Frequently asked questions

Short answers to the questions that come up most often. Anything about strategy lives in the guides; this page is about the product itself.

Is it free? Do I have to sign up?

It is free and there is no account — no email, no password, no sign-in of any kind. Open the page and play. Your settings and your cumulative statistics are stored in your own browser.

Do I have to install anything? Does it work offline?

Nothing to install, and no app store. The whole trainer is one self-contained page: once it has loaded, the engine runs inside your browser, which means your decisions are graded without a round trip to a server. Your statistics survive a reload because they live in local storage, not in a session.

What does GTO mean?

Game Theory Optimal: a strategy that cannot be exploited, because it makes your opponent's bluffs and their value bets equally unprofitable. In practice that means playing a mix of actions rather than one "correct" action per hand. There is a full explanation, including what GTO does not promise, in What GTO actually means.

Why does the same spot sometimes tell me to raise and sometimes to call?

Because a correct strategy is a mix, not a rule. When two actions have nearly the same EV, the equilibrium plays each of them a share of the time — that unpredictability is what stops an opponent from exploiting you. The trainer shows those shares as frequencies, and it will accept either action when the mix is close. What it will not accept is always taking the action that feels safer: a frequency of 30% raise does not mean "never raise", and never raising is exactly the leak the number is pointing at.

How do I read the range grid?

The grid is the 13x13 matrix of starting hands, one cell per combo class. The number or shading in a cell is how much of that hand the equilibrium plays the highlighted action with — so a cell at 100% is a pure play and a cell at 40% is a mixed one. Read it as a shape rather than cell by cell: you are looking for where the boundary sits, because the boundary is the thing you can carry to a table you have never seen before.

How accurate is the advice?

It depends on the street, and we would rather give you the measured numbers than a slogan. Against our own CFR+ solver, the median EV loss of the trainer's advice is 0.37% of the pot on the river (exploitability 0.000%), 0.54% on the turn (0.073%), and about 0.9% on the flop — where the number is explicitly marked as reference only, because that solve has not converged (3.34%). Heads-up river pots do not use the heuristic at all: they are solved live with CFR+ and the panel labels the remaining exploitability. Every decision you make is graded against those same numbers.

Which languages does the interface support?

Fourteen: English, 简体中文, 繁體中文, 日本語, 한국어, Deutsch, Español, Français, Italiano, Português (Brasil), Русский, ไทย, Türkçe and Tiếng Việt. Each language has its own address, and the root address adapts to your browser's language the first time you arrive.

Is this gambling? Can I win or lose money?

No. There is no real money anywhere in the product: you play against simulated opponents or against friends at a play-money table, and no payment method is involved. It is study software — the point is the feedback on your decisions, not the result of any single hand.

What data do you collect?

The hand you played — table settings, hole cards, every action, community cards and the result — plus a randomly generated device id that is not derived from any device property, and your interface language. Nothing else: no name, no email, no account, no location, no device fingerprint, no advertising identifier. The full list, including what deliberately stays on your machine, is in the privacy policy.

Can I turn the collection off?

Yes. Open the training page, press the "?" button, find the hand-data section and choose "Opt out". The trainer works exactly the same either way — collection is a default, not a feature you depend on.

Why are there ads? Where do they appear?

Ads pay for the servers that host the site and the collection service. They are manual units placed after the main content of each application page — never over the table, never over a button, and never inside the decision flow. The landing page has no ads at all, and we do not use auto-placement.

Does it work on a phone?

Yes. The layout is built for small screens, and the range grid, the action bar and the review timeline all work with touch. A phone is fine for training and for the equity questions you meet away from your desk; a larger screen is simply more comfortable for entering a hand you played live.

How do the tables with friends work?

One of you creates a table and gets a six-character room code; everyone else joins with that code. No sign-up, nothing to install. When you leave a table your seat is released immediately, and if your connection drops your seat is held for you so you can come back.

Can I review a hand I played somewhere else?

Yes — that is what the review page is for. Enter the table size, the positions, the stacks, your hole cards and everyone's actions, and each decision is replayed against equilibrium and graded. You can also paste a hand record in the format the trainer exports.

Are the opponents realistic?

They are twelve profiles built from the styles that actually show up in low and mid stakes games — the player who calls too much, the one who folds to any pressure, the one who only raises with the nuts, and so on. Each of them plays a different mix of tightness, aggression and bluff frequency, so the same hand can be a clear call against one and a clear fold against another. That variation is the point: a trainer where everyone plays well teaches you a game nobody plays.

Does it cover tournaments?

Not currently. The model is cash-game: fixed stacks, no blinds increase, and no ICM — the trainer does not adjust for payout structures or for the fact that chips you win are worth less than chips you lose near a bubble. Playing it will still sharpen the decisions that tournaments are made of, but treat the preflop ranges as cash-game ranges.

How long before I see a difference?

The honest answer is that the feedback arrives immediately and the improvement does not. What changes fast is your ability to name why a decision was wrong — usually within a couple of sessions, because the EV panel puts a number on it. What changes slowly is the habit: recognising the spot before you have to think about it takes weeks of the same loop, which is why the guides push one decision type at a time rather than "study everything".

If I get good at this, will I be a winning player?

Not by itself, and we will not pretend otherwise. Making fewer mistakes is necessary and it is not sufficient: table selection, the stakes you play, how much you play, and whether you actually stick to the decisions you know are right all sit outside anything a trainer can grade. What the trainer can do is show you the cost of each mistake in a number small enough to act on — 0.4bb here, 1.2bb there — which is the part most players never get to see.

Something is broken, or an answer looks wrong. How do I tell you?

Write to us through the contact page. A hand record helps a lot — it is the difference between "the grading felt off" and a reproducible case we can check against the solver.

Where should I start?

Open the trainer and play twenty hands without touching the settings. Then read How to use this trainer and pick one thing to fix — usually the decision type that got graded badly most often. That loop, repeated, is the whole method.