Win Rate

Win-rate

A player won $500 at NL10 over 5,000 hands, then jumped on a forum: "I'm running at +10 BB/100, I'm basically a poker pro." After 50,000 hands his win rate had dropped to +2.3 BB/100. After 200,000 hands it settled at +1.8 BB/100. That original number was variance noise, his real win rate five times lower. This isn't pessimism — it's sample-size math. If you don't understand how win rate works or how many hands you need for a reliable estimate, you'll systematically overrate your skill on upswings and tilt off your stack on downswings.

TL;DR — What It Actually Is

Win rate (sometimes abbreviated WR) is your average profit per unit of play over the long run. Not your result for a day, a week, or a session. Win rate is a converging statistic that only stabilizes over a large hand sample.

How it's measured depends on the game type:

• Cash games: BB/100 (big blinds per 100 hands). NL50 at +3 BB/100 means +$1.50 per 100 hands in dollar terms.
• MTTs (multi-table tournaments): ROI (return on investment), the percentage return on your buy-in. A 30% ROI means for every $1 you put in you get back $1.30 on average over the long run.
• SNGs (sit-and-gos): same ROI format, generally lower because of rake and faster structure.
• $/hour: used for live cash and for evaluating hourly earnings. Useful when comparing poker to other income options.

All these metrics measure one thing: your real skill, separated from luck. The larger the sample, the more accurately win rate reflects your actual edge.

Understanding win rate solves a fundamental problem. Beginners look at a session result and draw conclusions. "Won $200 at NL10 today, I'm a player." "Lost $200 at NL10 today, I'm a degenerate." On a real sample, 200 BB over 1,000 hands is roughly plus or minus 2 sigma of normal variance, not a meaningful skill signal. Without a clear picture of win rate, you're reacting emotionally to noise.

Metrics: BB/100, ROI, $/Hour

BB/100 (cash games). The most widely used metric. Take your total winnings or losses in BBs, divide by hands played, multiply by 100. If you're at NL50 (1 BB = $0.50) and won $250 over 5,000 hands, your win rate = (500 BB / 5,000 hands) × 100 = 10 BB/100. Works well for cash because it normalizes across stakes. 10 BB/100 at NL10 is $1/100 hands; at NL500 it's $50/100.

ROI (tournaments). ROI = (winnings - buy-ins) / buy-ins × 100%. If you played 100 tournaments at $50 each ($5,000 invested) and cashed $6,500, your ROI = (6,500 - 5,000) / 5,000 = 30%. Standard for MTTs and sngs. Doesn't account for time, so a 30% ROI in MTTs can be less profitable than cash at the same hourly rate if a tournament runs 8 hours instead of 2.

$/Hour. Most useful for live cash or when comparing formats on a single scale. Hourly rate = average daily earnings / hours at the table. A top reg at NL2/5 live in Vegas might clear $30–50/hour. NL10 online across 4 tables: $15–25/hour. Live $100/200 for the strongest pros: $200–1,000/hour.

Net BB/100 (after rake). The most honest metric. Includes rake paid. If you're winning 5 BB/100 before rake but paying 3 BB/100 in rake, your net is 2 BB/100. Varies by room: PokerStars takes less rake than ggpoker; live cash takes less than online.

All these metrics connect. At NL50 with a 3 BB/100 win rate and 1,500 hands/hour, you're making $22.50/hour. At NL200 (1 BB = $2) with a 2 BB/100 win rate and 1,000 hands/hour, that's $40/hour. Comparing hourly often tells you more about real profitability than raw BB/100 alone.

Sample Size: How Many Hands You Need for a Reliable Estimate

Win rate is a statistical estimate with a confidence interval. Fewer hands means a wider interval and more noise.

The standard 95% confidence interval formula: CI = real_winrate ± 1.96 × sigma / sqrt(N)

Where sigma is the standard deviation of your win rate (typically 80–120 BB/100 for NL cash) and N is hands played.

