, so even a five hand session keeps a trace of signal without being allowed to speak for your game.
A cap of 40 points per session
Whatever the components say, one session moves the total by at most 40 points either way. The cost of that protection is honest to state: after a genuine change in your play, the score needs several sessions before it agrees with you.
how much to trust it
The score carries a label based on your total hand count, and the label is part of the answer. The thresholds are fixed, and tournaments need more hands than cash because the decisions are spread across changing stack depths.
| Label | Cash, from | Tournament, from |
|---|---|---|
| Rough read | under 500 hands | under 1,000 hands |
| Focusing | 500 hands | 1,000 hands |
| Locked in | 2,000 hands | 5,000 hands |
| Stone cold | 10,000 hands | 20,000 hands |
For a sense of the scale in practice: on the test account used during development, a recomputation of the cash score returned 655 out of 1,000. That is one account on one day, and it is on this page as a scale reference and nothing else.
what it cannot do
A score that only advertises its strengths is a sales pitch. These are the limits, and they are structural rather than temporary.
It compares you to reference values, not to a measurement
The anchors the score leans on, 22 percent VPIP at six handed and the values that scale from there to heads up and full ring, are conventional numbers. The code that holds them says so in its own comment: they are common reference values, not something measured on this database. A score of 700 means your frequencies sit close to those anchors. It does not mean 700 players out of 1,000 play worse than you.
Thirty percent of the score is a winrate, and a winrate is noisy
The result component is 60 of the 200 raw points. Over a thousand hands the margin of error on a winrate is usually wider than the winrate itself, so on small volume this component is largely measuring luck. That is also why one session can move the total by no more than 40 points.
Frequency is not correctness
The score sees that you continuation bet 65 percent of flops. It does not see whether those were the right 65 percent. A player who c-bets every board scores the same as a player who picks the boards that fit their range, as long as the percentage lands in the same place.
It cannot see your opponents
GGPoker encrypts opponent names in the hand history, and the same opponent gets a different code in your next session. So nothing about table quality, opponent tendencies, or who was actually in the pot enters the score. Two players with identical statistics against very different opposition get the same number.
Short sessions barely count, on purpose
Below 30 hands in a session every component gets pulled back toward its own midpoint, by a factor that falls to a floor of one fifth. A five hand session cannot say much, so it is not allowed to.
The same hands feed two of the components
Your frequencies decide the preflop and postflop components, and those same frequencies are what the leak detectors look at. A single loose preflop pattern can therefore cost you points twice, once directly and once through an open leak.
The leaks component is the one that tells you what to do next, because it points at a named pattern instead of a number. How those patterns are found, including the two rules that stop a detector from firing on noise, is on the leak finder page. An actual report, with the measured numbers from one account, is here.
next
See your own score.
The formula above runs on hands you have already played. Create an account, upload a session of your own hand history, and the five components are calculated on your numbers instead of an example. Under 500 hands it will tell you it is a rough read, because it is.
questions
What is a Poker IQ score?
A number from 0 to 1,000 built from five components of your own play: preflop frequencies, flop continuation betting, showdown results, open leaks, and winrate. The components are worth 40, 40, 30, 30 and 60 raw points, which sum to 200 and are multiplied by five to reach the 0 to 1,000 scale. The midpoint, and the value a new account starts from, is 500.
How many hands do I need before the score means anything?
The app labels anything under 500 cash hands a rough read, and it is right to. From 500 hands the frequency based components start to settle; the winrate component needs far more than that. For tournaments the same labels sit at 1,000, 5,000 and 20,000 hands, because tournament volume is spread across changing stack depths.
Can one bad session tank my score?
No. A single session can move the total by at most 40 points in either direction, up or down, and the score is then clamped to the 0 to 1,000 range. That cap exists so a losing night does not rewrite a measurement built on thousands of hands. The trade off is that the score lags: after a real change in your play it takes several sessions to catch up.
Are cash games and tournaments scored together?
No. Hands that carry a tournament phase are excluded from the cash score entirely, and tournaments have their own score with their own reliability thresholds. Mixing them would compare a 100 big blind cash decision with a 12 big blind push or fold decision as if they were the same problem.
Does the score go up if I win more?
Partly. Winrate is 60 of the 200 raw points, so it moves the number, but the other 140 points come from how you played rather than how it turned out. That split is deliberate: results over a few thousand hands are mostly variance, and a score that followed them would tell you to change your game every time you ran badly.