Poker Win Rates Explained: What Good bb/100 Looks Like
5bb/100 is strong online at NL50 while live 1/2 regs post double that, and you need roughly 100k hands before the number means anything.
6 min read · Published
A good online win rate is roughly 5bb/100 at NL50, and anything above 2bb/100 is respectable once you reach NL200 and higher. Live 1/2 regulars routinely post double the equivalent online number, because the games are softer and the pace is slower. The awkward part is the sample: below about 100,000 hands, your win rate tells you almost nothing that variance cannot explain on its own.
What bb/100 actually measures
Big blinds per 100 hands is the only win rate worth comparing across games. It normalises for stake and for volume, so a NL10 grinder and a NL200 regular can talk about the same number.
Win rate = (profit in big blinds ÷ hands played) × 100
Win $180 over 6,000 hands at NL50, where a big blind is $0.50, and you have made 360bb. Divide by 6,000, multiply by 100, and your win rate is 6bb/100.
Two older measures still float around and both are worse. BB/100 with a capital B counts big bets, a limit-poker unit that halves every number and confuses everyone. Dollars per hour mixes your edge with your table count and cannot be compared between players at all.
One detail catches people out when they compare figures with other players. Most tracking software reports your win rate net of rake but before any rakeback, cashback or bonus money, so two players quoting the same 4bb/100 can be taking home visibly different amounts. Check which number someone is quoting before you decide you are behind them.
What counts as a good win rate online
Win rates fall as you move up, and they fall sharply. The same player who beats NL25 for 7bb/100 will often be a 2bb/100 winner at NL200 against opponents who no longer donate.
| Stake (6-max online) | Solid winner | Strong regular |
|---|---|---|
| NL2 to NL10 | 5bb/100 | 10bb/100+ |
| NL25 to NL50 | 3bb/100 | 6-8bb/100 |
| NL100 | 2bb/100 | 4-5bb/100 |
| NL200 to NL500 | 1bb/100 | 3bb/100 |
| NL1000+ | Breakeven plus rakeback | 2bb/100 |
Three things follow from that table.
Big win rates only exist where big mistakes exist. Double-digit bb/100 at NL5 is not evidence of mastery, it is evidence of opponents calling three streets with third pair. The realistic ceiling for micro stakes win rates is high precisely because the leaks are enormous.
Rakeback stops being a bonus. At NL500 a 1bb/100 edge on the tables plus 1bb/100 back from the site is a doubling of your income. Serious players choose sites on rake structure for exactly this reason.
Moving up costs win rate but earns money. Dropping from 6bb/100 at NL50 to 2.5bb/100 at NL200 sounds like a collapse. In dollars it is $30 per thousand hands becoming $50 per thousand hands.
Live win rates are higher and worth much less
Live cash games produce win rates that look absurd next to online numbers. A competent 1/2 regular beating the game for 10bb/100 is entirely normal, and 15bb/100 at a loose table is not a fantasy.
Then you divide by volume. Live you see roughly 25 to 30 hands per hour; online, several hundred across tables. That 10bb/100 at 1/2 works out to around $5 to $6 an hour before tips and travel.
The comparison between formats is mostly a comparison of hourly rates rather than bb/100, and the trade-offs between online and live hourly rates rarely favour whichever one you assumed.
Why you need 100,000 hands before believing anything
Your observed win rate is a measurement with an error bar, and at small samples the error bar swallows the measurement. Standard deviation in online 6-max runs around 90 to 100bb/100, which produces confidence intervals like these:
| Hands played | 95% range around a 5bb/100 result |
|---|---|
| 10,000 | -15 to +25bb/100 |
| 50,000 | -3.8 to +13.8bb/100 |
| 100,000 | -1.2 to +11.2bb/100 |
| 500,000 | +2.2 to +7.8bb/100 |
| 1,000,000 | +3 to +7bb/100 |
Read the first row again. After 10,000 hands, a player showing 5bb/100 could genuinely be a 15bb/100 loser having a good month. That is not a technicality, it happens constantly.
At 100,000 hands the interval finally excludes zero, which is why that figure gets quoted as the threshold for taking your own graph seriously. Even then you know you are a winner, not by how much.
Running your own numbers through win rate confidence intervals is more sobering than any article can be, and it is the fastest cure for both overconfidence after a heater and panic during a downswing. The mechanics of how variance masks your edge explain why 20 buy-in swings are ordinary for a genuine winner.
