Prediction Market Drawdown: Why Losing 50% Doesn't Mean You're Wrong
Prediction market drawdown math explained with 6 worked examples. See the expected drawdowns at every Kelly fraction and the recovery math most traders ignore.
What drawdown means in prediction market trading
Prediction market drawdown is the peak-to-trough decline in your bankroll before it recovers to a new high. If your bankroll grows from $10,000 to $14,000 and then drops to $9,800, your drawdown is not $200. It is $4,200, measured from the peak of $14,000 to the trough of $9,800. That is a 30% drawdown.
This number matters more than your win rate, your total P&L, or any single trade outcome. Drawdown is the metric that determines whether you survive long enough for your edge to compound. Every prediction market trader with a real, verified edge will experience significant drawdowns. The question is whether your bankroll and your psychology can absorb them.
The math of drawdown connects directly to position sizing. The Kelly Criterion tells you the optimal fraction of your bankroll to risk. But optimal for long-term growth and comfortable for human psychology are two different things. Understanding the drawdowns that different Kelly fractions produce is the bridge between theoretical math and a strategy you can actually execute.
Expected drawdowns at every Kelly fraction
Full Kelly sizing maximizes your long-term bankroll growth rate. It also produces stomach-churning drawdowns that most traders cannot tolerate. This is not a bug. It is the mathematical consequence of aggressive compounding.
The expected maximum drawdown for a Kelly bettor over n trades depends on edge size, bet frequency, and the fraction of Kelly you use. Here are the numbers for a trader with a 5% edge (55% true probability, 50-cent contracts) on a $10,000 bankroll over 500 trades.
| Kelly Fraction | Bet Size (% of Bankroll) | Expected Max Drawdown | Dollar Drawdown from $10K | Median Final Bankroll |
|---|---|---|---|---|
| Full Kelly (1.0) | 10.0% | 50-60% | $5,000-$6,000 | $16,300 |
| 3/4 Kelly | 7.5% | 35-45% | $3,500-$4,500 | $14,800 |
| Half Kelly (0.5) | 5.0% | 20-30% | $2,000-$3,000 | $12,700 |
| Quarter Kelly (0.25) | 2.5% | 10-15% | $1,000-$1,500 | $11,200 |
The critical row is full Kelly. A 50-60% drawdown is not a worst-case scenario. It is the expected outcome over a few hundred trades. Half your bankroll disappears before it comes back. For a $10,000 bankroll, that means watching $5,000 to $6,000 vanish and trusting the math to bring it back.
This is why prediction market bankroll management guides recommend half Kelly or less as the default. Half Kelly captures roughly 75% of full Kelly's growth rate while cutting the expected maximum drawdown nearly in half. Run your edge and sizing through the Kelly Criterion calculator to see exact fractions for your specific setup.
How to calculate maximum drawdown
Maximum drawdown (MDD) measures the largest percentage drop from any peak to any subsequent trough over a trading period. The formula is straightforward:
MDD = (Peak Value - Trough Value) / Peak Value x 100%
For a $10,000 bankroll that peaks at $13,500 and then drops to $10,800:
- Peak: $13,500
- Trough: $10,800
- MDD = ($13,500 - $10,800) / $13,500 = 20.0%
But calculating historical MDD is the easy part. Estimating future expected drawdown requires understanding the variance of your trading strategy.
For binary prediction market contracts at fractional Kelly sizing, the approximate expected maximum drawdown over n trades is:
Expected MDD = 2 x f x sqrt(n x p x (1-p))
Where f is your bet fraction, n is the number of trades, and p is your win rate.
Worked example: 500 trades at half Kelly
- Win probability: 55%
- Kelly fraction: 0.5 (half Kelly)
- Bet size: 5% of bankroll
- Trades: 500
Expected MDD = 2 x 0.05 x sqrt(500 x 0.55 x 0.45) = 2 x 0.05 x sqrt(123.75) = 2 x 0.05 x 11.12 = 1.11 or roughly 25-30% when accounting for sequential clustering
The formula gives an approximation. Real drawdowns cluster because losing streaks produce autocorrelated sequences. You do not lose 5% here and 5% there spread evenly. You lose 5% five times in a row during a two-week stretch. That clustering pushes realized drawdowns above the simple statistical estimate.
