How to review a session with hand history and player profiles

1. What was recorded

GTO MAX AI keeps, under your login, every hand it saw: the cards, the board, the action, the pot, and for each of your decisions the GTO line, the exploit line and the explanation as they were shown at the time. It also keeps every opponent's profile as it evolved: the tendencies with their samples, the archetype and its confidence, your notes. The review uses both: what the advice was, and what was known about the player when it was given.

2. Step 1: the deviations

Filter the session to decisions where your action differed from the top action of the advice. Not the big pots; the deviations. Most sessions have ten to thirty. These are the only hands worth your time, because the rest were played as the theory and the profile suggested, and whether they won or lost was the cards.

3. Step 2: sort them into three piles

• Wrong. You deviated, the advice was better, and you can see why now. The common ones: a river fold with the right price, a call against a player whose profile said he never bluffs, a bluff into a station. These go into step 4.
• Right. You deviated and you were right for a reason the tool did not have: a live read, a note you had not written yet, a timing tell. These are the most valuable hands in the review, because they are the information the model lacked. Write the note now, in the Players tab, so it has it next time.
• Lucky. You deviated, you won, and you cannot name a reason. These are the dangerous ones. Winning hands teach bad habits faster than losing ones. Mark them wrong.

4. Step 3: read the profile at the time

For each deviation, open the opponent's profile and look at the number the advice rested on: "folds to river bet 71% over 18 hands". Ask whether 18 hands justified the move the advice made, and whether the hands since have moved the number. Sometimes the review shows the tool was right for the wrong reason: a small sample that happened to point the right way. That is worth knowing too, because the next time the sample is small it may point the other way.

5. Step 4: three spots into study

From the wrong pile, pick the three that recur. Not the exotic ones; the pattern ones. Over-folding rivers against three-quarter-pot bets. Flatting 3-bets with hands that are pure folds. Checking back thin value against stations. Take each to a study tool and look at the spot in general, then come back to the hands and see the general rule in the specific case. Three spots a week, every week, is how a player's game changes in a quarter.

6. Step 5: the players

Scan the Players tab for anyone whose archetype changed this session, or whose confidence crossed from guess to conviction. A player who was "unknown" and is now "calling station, 78% confidence" is a player you will meet again, and the exploit line against them will be different next time. Add a note where the numbers do not tell the story: a player who only became a station after losing a big pot is two players, and the model will take a while to work that out alone.

7. What not to do

Do not review the coolers. Do not review the hands where you followed the advice and lost; those are variance and the review cannot help. Do not change your strategy because of one session's results against one player. The procedure above is dull and it works, and the hand history makes it twenty minutes instead of two hours.

How the profiles that drive this are built: player profiling. Why the deviations matter: GTO vs exploitative play.

More from the blog

• Is poker RTA detectable? How sites catch it in 2026
• GTO vs exploitative poker: when to leave the equilibrium, and by how much
• AI in poker: what RTA, bots and solvers actually do, and what they do not
• Playing GTO with real-time assistance: where the edge is, and where it is not

Longer reads: the poker RTA guide and all guides.