Expected Value vs. Survival Probability: The Trader's Dilemma

Expected Value Is Not Enough

Expected value (EV) is the average profit or loss per trade calculated over an infinite number of trades. It tells you whether your strategy has an edge but says nothing about whether you can survive long enough to realize it. The distinction between EV and survival was formalized by Ole Peters and Murray Gell-Mann in *Nature Physics* (2016): "the expected value of a multiplicative process diverges from the time-average growth rate — what matters for a real trader is the latter, not the former."

EV = (Win% × Avg Win) - (Loss% × Avg Loss)

A strategy with +$50 EV per trade will eventually be profitable over a very large sample. But "eventually" might require surviving a 60% drawdown first.

The Gap Between Theory and Practice

In theory, every positive EV strategy is worth trading. In practice:

• You have limited capital (finite bankroll, not infinite)
• Drawdowns trigger emotional responses (loss aversion is 2.5x stronger than gain satisfaction, per Kahneman & Tversky, 1979)
• Margin calls don't wait for the law of large numbers
• Compounding losses hurt more than compounding wins help (the asymmetry problem)

How Survival Probability Fills the Gap

Survival probability answers a different question: "What are the odds I make it through the next N trades without hitting my max drawdown?"

A strategy can have:

| EV | Survival Probability | Verdict |

|----|---------------------|---------|

| Positive | High (>90%) | Trade it |

| Positive | Low (<50%) | Reduce position size first |

| Negative | High | Slow death — avoid |

| Negative | Low | Worst case — avoid immediately |

Optimizing for Both

The goal is to find the position size where:

1. EV remains positive (edge preserved)

2. Survival probability exceeds 90% (ruin unlikely)

3. Drawdown risk is psychologically manageable (you can execute the system)

Simulation-based analysis is the tool for this. Testing different risk levels and comparing the survival probability of each highlights where the balance lies.

Use the Risk of Ruin Calculator for a quick analysis. RangeSight provides full simulation paths.

Conceptually this provides a framework. In practice, outcomes vary across simulations. Testing multiple scenarios reveals how consistent a strategy actually is.