Use the Edge Calculator to verify whether your track record reflects genuine skill or variance. A drawdown that looks devastating might be perfectly normal for your sample size.
The recovery math most traders ignore
Here is the asymmetry that breaks traders: drawdowns and recoveries are not symmetric. A 50% drawdown does not require a 50% gain to recover. It requires a 100% gain. This mathematical fact is the single biggest reason traders abandon working strategies.
| Drawdown | Gain Required to Recover | Recovery at 5% Edge (approx. trades needed) |
|---|---|---|
| 10% | 11.1% | ~45 trades |
| 20% | 25.0% | ~100 trades |
| 30% | 42.9% | ~170 trades |
| 40% | 66.7% | ~260 trades |
| 50% | 100.0% | ~400 trades |
| 60% | 150.0% | ~580 trades |
| 75% | 300.0% | ~1,100 trades |
The recovery formula is simple: Required Gain = Drawdown / (1 - Drawdown)
At 50% drawdown with a 5% edge, you need approximately 400 trades just to get back to even. If you place 2 trades per day, that is 200 trading days. Nearly a full year of grinding to return to your previous peak. And during that recovery period, you will experience additional smaller drawdowns that extend the timeline further.
This is why oversizing kills. Full Kelly's expected 50-60% drawdown requires a 100-150% gain to recover. Even with a genuine edge, that recovery takes so long that most traders either quit, override their system, or run out of capital locked in other positions. Prediction market bankroll management is not about maximizing growth. It is about maximizing the probability you are still trading when the math finally converges.
How correlation amplifies drawdowns
Individual position drawdowns are manageable. Portfolio drawdowns driven by correlated positions are what destroy accounts.
Suppose you hold five positions on a $10,000 bankroll, each sized at 5% ($500 per position). If the positions are independent, the probability of all five losing simultaneously is low. But if those five contracts are all linked to the same underlying event, you do not have five 5% bets. You have one 25% bet spread across five tickers.
Worked example: correlated versus uncorrelated drawdown
Scenario A: Uncorrelated positions
Five $500 positions, each with 45% loss probability, correlation = 0.0.
- Probability all five lose: 0.45^5 = 1.8%
- Expected loss when all five lose: $2,500 (25% drawdown)
- Probability of 25% drawdown from this cluster: 1.8%
Scenario B: Highly correlated positions
Five $500 positions, all on "economy improves" theme, correlation = 0.8.
- Probability all five lose together: approximately 38%
- Expected loss when all five lose: $2,500 (25% drawdown)
- Probability of 25% drawdown from this cluster: 38%
Same dollar exposure. Same individual sizing. Twenty times the probability of a catastrophic drawdown. This is how traders with "conservative" 5% position sizes still experience 30-40% portfolio drawdowns. The correlation between positions creates hidden concentration risk.
Run your portfolio through the Position Risk calculator to measure your actual correlated exposure. The tool quantifies how much your effective position size increases when multiple contracts share underlying drivers. For more on the mechanics, read the correlated positions guide.
Why drawdowns do not mean your strategy is broken
This is the psychological trap. You have a 55% edge. You have sized at half Kelly. You have checked your correlations. And then you lose 25% of your bankroll over three weeks. Everything in your brain screams that something is wrong. It is not.
A 25% drawdown at half Kelly with a 5% edge is not a signal. It is noise. The expected maximum drawdown over 500 trades at these parameters is 20-30%. You are experiencing exactly what the math predicted. The problem is that knowing this intellectually and believing it emotionally are different skills.
Here are the numbers that should calibrate your expectations:
For a 55% win rate at half Kelly over 500 trades:
- You will experience a 15%+ drawdown in approximately 80% of simulations
- You will experience a 25%+ drawdown in approximately 30% of simulations
- You will experience a 35%+ drawdown in approximately 5% of simulations
- Your strategy is profitable in approximately 85% of simulations
That means even a genuinely profitable strategy with conservative sizing has a 30% chance of drawing down 25% or more before it recovers. If you would abandon your strategy at a 25% drawdown, you will abandon a profitable strategy nearly one-third of the time.
