AI LEAK REPORT

Find your biggest leaks in 30 seconds

Paste your last 50-100 hand histories. The AI aggregates the stats — money lost per hand class, position breakdowns, c-bet patterns — then writes a structured report on the top 3 leaks bleeding your bankroll.

Hand history

PokerStars / GG format (text). Multiple hands separated by blank lines. Aim for at least 30-50 hands for a useful report.

0 chars · —3 free reports / month

📊

What it analyzes

Hands won/lost per starting hand class, position-by-position counts, c-bet frequency, limp detection, all-in spots.

🔒

Your data

Your raw hand history is parsed locally in our serverless function. Only aggregated stats (numbers) get sent to the AI — never the raw hands themselves.

⚡

Output

Top 3 leaks with estimated dollar cost · what you're doing right · one focus item for next month.

Coming with PRO / Elite

Unlimited reports (vs 3/month free)
File upload (drag-and-drop your full hand history archive)
Trend analysis across multiple sessions
Per-stake leak breakdown when playing multiple stakes
Side-by-side comparisons month-over-month

See PRO + Elite features →

Guide

What the leak report finds — and what it can't

A leak is a repeated mistake, and repeated mistakes hide in volume. You won't notice that you over-call 3-bets from the blinds by reviewing one hand — you notice it when 60 hands sit side by side. That's the job of this tool: paste your last 50-100 hands and get a structured report on the top 3 patterns costing you money, written by Claude from an aggregate view of your sample.

It's aimed at micro-stakes players who don't run a paid tracker. If you play NL2-NL25 on GG or Stars and export hand histories, you have everything you need. The report also tells you what you're doing right — knowing which habits to keep matters as much as knowing which to fix.

How to use it, step by step

1.Export a hand history file from your poker site (PokerStars or GG text format).
2.Open the file in any text editor, copy everything, and paste it into the box above. Aim for 50-100 hands.
3.Generate the report. Analysis takes about 30 seconds.
4.Fix leak #1 first. Ignore #2 and #3 until the next report shows #1 is gone — one habit at a time actually sticks.

The methodology, honestly

Your pasted hands are analyzed by Claude, which aggregates money lost per hand class, position breakdowns, and c-bet/aggression patterns, then writes the findings up as a report. It is pattern recognition over a small sample, not a solver audit — and a 50-100 hand sample is small. Frequency leaks (limping, calling too wide, never 3-betting) are reliably visible at this size; anything win-rate-shaped is not, because variance dominates short samples. Run a report every few weeks, fix the top item, and compare against the micro-stakes strategy guide to see where your game should be heading. Free users get 3 reports per month; PRO ($9/mo) is unlimited.

Frequently asked questions

How many hands do I need?›

At least 30-50 for a useful report; 50-100 is the sweet spot. Below that, the AI can still spot frequency leaks (limping, over-calling) but position and sizing patterns get noisy.

How many free reports do I get?›

3 leak reports per month on the free tier. PRO ($9/mo) removes the limit. One report a week is genuinely enough for most players — leaks change slowly.

What hand history formats work?›

PokerStars and GG-format text files. Paste the raw text with hands separated by blank lines — no conversion needed. If your site exports a different format, the parser may miss fields; PokerStars format is the safest bet.

How is this different from the Hand History Analyzer?›

The Hand History Analyzer runs entirely in your browser and computes fixed stats (P/L per hand, leak heuristics). The Leak Report sends your sample to Claude, which aggregates the patterns and writes a coaching-style narrative: your top 3 leaks, what they cost, what you're doing right, and one thing to focus on. Use the analyzer for numbers, the report for direction.

Is 50 hands really enough to find leaks?›

For frequency leaks, yes — if you limp 20% of hands, that shows up in 50 hands. For win-rate conclusions, no — bb/100 needs tens of thousands of hands to converge (see our variance calculator). Treat the report as directional coaching, not a statistical verdict.