Statistical Inferencing for Quantitative Trading Strategies

Have you asked:

1. Is my quant trading strategy performance statistically significant ?

2. Are my in-sample performances statistically significant while controlling for model complexity and bias? Is my ML model an inefficiency detector or a piece of overfitting poppycock software?

3. If I backtest 10 strategies, pick those with Sharpe > 1, am I headed for wealth or ruin?

   
   

Statistical Inferencing for Quantitative Trading Strategies is one-of-a-kind quantitative lecture series on applying probability theory and statistical methods to construct robust hypothesis tests for validation of trading strategies using distribution-free methods.

The course takes the student on a whirlwind tour of finance basics, statistics basics as well as more advanced and modern techniques in statistical decision/inferencing theory.

Hypothesis testing concepts, Type I/II errors, powers, FWER control, multiple testing frameworks are introduced under both parametric and non-parametric assumptions for quantitative research.

Classical location tests (t,sign,rank-sum) tests are discussed in addition to cutting edge techniques using monte-carlo permutation methods. The lectures take you through the motivation for the need to employ rigorous scientific procedures in validating trading strategies.

In pharmaceuticals, medicine and other high-stakes industries, experimental design and implementation are key to decision-making, such as the acceptance of new chemicals in treatments. Unfortunately - hardly the same amount of scientific rigour is paid in deciding whether to take a trading strategy live. Apparently, moon cycles and lunar phases are enough! For these people, the writing is in the wall.