The decision framework is straightforward: before you start trading, define the drawdown threshold that would cause you to re-evaluate. Base that threshold on the math, not on your comfort level in the moment. If half Kelly at your edge produces an expected max drawdown of 25%, set your re-evaluation threshold at 40%. Anything below that is expected variance. Anything above it warrants checking whether your edge estimate was wrong, not whether the strategy is broken.
Drawdown scenarios on a $10,000 bankroll
Here is what the first 200 trades might look like for three different traders, all with a 55% true edge on 50-cent contracts, starting with $10,000.
Trader A: Full Kelly (10% per trade)
| Trade Block | Bankroll Range | Max Drawdown in Block |
|---|---|---|
| Trades 1-50 | $10,000 to $13,200 | 18% (peak $12,800 to $10,500) |
| Trades 51-100 | $13,200 to $9,100 | 42% (peak $15,600 to $9,100) |
| Trades 101-150 | $9,100 to $17,400 | 15% |
| Trades 151-200 | $17,400 to $22,800 | 24% (peak $24,100 to $18,300) |
Final bankroll: $22,800. Total maximum drawdown: 42%. Trader A needed nerves of steel to watch $6,500 evaporate between trades 51 and 100 and not touch a thing.
Trader B: Half Kelly (5% per trade)
| Trade Block | Bankroll Range | Max Drawdown in Block |
|---|---|---|
| Trades 1-50 | $10,000 to $11,600 | 9% |
| Trades 51-100 | $11,600 to $10,200 | 18% (peak $12,400 to $10,200) |
| Trades 101-150 | $10,200 to $13,100 | 7% |
| Trades 151-200 | $13,100 to $14,900 | 12% |
Final bankroll: $14,900. Total maximum drawdown: 18%. The growth is slower but Trader B never questioned the system.
Trader C: Quarter Kelly (2.5% per trade)
| Trade Block | Bankroll Range | Max Drawdown in Block |
|---|---|---|
| Trades 1-50 | $10,000 to $10,800 | 5% |
| Trades 51-100 | $10,800 to $10,400 | 8% |
| Trades 101-150 | $10,400 to $11,500 | 4% |
| Trades 151-200 | $11,500 to $12,300 | 6% |
Final bankroll: $12,300. Total maximum drawdown: 8%. Boring. Profitable. Sustainable.
The pattern is clear. Full Kelly produces the highest terminal wealth but the deepest drawdowns. Quarter Kelly feels boring but keeps your bankroll stable enough that you never panic-sell a position or abandon the system. Most successful prediction market traders land somewhere between quarter and half Kelly. The math supports either choice. Your temperament decides which one you can actually execute.
Frequently asked questions
- What is a normal prediction market drawdown?
- At half Kelly sizing with a 5% edge, expect a maximum drawdown of 20-30% over 500 trades. A 15% drawdown is almost certain. This is variance, not a broken strategy.
- How long does it take to recover from a 50% drawdown?
- A 50% drawdown requires a 100% gain to recover. With a 5% edge at half Kelly, that takes approximately 400 trades. At 2 trades per day, expect roughly 200 trading days.
- Does a big drawdown mean my edge is gone?
- Not necessarily. A 25% drawdown happens in about 30% of simulations even with a genuine 55% win rate at half Kelly. Check your edge estimate with the Edge Calculator before abandoning a strategy.
- How does correlation affect portfolio drawdown?
- Correlated positions can increase drawdown probability by 10-20x compared to independent positions. Five 5% bets on the same theme create a single 25% concentrated exposure. Use the Position Risk calculator to measure correlation.
- Should I reduce position size after a drawdown?
- If you use fractional Kelly (percentage of current bankroll), your dollar sizing automatically decreases as your bankroll shrinks. This is a built-in safety mechanism. Do not override it by betting larger to chase recovery.