Position Sizing in Weather Markets: Applying the Kelly Criterion
TL;DR / Key Takeaways
- Kelly sizing is useful for thinking about edge, but full Kelly is too aggressive for noisy weather markets.
- Capital preservation matters because even positive-expectancy systems experience losing streaks.
- Fractional Kelly reduces drawdown while still scaling position size with estimated edge.
- The sizing model must account for probability error, fees, and binary contract payoff structure.
The Mathematics of Capital Preservation
The most sophisticated NOAA weather ensemble pipeline in the world is useless if you risk twenty percent of your account on a single trade. In algorithmic trading, your edge dictates your frequency, but your risk management dictates your survival.
To optimize growth while minimizing the risk of ruin, quantitative systems rely on variations of the Kelly Criterion. This formula determines the optimal size of a series of bets based on the probability of winning and the payout odds.
The Formula and the Edge
The standard Kelly formula is: K = W - [(1 - W) / R]
Where W is your winning probability, and R is your win/loss ratio.
If our 62-member HGEFS ensemble indicates an 80 percent probability of convergence (W = 0.8), and the Kalshi market is pricing the contract at 60 cents (meaning a 40-cent profit against a 60-cent risk, or R = 0.66), the formula outputs a positive percentage.
If the Kelly formula outputs a negative number, your perceived edge is an illusion, and the system must abort the trade entirely.
The Fractional Kelly Approach
While the pure Kelly Criterion is mathematically optimal, it is highly aggressive and assumes zero variance in your probability calculations. Weather modeling inherently contains variance.
Therefore, the Predict & Profit system utilizes a "Fractional Kelly" approach. We calculate the optimal mathematical size, and then strictly divide it by four. Furthermore, we implement a hard global cap: no single weather contract can ever exceed two percent of the total portfolio equity. This strict, unemotional sizing protocol ensures the algorithm survives anomalous weather events.
Frequently Asked Questions
Q: Why not use full Kelly sizing for weather markets?
A: Full Kelly assumes the estimated edge is accurate. Weather probabilities are noisy, and binary markets can have clustered losses, so full Kelly can create unacceptable drawdowns.
Q: What does fractional Kelly solve?
A: Fractional Kelly scales position size down from the theoretical optimum. That preserves some edge-based sizing while reducing volatility and model-error sensitivity.
Q: How should fees affect Kelly sizing?
A: The input edge should be calculated after fees and execution costs. Using pre-fee probabilities overstates the bankroll fraction the system can safely risk.