#Definition
Drawdown measures the decline from a portfolio's peak value to its lowest point before recovering to a new peak. In prediction markets, drawdown quantifies the worst loss a trader experiences during a specific period, expressed as a percentage or absolute value.
A trader whose account grows from 15,000, then falls to 3,000 decline from the $15,000 peak). Drawdown is the essential metric for understanding downside risk and the psychological stress of a trading strategy.
#Why It Matters in Prediction Markets
Drawdown captures what raw returns cannot: the pain of losing. Two traders might both end the year up 50%, but one experienced a smooth path while the other suffered a 40% drawdown mid-year. Their experiences—and the likelihood of abandoning their strategy—differ dramatically.
Risk assessment: Expected value tells you the average outcome; drawdown tells you how bad it can get. A strategy with high EV but extreme drawdowns may not be survivable in practice.
Position sizing: Maximum acceptable drawdown should inform position sizing. If you can't tolerate a 30% drawdown psychologically or financially, you must size positions to make that level unlikely.
Strategy evaluation: Comparing strategies requires risk-adjusted metrics. A strategy returning 40% with 10% max drawdown is superior to one returning 50% with 50% max drawdown for most traders.
Capital preservation: In prediction markets, unlike diversified portfolios, individual positions can go to zero. Understanding drawdown dynamics is crucial for survival.
Psychological preparation: Knowing your strategy's historical drawdowns prepares you mentally. Traders who encounter unexpected drawdowns often panic and abandon sound strategies at the worst time.
#How It Works
#Calculating Drawdown
At any point in time:
Current Drawdown = (Peak Value - Current Value) / Peak Value × 100%
Example trajectory:
| Date | Portfolio Value | Peak Value | Drawdown |
|---|---|---|---|
| Jan 1 | $10,000 | $10,000 | 0% |
| Feb 1 | $12,000 | $12,000 | 0% |
| Mar 1 | $11,000 | $12,000 | 8.3% |
| Apr 1 | $9,500 | $12,000 | 20.8% |
| May 1 | $10,500 | $12,000 | 12.5% |
| Jun 1 | $13,000 | $13,000 | 0% |
The maximum drawdown for this period is 20.8% (the deepest trough before recovery).
#Key Drawdown Metrics
Maximum Drawdown (Max DD): The largest peak-to-trough decline over the entire period. This is the single most important drawdown metric.
Average Drawdown: The mean of all drawdown periods, indicating typical decline severity.
Drawdown Duration: How long until recovery to a new peak. Long durations test patience even with modest percentage declines.
Recovery Factor: Total return divided by maximum drawdown. Higher is better—it measures how much return you earn per unit of worst-case pain.
Recovery Factor = Total Return / Maximum Drawdown
A strategy returning 60% with 20% max drawdown has a recovery factor of 3.0.
#Visualizing the Pain
#Python: Analyzing Drawdowns
This script calculates the Maximum Drawdown from a list of equity values.
import numpy as np
def calculate_max_drawdown(equity_curve):
"""
Calculates Max Drawdown % and Duration.
equity_curve: List of portfolio values over time.
"""
peak = equity_curve[0]
max_dd = 0
for value in equity_curve:
if value > peak:
peak = value
dd = (peak - value) / peak
if dd > max_dd:
max_dd = dd
return max_dd * 100
# Example Portfolio History
portfolio = [10000, 12000, 11000, 9500, 10500, 13000]
mdd = calculate_max_drawdown(portfolio)
print(f"Maximum Drawdown: {mdd:.2f}%")
#Drawdown in Prediction Markets
Prediction market drawdowns often behave differently than traditional portfolios:
- Binary outcomes: Individual positions can lose 100%, creating sharp drawdowns
- Event clustering: Multiple correlated positions can fail simultaneously
- Timing concentration: Drawdowns often cluster around major events (elections, announcements)
- Recovery dependency: Unlike stocks, losing prediction market positions don't recover—new positions must generate returns
Example with binary outcomes:
A trader holds three 10,000 portfolio):
- Position A: 60% probability, loses → -$1,000
- Position B: 70% probability, wins → +700, received $1,000)
- Position C: 55% probability, loses → -$1,000
Net result: -$1,571 (15.7% drawdown from a single event cluster)
Event Clustering Risk: Unlike stocks where earnings are scattered, prediction markets often resolve simultaneously (e.g., Election Day). This forces "step-function" equity curves. You might be flat for months, then instantly jump +50% or drop -30% in a single night. Drawdown management here means managing aggregate exposure to single dates.
#Examples
Concentrated election exposure: A trader allocates 40% of their portfolio to related election markets. When unexpected results emerge, all positions lose simultaneously. The portfolio drops 35% in hours—a drawdown that takes months of profitable trading to recover.