Concrete numbers for NL cash with sigma = 100:

• 10,000 hands: CI ±19.6 BB/100. If you're showing "+10 BB/100," your real win rate could be anywhere from -10 to +30. Useless estimate.
• 50,000 hands: CI ±8.8 BB/100. Real win rate within ±9 BB/100 of observed. Starting to mean something.
• 100,000 hands: CI ±6.2 BB/100. A useful estimate, though not precise.
• 250,000 hands: CI ±3.9 BB/100. Accurate enough for most practical purposes.
• 500,000 hands: CI ±2.8 BB/100. Confident estimate.
• 1,000,000+ hands: CI ±2.0 BB/100. Professional-grade reliability.

What this means in practice:

• 10k hands: draw no conclusions about skill. It's noise.
• 50k hands: you have a rough estimate, work with it, but don't over-weight it.
• 100k hands: solid enough for bankroll planning.
• 250k+ hands: real read on your edge; can make move-up or move-down decisions.
• 500k+ hands: confident estimate for long-term planning.

Tournaments are even harder because cashes are rare. Sample size targets for MTT ROI:

• 100 MTTs: interval is massive, estimate is worthless.
• 500 MTTs: starting to tell you something.
• 1,000 MTTs: can begin drawing conclusions about edge.
• 3,000+ MTTs: confident estimate.

Use the variance simulator to model your own sample and see how your current results fit within expected variance.

Win Rates by Stakes in 2026

Realistic win rate ranges for 2024–2026 in major online cash games (PokerStars, GGPoker, WPT Global):

NL2–NL5 (microstakes): top regs 8–15 BB/100, average regs 3–7 BB/100, fish from −10 to −30 BB/100. Massive edge for a disciplined player, thanks to the abundance of recreational players and beginners.

NL10–NL25: top regs 5–9 BB/100, average 2–4 BB/100. The start of the "professional" level. You can make $15–30/hour playing four tables.

NL50–NL100: top regs 3–6 BB/100, average 1–3 BB/100. Regs already know the basics of GTO and the edge gets thinner. $30–60/hour at a solid win rate.

NL200–NL500: top 2–4 BB/100, average 0.5–2 BB/100. Professional-level play. Requires range balancing, mixed strategies, and solver work. $50–150/hour.

NL1K+: top 1–3 BB/100, average regs are around breakeven or marginally positive. High-stakes regs earn not just from win rate but also from rakeback, soft fish events, and promo rotations. $100–1,000/hour.

MTT ROI ranges:

• Micro MTT ($1–5): top players 50–100% ROI, average 20–40%.
• Low MTT ($10–30): top 30–60%, average 10–25%.
• Mid MTT ($50–200): top 15–30%, average 5–15%.
• High MTT ($500–2K): top 8–15%, average 2–8%.
• Super high MTT ($5K+): top 5–10%, the rest often around breakeven.

Win rates compress at higher stakes, but $/hour climbs because the big blind is worth more in absolute terms. That dynamic has held throughout poker history.

Variance: Win Rate Is Not a Steady Paycheck

Win rate is an average. Actual results swing around that average due to variance. Standard deviation in NL cash games is typically 80–120 BB/100, which means your real result over 100 hands can range from −100 to +110 BB at a true win rate of +5 BB/100.

Concrete examples of downswings at different win rates:

Win rate +5 BB/100, sigma 100, 10,000 hands:

• Expected result: +50 BB
• Median downswing over 10,000 hands: −200 BB
• Worst downswing (95th percentile): −400 BB
• Career worst (100K hands): −700 BB

What this means in practice: even a winning player at +5 BB/100 will regularly go through −400 BB downswings. At NL50, that's −$200 over a week or two. Plenty of players can't handle it mentally and quit the game.

Win rate +2 BB/100, sigma 100, 10,000 hands:

• Worst downswing: −600 BB.
• At a low win rate, downswings get brutal.

