Marcos López de prado and Frank J. Fabozzi - 2019
Avoid backtest overfitting
In tournaments, researchers can't run multiple backtest until finding a strategy that fits the past curve, but will obviously fail going forward.
and breath for analysis
Tournament helps explore more investment opportunities and produce more forecasts per investment opportunity.
Avoiding false discovery
By obfuscating our dataset, data-scientists are not subject to selection biases or false assumptions.
Each data-scientist brings its unique skills and approach. It increases precision, originality and brings an edge to our signal.