See your real expected win rate vs actual results
Sharkling's analytics separate the run-bad from the actual leaks, hand by hand, against a solver baseline.
What actually moves the number
Win rate is not one skill, and the components are not equally weighted.
- Preflop ranges. The cheapest fixes live here. Opening 30% from UTG instead of 16% costs several big blinds per 100 on its own, and no amount of postflop brilliance repairs it.
- Big blind defence. Folding too much for the price you are offered is the single most common leak at low stakes. Every fold you make below the break-even threshold is money handed back.
- Rake. At NL5, rake can exceed 8bb/100 taken from the pool. Your edge is what remains after it, not before it.
- Table selection. Sitting with two regulars rather than two recreational players can halve your win rate without a single strategic error.
- Tilt and session length. The 4bb/100 you earn across a solid session vanishes in twenty minutes of chasing losses at stakes you do not normally play.
How to track it without fooling yourself
Tracking software makes it easy to slice data until something looks profitable. Guard against that.
Never split by hand. Your win rate with AKs over 800 hands is noise wearing a costume. Anything below a few thousand occurrences of a spot tells you nothing.
Do split by position. Position samples grow fast and the pattern is stable: you should lose in the blinds, win moderately in the middle seats, and win most from the BTN. A losing BTN win rate over a decent sample is a real, findable problem.
Watch your rate stats before your money stats. VPIP and PFR stats stabilise within a few thousand hands, while win rate needs six figures. If your VPIP is 34% in a 6-max game, you already know something is wrong without waiting for the graph to confirm it.
Separate all-in adjusted from actual. The gap between them is short-term luck. Over 100,000 hands the two lines should sit close together, and a large persistent gap usually means your sample is smaller than you think.
What to do with the number
Your win rate has exactly two honest uses.
The first is bankroll. A 5bb/100 winner with 100bb/100 standard deviation needs roughly 30 buy-ins to keep the risk of ruin low; a 2bb/100 winner needs considerably more. Lower edge means longer swings, so moving up demands more buy-ins rather than fewer.
The second is deciding when to move up. Twenty thousand winning hands at NL25 is not proof of anything, but combined with a bankroll and a plan to drop back down it is enough. Waiting for statistical certainty means waiting for 100,000 hands at a stake you have already outgrown.
What your win rate cannot do is tell you how to play better. It is an outcome, measured slowly and reported late. The decisions that produce it — the ranges, the sizings, the folds you should not have made — get fixed by studying spots, not by refreshing a graph.
The practical habit is to check the number monthly and act on it quarterly. Anything more frequent turns a slow measurement into an emotional one, and you end up moving down after a bad week or moving up after a good one. Set a review interval, hold to it, and spend the time you save on hands instead.
See your real expected win rate vs actual results
Sharkling's analytics separate the run-bad from the actual leaks, hand by hand, against a solver baseline.
Get The Sharkling Edge
Twelve issues, one poker idea each — position, pot odds, board texture, blockers. Free, and you can leave whenever you like.
No spam, one-click unsubscribe in every email.
Frequently asked questions
What is a good poker win rate in bb/100?
Online 6-max, around 5bb/100 is strong at NL50 and anything above 2bb/100 is respectable at NL200 and higher. At the micros a good regular can run 8bb/100 or more. Live 1/2 players often post double the equivalent online figure because the games are much softer.
How many hands do you need to know your true win rate?
Roughly 100,000 hands before the number is worth quoting, and even then the 95% confidence interval is about plus or minus 6bb/100. At 10,000 hands the interval is wider than almost any real edge, so a winning graph over that sample proves very little.
Is 3bb/100 a good win rate?
At NL100 and above, yes. 3bb/100 at NL200 is roughly $60 per thousand hands before rakeback, which is a solid professional edge at that level. At NL5 the same number would suggest you are leaving a lot on the table.
Why is my live win rate higher than my online win rate?
Live games have weaker opposition, more limping, and far less multi-tabling, so the average edge per hand is larger. The trade-off is volume: you see around 25 to 30 hands per hour live versus several hundred per hour online, so a big bb/100 turns into a modest hourly rate.
Does rake affect your win rate much?
Enormously at small stakes. Rake can cost several big blinds per 100 hands at NL5 and NL10, which is often larger than the edge itself. Rakeback and loyalty schemes are a real part of a low-stakes win rate rather than a bonus.
Try it yourself
Simulate thousands of hands at your win rate to see confidence intervals, downswing sizes, risk of ruin, and how many hands it really takes to know if you're a winner.