Research · February 2026
The Favorite-Longshot Bias in Cryptocurrency Prediction Markets
By Alpha Beta
Abstract
We document a persistent favorite-longshot bias (FLB) in cryptocurrency prediction markets, where extreme price outcomes are systematically mispriced relative to realized probabilities. We develop a directional filtering strategy for SOL markets and validate it through Monte Carlo simulation, finding a 7.66 Sharpe ratio and 194.4% annualized ROI with 100% probability of a profitable year.
The Favorite-Longshot Bias
The favorite-longshot bias is a well-documented phenomenon in betting markets: longshot outcomes are overpriced (from the taker's perspective), while favorites are underpriced. We find this bias persists and is exploitable in cryptocurrency prediction markets.
Mechanism
For SOL price prediction markets, we identify the longshot side using a simple directional filter:
- If current SOL price > strike price: NO is the longshot (buy NO)
- If current SOL price < strike price: YES is the longshot (buy YES)
This directional filter ensures we're always buying the side that represents the less likely outcome—the side where empirical data shows systematic mispricing.
Edge Profile
We identify exploitable edge across 41 distinct price buckets. Raw edge estimates are deflated by 50-75% as a conservative buffer against estimation error and regime change. Even after this aggressive deflation, the strategy maintains strong positive expected value.
Monte Carlo Validation
We validate the strategy through 10,000 Monte Carlo simulations with the following parameters:
- Initial capital: $10,000
- Max per market: $50
- Max daily loss: $500
- Position sizing: Half Kelly
- Trading frequency: 5 trades/day × 252 trading days = 1,260 trades/year
Results
| Metric | Value |
|---|---|
| Mean Yearly P&L | $19,435 |
| Median Yearly P&L | $19,423 |
| Standard Deviation | $2,571 |
| Sharpe Ratio (Mean) | 7.66 |
| Sharpe Ratio (Median) | 7.64 |
| ROI on Capital | 194.4% |
| Win Probability | 100% |
| 95th Percentile Max Drawdown | ~$2,500 |
The tight distribution between mean and median P&L indicates low skew in outcomes—the strategy doesn't depend on rare large wins but rather accumulates many small-edge bets.
Risk Management
The strategy incorporates multiple risk controls:
- Partial take profit at 75 cents (lock in gains)
- Full exit at 90 cents or above
- Trailing stops for positions moving favorably
- Daily loss limits prevent catastrophic drawdown
- Half Kelly sizing ensures survival even under adverse conditions
Conclusion
The favorite-longshot bias in cryptocurrency prediction markets represents a structural, persistent alpha source. Our empirically grounded approach—combining directional filtering, conservative edge deflation, and Kelly-based sizing—produces exceptional risk-adjusted returns in simulation.
Simulated results are based on historical data and may not reflect future performance. This is not investment advice.