Variance and Bankroll Management: The Fundamentals

The most confusing thing about poker is this: good players lose money in the short term, and bad players win money in the short term. This isn't unfair — it is the core mathematical feature of poker, variance. Variance is the intermediate variable between skill and result; without it, poker would be a deterministic "the best player always wins" game and nobody would play. But variance affects more than mindset — it also determines how much money you need to survive long-term in poker. This article covers where variance comes from, how sample size determines whether a result reflects skill, the basics of bankroll management, and the principles for moving up and down.

Where Variance Comes From

Variance = how widely actual results deviate from EV. Poker variance comes from two stacked sources:

Incomplete Information

You don't know your opponent's hand. Even if you make a +EV decision, the opponent may happen to hit an extreme combination (e.g. AA vs KK cooler), making this hand's outcome diverge from EV.

Random Cards

Flop, turn, and river are completely random. With AA all-in preflop against 27o you have 85% raw equity, but you still lose 15% of the time. Those 15% events can "cluster" in single-event sequences, producing the downswings you observe.

Quantifying Variance

The standard measure of variance is standard deviation, usually in bb/100 hands. Typical values for 6-max NLHE cash:

• Loose-aggressive players: 130-160 bb/100
• TAG players: 100-130 bb/100
• Nits (extremely tight): 80-100 bb/100

The more aggressive the style, the higher the variance — more 3bet bluffs and barrels increase both +EV and volatility.

Short Term vs Long Term: Sample Size Decides

Confidence Interval at 1,000 Hands

Suppose your true win rate is 5 bb/100 and your standard deviation is 120 bb/100. The 95% confidence interval at 1,000 hands:

• Mean ± 2σ/√n = 5 ± 2 × 120/√10 ≈ 5 ± 76 bb/100

In other words, after 1,000 hands your result lies somewhere between -71 and +81 bb/100 — the signal of your true 5 bb/100 win rate is completely buried in noise.

Confidence Interval at 100,000 Hands

Same 5 bb/100 win rate, 100k hands:

• 5 ± 2 × 120/√1000 ≈ 5 ± 7.6 bb/100

After 100k hands the interval narrows to -2.6 to +12.6 — you can start to roughly tell whether you're a positive-EV player.

Confidence Interval at 1,000,000 Hands

• 5 ± 2 × 120/√10000 ≈ 5 ± 2.4 bb/100

Only at the 1-million-hand level does the precise win-rate value become trustworthy.

Practical Implications

• A few hundred to a few thousand hands tell you nothing — wins or losses are just variance
• 10k-50k hands give a rough trend, not a precise number
• Real skill evaluation requires at least 100k+ hands

A common beginner mistake: "I'm down 5 buy-ins this month — am I not cut out for poker?" — 5 buy-ins = 500bb, possibly within a few thousand hands, sample size insufficient. Whether you're suited to poker is something to discuss after 100k hands.

Downswings Are Inevitable

Monte Carlo simulation: a 5 bb/100 TAG player over 100,000 hands:

• 80% probability the deepest downswing ≥ 10 buy-ins
• 50% probability the deepest downswing ≥ 15 buy-ins
• 10% probability the deepest downswing ≥ 25 buy-ins

These numbers are inevitable for a long-run +EV player — they have nothing to do with skill. Pros routinely experience 20+ buy-in downswings within a year, and that doesn't mean they suddenly got worse.

Two Wrong Reactions to a Downswing

1. Concluding your skill is broken and rewriting strategy: real skill hasn't changed; the result is just variance. Changing strategy actually breaks an already-tuned game
2. Increasing stakes to chase losses: jumping from 200NL to 500NL to "win it back in one hand" is the fastest way to amplify variance, often turning a small downswing into a catastrophic one

The correct reaction: keep your strategy, keep your stake, and reassess only after sample size is sufficient.

Bankroll Basics

Bankroll management is fundamentally "having enough money to weather the deepest downswing." Conventional standards by format:

"Minimum buy-ins" Is Not a Recommendation

"Minimum" means you accept the bust risk (bust risk = your bankroll going to zero in a rare downswing). For most players this isn't a "suggestion" but a "minimum compliance line."

Bankroll Size Scales with Aggression

The more aggressive your style, the higher the variance and the more buy-ins you need. If you win 5 bb/100 with a standard deviation of 160 bb/100, you need much more bankroll than someone with 5 bb/100 and SD 100.

Mental Bankroll vs Math Bankroll

50 buy-ins might be mathematically sufficient, but if you can't sleep after losing 1 buy-in, your mental bankroll need is far above your math need. Adding a 10-20% buffer is reasonable.

Moving Up and Down

Move-Up Trigger

A general guideline: when your bankroll is 1.5× the buy-in count required for the target stake, you can try moving up. Example: at 50NL you want to move to 100NL — if 50NL needs 30 buy-ins ($1,500), 100NL needs at least 30 × $100 = $3,000, plus a 50% buffer = $4,500 before taking the shot.

Move-Down Trigger

When your bankroll drops to 0.7-0.75× the buy-in requirement of your current stake, move down decisively. Example: 100NL needs 30 buy-ins ($3,000); when you fall to $2,200-$2,300, move back to 50NL.

Mental Factors

• Don't move up mid-downswing (chasing losses)
• Don't move down at the end of an upswing (missing the chance to keep climbing)
• Every move-up comes with a skill asymmetry — new opponents are tougher, and your real win rate at the new stake will be 1-3 bb/100 lower than the previous one

Variance and Mindset

Accept Variance

Pro-level mental training centers on "accept that variance is unavoidable." This isn't a feel-good slogan — it's the prerequisite for letting decisions stay clean from results.

Don't Be Result-Oriented

Before every decision, ask only: "is this +EV?" After the decision, win or lose, don't second-guess the decision itself (unless you're missing information or made an obvious mistake). Reverse-engineering decision quality from results is result-oriented thinking — the biggest mental trap for beginners.

Don't Tilt-Up

Increasing stakes after tilt to "make it back" can turn a normal 5 buy-in downswing into a 20 buy-in bankroll-destroying event. This is the most common watershed between amateur and pro.

Common Mistakes

Mistake 1: "I'm good, I don't need that much bankroll." Variance is independent of skill — a high win rate doesn't mean low variance; they're independent metrics.

Mistake 2: Evaluating yourself by short-term results. Up 10 buy-ins over 3,000 hands doesn't make you a +10 bb/100 player — it just means today's variance was favorable.

Mistake 3: Changing your game mid-downswing. Your strategy was validated by 100k hands of sample; one downswing (maybe 5,000 hands) isn't enough to overturn it.

Mistake 4: Relying on mindset without a math buffer. Even strong mindsets can bust in extreme downswings. The math buffer (enough buy-ins) is the physical foundation for the mental game.

Mistake 5: Bankroll size = progress. A growing bankroll might just be an upswing. Evaluate using win rate (bb/100), not total bankroll.

Summary

Variance and bankroll management are two sides of the same coin:

1. Variance is the essence of poker — it can't be eliminated, only weathered
2. Sample size determines meaning — most results below 100k hands are noise
3. Downswings are inevitable — even +EV players experience 10-20 buy-in declines
4. Bankroll management = variance management — start at 30-50 buy-ins by stake and aggression; add a 20% buffer if you're mentally sensitive
5. Move-up and move-down have clear triggers — don't make those decisions emotionally

Variance awareness, EV thinking, and ICM form the "math trio" of poker — they don't tell you how to play any specific hand, but they define the rules of the game itself. Without these foundations, even great preflop and postflop technique gets diluted by emotion and miscalibrated variance over the long run.