Loading...
UNDERSTANDING SHARPE RATIOSv1.0

>What is a Sharpe Ratio?_

The Sharpe ratio measures risk-adjusted returns. It tells you how much excess return you receive for the extra volatility of holding a risky asset.

Sharpe Ratio = (Return - Risk Free Rate) / Volatility

In our game, we assume risk-free rate = 0 for simplicity

>True vs Sample Sharpe_

Understanding the difference is key to mastering this game:

TRUE SHARPE (Population Parameter)

The "true" Sharpe is the parameter we use to generate the returns. It's like the actual skill of a trader - constant but unknown.

Example: If true Sharpe = 1.5, the data generator uses 1.5 as the target

SAMPLE SHARPE (Observed)

The "sample" Sharpe is calculated from the actual returns you see. Due to randomness, it usually differs from the true value.

Example: True Sharpe = 1.5, but observed returns might give sample Sharpe = 1.3

>How Returns Are Generated_

We use Arithmetic Brownian Motion to generate realistic financial returns:

1. Start with a target "true" Sharpe ratio (-3 to 3)

2. Convert to annual drift and volatility

3. Generate 504 daily returns (2 trading years)

4. Each day: Return = drift + volatility × random_noise

Technical: We use Box-Muller transform for normally distributed noise

>Concrete Example_

Scenario: True Sharpe = 2.0 (Excellent strategy)

• Annual volatility: 20% (randomly chosen)

• Annual drift: 2.0 × 20% = 40%

• Daily drift: 40% / 252 ≈ 0.16%

• Daily volatility: 20% / √252 ≈ 1.26%

After generating 504 days of returns:

• Calculated mean return: 38.5% (close to 40%)

• Calculated volatility: 21.2% (close to 20%)

• Sample Sharpe: 38.5% / 21.2% = 1.82

True Sharpe: 2.0 → Sample Sharpe: 1.82

The difference is due to random variation in finite samples

>Why This Matters_

In TRUE mode: You're estimating the underlying parameter - like guessing a trader's actual skill level.

In SAMPLE mode: You're calculating from observed data - like evaluating past performance.

Real traders face this constantly: Is good performance skill (high true Sharpe) or just luck (sample Sharpe > true Sharpe)?

>Tips for Better Guessing_

• Sample Sharpe has more variation with volatile returns

• Over 2 years, sample usually converges toward true

• Extreme sample values (near ±3) often indicate lower true values

• Look at the consistency of returns, not just the average