Variance and drawdown

Variance is the difference between the result you expect and the result you get. Drawdown is the worst-case version of that difference. Both are real and unavoidable. A bettor who refuses to internalize them will quit a winning strategy on its way to paying off, and stick with a losing one until the bankroll is gone.

Why variance feels worse than it is

Human intuition is bad at random sequences. We see meaningful patterns in noise. We are surprised by stretches that the math says should happen often. The result is that bettors look at a 4-12 cold stretch and conclude 'my strategy is broken' when in fact a 4-12 stretch is normal variance for many real strategies.

The stat that helps is the binomial distribution. For a true 55% win-rate on -110 lines (a strong but realistic edge), the probability of various worst-case stretches looks like this.

55% TRUE WIN-RATE BETTOR, -110 LINES
  prob(0 wins in next 5)    = 1.8%
  prob(1 or fewer in 10)    = 2.3%
  prob(4 or fewer in 16)    ≈ 11%
  prob(25 or fewer in 60)   ≈ 12%
  prob(45 or fewer in 100)  ≈ 7%

  At any moment a winning bettor has
  meaningful probability of looking
  like a losing bettor. Variance is the
  default; results are noisy.

Drawdown

Drawdown is the peak-to-trough decline in your bankroll over a period. A bettor whose bankroll runs from $10,000 up to $12,000 and back down to $9,000 has a peak drawdown of $3,000 from the high, or 25% of the peak. Drawdown is a normal feature of every betting strategy.

Sizing decides how big the drawdowns get. Bigger units cause bigger drawdowns at the same edge. The math compounds: a strategy that runs +3% EV at quarter Kelly might see 12% drawdowns from peak. The same strategy at full Kelly might see 50% drawdowns. A bettor who cannot psychologically tolerate a 50% drawdown should not be running full Kelly.

What you can actually expect

What the table tells you is that even at strong edge with conservative sizing, drawdowns of 10 to 20% are normal across a season. A bettor who panics at 5% drawdown will be panicking constantly. Internalize the numbers before they happen, not during.

Why bettors quit at the wrong time

Two cognitive failures combine to make exit timing terrible.

First, recency bias. The most recent results dominate the bettor's perception of strategy quality. A bettor with a 1,000-bet positive history will quit a strategy after 30 bad bets if those bad bets are recent enough. The strategy did not change. The data did not change. The 30-bet sample is just noise.

Second, sunk-cost reasoning in the wrong direction. After a drawdown, bettors often increase unit size to 'make it back faster.' This compounds the risk of ruin. Mathematically, doubling unit size on a -1% EV strategy doubles the rate at which the bankroll bleeds. There is no version where this works out.

How long until variance fades

Variance decreases proportional to the square root of the sample size. A 100-bet sample has a standard deviation of result roughly 10 times that of a 10,000-bet sample (relative to expected value). Concretely:

• 100 bets is too few to evaluate strategy quality reliably.
• 500 bets is enough to see a clear strategy with strong edge, but real strategies often look ambiguous.
• 1,000+ bets is where genuine signal starts to dominate noise on most realistic edges.
• 10,000+ bets eliminates almost all variance noise on realistic edges.

For most retail bettors, that means 'do not draw conclusions about your strategy from less than a season's worth of bets.' For sharp bettors with high volume, the timeline compresses. For low-volume bettors taking 100 bets a year, the strategy needs a decade of data to evaluate cleanly.

This is also why closing line value is so useful. CLV stabilizes much faster than ROI and gives you signal on edge quality long before the equity curve has settled.

Plan the drawdown before it happens

1. Pick a sizing strategy you can stomach. Quarter Kelly is the default. If a 12% drawdown is your real psychological limit, size to that.
2. Write down your strategy and your sizing in advance. The act of writing makes deviation visible later.
3. Track CLV alongside ROI. CLV gives you an early signal that the strategy is still healthy even when the equity curve says otherwise.
4. Resist increasing size during a drawdown. The math does not support it. The math supports staying the course.

What to read next

Kelly criterion is the formal sizing framework. Hit rate vs ROI explains why looking at hit rate alone tells you almost nothing about strategy quality. Closing line value is the metric that helps you stay disciplined during drawdowns.