Overconfident position sizing: Using Kelly Criterion aggressively, a trader bets 25% of their bankroll on a 65% probability event. The event resolves unfavorably. This single loss creates an immediate 25% drawdown, requiring a 33% gain just to break even.
Extended drawdown duration: A trader correctly predicts several long-dated markets, but most positions won't resolve for 6+ months. Meanwhile, short-term losing trades create a 15% drawdown that persists for months until the profitable positions settle. The extended duration—though the percentage is modest—strains confidence.
Correlated market failure: A trader holds positions across 10 apparently different markets, but all are correlated to the same underlying economic outcome. When economic conditions shift unexpectedly, 8 of 10 positions move against the trader simultaneously, creating a 25% drawdown despite apparent diversification.
#Risks and Common Mistakes
Ignoring drawdown in strategy selection: Choosing strategies based purely on returns without considering drawdowns leads to selecting unsustainable approaches. High returns often come with high drawdowns.
Underestimating correlation: Seemingly independent prediction markets often correlate during stress events. Diversification that looks good on paper fails when everything moves together.
Psychological capitulation: The most common drawdown mistake is abandoning a sound strategy during a normal drawdown. Strategies with 20% expected max drawdown will eventually hit that level—panicking and selling locks in losses.
Increasing size during drawdowns: Attempting to recover losses by increasing position sizes ("doubling down") transforms manageable drawdowns into account-destroying spirals.
Ignoring duration: A 10% drawdown lasting one week is very different from one lasting six months. Extended drawdowns erode confidence and opportunity cost even when percentage losses are modest.
Survivorship bias: Historical drawdown data often excludes strategies that suffered catastrophic drawdowns and were abandoned. The strategies you can study are the ones that survived.
#Practical Tips for Traders
-
Define your maximum acceptable drawdown before trading. If 20% is your limit, size positions so this level is unlikely (not impossible) to breach.
-
Track drawdown in real-time, not just returns. Seeing "15,000 peak" feels different than "up $2,000 from start."
-
Use fractional Kelly sizing (half-Kelly or quarter-Kelly) to reduce drawdowns while maintaining most of the expected return.
-
Stress test for correlation: Assume your positions are more correlated than they appear. What happens if 70% of positions lose simultaneously?
-
Plan for drawdown duration: Ensure you don't need the capital for other purposes during potential extended drawdown periods.
-
Document your strategy in advance so you can distinguish a normal drawdown from a broken strategy. "20% drawdown is expected" is very different from "something is wrong."
-
Review drawdowns analytically: After a drawdown, assess whether it resulted from bad luck, bad sizing, or flawed analysis. Only the latter requires strategy changes.
-
Set hard stop-losses on total portfolio drawdown if you cannot tolerate extreme outcomes. Accept reduced returns for reduced risk.
#Related Terms
#FAQ
#What is a good maximum drawdown for prediction market trading?
There's no universal answer—it depends on risk tolerance and strategy type. Conservative approaches target max drawdowns under 15%. Aggressive strategies might accept 30-40%. Few traders can psychologically survive drawdowns exceeding 50%, regardless of expected returns. Choose a level you can genuinely tolerate, not the theoretical maximum.
#How does drawdown relate to Kelly Criterion?
Full Kelly sizing maximizes long-term growth but produces large drawdowns—often 30-50% or more. Most practitioners use fractional Kelly (half or quarter) to reduce drawdowns significantly while sacrificing only modest expected returns. The relationship is direct: larger bet sizes create larger drawdowns.
#Why is maximum drawdown more important than average drawdown?
Maximum drawdown captures the worst-case scenario you'll actually experience. Average drawdown understates risk because extreme events are rare but devastating. You don't go broke from average drawdowns—you go broke from the maximum. Plan for the worst, not the typical.
#How long should I expect drawdown recovery to take?
Recovery time depends on drawdown depth and strategy returns. A rough formula: if your strategy returns R% annually and you suffer a D% drawdown, recovery takes approximately D/R years. A 20% drawdown with 40% annual returns recovers in ~6 months. A 20% drawdown with 10% returns takes ~2 years. Compounding complicates this—deep drawdowns require larger percentage gains to recover.
#Should I stop trading during a drawdown?
Usually no—if your strategy has positive expected value, stopping during a normal drawdown just delays recovery. However, you should: (1) verify the drawdown is within expected ranges for your strategy, (2) confirm nothing fundamental has changed, and (3) consider whether reduced position sizing (not stopping) is appropriate if the drawdown affects your psychology. Stop only if you discover the strategy itself is flawed.