More detail in Variance and Standard Deviation.

The relationship between win rate and variance sits at the core of bankroll planning. The lower your win rate relative to variance, the bigger the bankroll you need to survive downswings. The formula: for 5% Risk of Ruin, bankroll = 4 × variance² / win rate (in BB). At +5 BB/100 with sigma 100, that's 4 × 10,000 / 5 = 8,000 BB = 80 buy-ins. At +2 BB/100 with the same sigma, you need 200 buy-ins. A thin win rate demands a massive bankroll.

Win Rate Tracking Tools

PokerTracker 4 (PT4). The most popular tracker out there. Imports hand histories from PokerStars, GGPoker, and partypoker. Calculates BB/100, EV BB/100 (theoretical win rate stripped of variance), broken down by position and stake. One-time license: $99.

Holdem Manager 3 (HM3). PT4's main competitor. Same core functionality, different interface. Many players prefer HM3 for its faster hand search. License runs $80–120.

DriveHUD. Cheaper at $30–50, a basic tracker built for micro- and low-stakes grinders. Fewer advanced features, but plenty for player reads and win rate tracking.

Hand2Note. A modern tracker with built-in solver integration. License is $60–100. Popular among serious regulars for its flexible HUD customization.

Built-in site tracker. PokerStars shows basic stats inside the client, but without deep HUD functionality or filters. Fine for a quick glance, useless for actual study.

For MTT players, the same trackers apply, but you also have PokerNet (Sharkscope) and PocketFives for evaluating your ROI and benchmarking against the tournament community. PocketFives rankings show where you stand in the global MTT leaderboard.

Live cash is a different story entirely, trackers don't apply. Use a mobile app like Poker Income, or just keep session notes and calculate BB/100 by hand. Live sample sizes are far smaller than online, so your confidence interval stays wide even after a full year of playing.

Cash vs. MTT: Completely Different Metrics

Cash games and MTT results can't be compared using the same numbers.

Cash games: win rate is measured per 100 hands, with a sigma of 80–120 BB/100. Large samples accumulate quickly. 10k hands in a day is realistic when multi-tabling.

MTT: win rate is measured per tournament as ROI. Variance is dramatically higher because 85–95% of tournaments return nothing (busting before the money). Only 5–15% of entries produce a positive result. You need thousands of tournaments before you can draw confident conclusions.

To put some numbers on it: over 1,000 MTTs at 20% ROI, the actual 95% CI on that ROI runs from roughly -10% to +50%. After 1,000 tournaments, a player with a true +20% ROI can look like a losing player on paper. That level of variance simply doesn't exist in cash games. MTT demands seriously thick skin.

One alternative worth considering is sng. They run faster (30–60 minutes versus 4–8 hours for MTT), variance is lower because you cash in 30–40% of entries instead of 10–15%, and the ROI ceiling is lower too (20–40% versus 20–100% in good MTTs). A solid option for building sample size faster if you're a newcomer working with a limited bankroll.

Live vs Online: Different Rhythms and Win Rates

Live cash: considerably slower at 25–40 hands/hour compared to 1,000+ online. Win rates in BB/100 can look massive (+15 BB/100 for top players), but $/hour is comparable to online NL100–200 because of the low hand frequency. The edge potential is enormous, recreational players in live games are far weaker than the average online player. $1/3 live equals NL100 online in terms of edge, but you're pulling $30–60/hour in actual dollars thanks to a deeper pool of weak players.

Live MTT: ROI runs higher than online MTTs at the same buy-in because of the soft field at high buy-in events. The WSOP Main Event ($10K buy-in) draws roughly 30% recreational players every year. Top pros post 50–100% ROI at these tournaments versus 10–15% at $10K online events. The downside is brutal variance from a tiny sample size. 10–30 high roller MTTs per year at most.

