Anki for Poker: Do Flashcards Actually Work for Ranges?

If you're trying to memorize your ranges, you've certainly come across Anki. It's the best-known flashcard software in the world, free on desktop and Android, and it runs on spaced repetition — exactly the right principle.

So the question deserves a real answer: does Anki work for poker ranges?

Short version: the engine is good, but the card format doesn't fit poker. Here's why — and what to use instead.

What Anki does very well

Credit where it's due. Anki's spaced repetition algorithm is solid, battle-tested across millions of users, and the principle it applies is the right one: review just before you forget, stretch the intervals with every success.

For vocabulary, formulas or dates, it's unbeatable. The engine isn't the problem. The problem is that poker doesn't fit inside a flashcard.

Limit #1 — The granularity problem

A range is 169 hands on a 13×13 grid. When you try to turn that into cards, you only have two options, and both are bad.

Option A: one card per range (front "BTN open 100bb", back an image of the grid). You create ten of them in an hour. Except that at review time, you flip the card, look at the image and think "yes, I knew that". You did not rebuild it — you recognised it. That's precisely the illusion of knowing, the most expensive bias in all range learning.

Option B: one card per hand (front "KJo on the BTN facing a CO open", back "fold"). Now you're genuinely doing active recall — but do the maths: 169 hands × 10 ranges = 1,690 cards to build by hand. You'll spend more time creating your deck than learning it. And you'll learn isolated hands out of context, never seeing the shape of the range — which is exactly what lets you generalise to a new situation.

There's no good middle option in Anki, because the unit it manipulates is the card, not the grid.

Limit #2 — Self-grading is subjective

In Anki, you decide whether you knew it: Again / Hard / Good / Easy. For vocabulary that works — you objectively know whether you retrieved the word.

For a range it's unworkable. You got 154 hands right out of 169: is that "Good" or "Hard"? And if all 15 mistakes are borderline hands — that is, exactly the ones that matter — the right answer is "Again", but your brain will tell you "Good, I had the essentials".

The result: you over-grade yourself, the algorithm stretches the intervals too far, and you lose the range without noticing.

Limit #3 — Anki doesn't know poker

Anki has no idea what a position, an effective stack or a frequency is. So it can't:

• display an interactive 13×13 grid you paint your answer onto;
• correct you cell by cell and tell you where you got it wrong;
• handle frequencies (mixed hands, e.g. 50% raise / 50% call) — and modern ranges are full of them;
• objectively measure your accuracy to drive the review schedule.

You can hack all of this together with add-ons and custom HTML in your cards. Budget several evenings of configuration — for a result that will stay approximate.

Limit #4 — Your ranges change

You switch format, you tweak an opening range, you move from 6-max to MTT: in Anki you have to rebuild the affected cards by hand. That's the kind of friction that, in practice, gets a deck abandoned after three weeks.

What Forge.poker does instead

Forge.poker starts from the same scientific principle as Anki — spaced repetition — but with the right unit: the range, not the card.

You build nothing. You import or draw your range once on the grid, and the training exists immediately. No deck to construct, no 1,690 cards to type.

The test is genuine active recall. You get an empty grid and the situation; you rebuild the range from memory, cell by cell. Impossible to merely "believe" you knew it: either you produce the answer or you don't.

Correction is objective and located. Forge compares your grid to the reference and shows you exactly which cells you missed. You immediately see that your errors cluster on the borderline hands — the information Anki can never give you.

Grading isn't subjective. Your real accuracy percentage drives the review schedule, not your impression. Solid ranges space themselves out; shaky ones come back fast.

Frequencies are handled natively. Mixed hands are part of the model, not a bolted-on workaround.

The cycle is progressive. Study → copy → rebuild from memory: you're not thrown in at the deep end, and you don't get stuck in passive reading either.

The verdict

Anki remains an excellent tool — for what it was designed for. If you're learning Japanese, don't change a thing.

But for poker ranges, you'd spend evenings building an approximate deck, with self-grading that lies to you and no way to see the shape of a range. A dedicated trainer does the same memorization work with zero preparation, and with a correction you can't fudge.

The principle is identical. It's the tool that has to fit the material — and a poker range belongs on a 13×13 grid, not on a two-sided rectangle.

Going further

• The full step-by-step method: learning your ranges for free
• Why your memory fails, and the three real causes: why you keep forgetting your ranges
• How long it really takes: how long to learn your ranges