Quantifying Backtest Overfitting in Alternative Beta Strategies