Online cash: fast sample accumulation, lower $/hour compared to live high stakes, but it scales. Playing 4–8 tables simultaneously multiplies your effective hourly rate.

Online MTT: same variance problems as live, but your sample builds much faster. If you want to grow a bankroll to $10K in a year, online MTTs are a realistic path after 2,000+ tournaments. In live poker, the same journey takes years.

How to Actually Improve Your Win Rate

Work with a solver. PioSolver, GTO+, and MonkerSolver model optimal play. Compare your real decisions against solver recommendations. Every deviation frequently costs real money. Beginners should start with basic spots: 3-bet pots and c-bet sizing.

Review hands using tracker filters. PT4/HM3 let you filter hands by type (open vs. 3-bet vs. c-bet) and check your win rate in each scenario. The weak spots in your game usually surface clearly once you break things down by situation.

Balance your ranges. At NL50 and above, unbalanced ranges are one of the main reasons for a weak win rate. If you open 22% on the button but only 3-bet QQ+/AK, regulars will exploit you through reverse-blocker lines. Study your ranges.

Bankroll management. Poor bankroll management often disguises a weak win rate as a "bad run." At +2 BB/100, you need 200 buy-ins for a 5% Risk of Ruin. If you have 50, you'll be sweating your bankroll even with a positive win rate.

Mental game. Tilt costs a lot of BB/100. A simple calculation: average tilt loss of $50 per session, 50 tilt sessions per year = $2,500. At NL50, that's 50 buy-ins burned purely on emotions. Reading: Jared Tendler's "The Mental Game of Poker."

Drop down in stakes for the sake of BB/100. Running -2 BB/100 at NL50? Go back to NL25 where you're running +4 BB/100. Your dollar earnings will be lower, but it's psychologically far more manageable and your skill progression will be more consistent.

Common Win-Rate Estimation Mistakes

Small sample size. A player runs +20 BB/100 over 5,000 hands and decides he's a crusher. That's noise. Actual win rates for regulars at micro and low stakes almost always land somewhere between -10 and +10 BB/100.

Gross ROI without accounting for rake. On ggpoker, rake can eat up 4–5 BB/100 at NL50, wiping out a big chunk of your gross win rate. Always calculate net BB/100 after rake, not gross.

Mixing stakes. If you're playing NL10, NL25, and NL50 simultaneously, your "combined" win rate tells you nothing useful. Track each stake separately.

Ignoring format. 6-max and 9-max have different optimal win-rate ranges. Zoom (fast-fold) versus regular cash is another distinct category. Your tracker needs to filter by these breakdowns or the numbers are meaningless.

Selection bias. A player only tracks the sites where he has accounts and ignores live games or other platforms. This systematically inflates his self-assessment.

Not being ready for downswings. A player running a true +5 BB/100 will still go through stretches of -10 BB/100 lasting 20,000+ hands. Completely normal. If you can't handle downswings mentally, you'll quit before variance evens out.

Where Win Rate Falls Short

Win rate is a statistical estimate, not a prediction. Knowing your true win rate says nothing about tomorrow's result. Variance can throw anything at you, including a 3-sigma month in either direction, and strong players sometimes post losing stretches while weaker players occasionally book massive wins.

Win rate doesn't transfer to new stakes. A player running +5 BB/100 at NL10 who moves up to NL50 often sees -2 BB/100. That doesn't mean his game got worse. The edge at the new level is just thinner. Move up carefully and expect your win rate to drop.

Win rate ignores hourly opportunity cost. Making $10/hour at poker when you could earn $30/hour doing something else is effectively -$20/hour, even if you're technically profitable. Factor in opportunity cost, especially if you have marketable skills outside the game.

Your long-term true win rate can drift over time. Players improve, so a win rate measured two years ago may no longer reflect current reality. Re-measure after major skill upgrades: solver work, a strategy overhaul, switching to 6-max. Win rate can also erode gradually as the player pool gets tougher.