Pitfalls of various trading systems

Financial Trading in R

Ilya Kipnis

Professional Quantitative Analyst and R programmer

Pitfalls in trading system development

Market data is a mix of fear, greed, and noise of millions
“Past performance is not indicative of future results.”
Overfit on past (in-sample) data means bad performance on future (out-of-sample) data

How to not overfit

Can cause a system to fail in the future
Minimize the number of moving objects!
GOOD strategy
BAD strategy

Stability with system settings

-System should behave similarly for similar settings levels

Stability with system settings

-System should behave similarly for similar settings levels

Hypothesis testing

Perform hypothesis tests
- Relationship between an indicator & future returns?
- Signal process to generates outperformance?
Most of these are beyond the scope of the course, but keep them in mind

Let's practice!

Financial Trading in R

Preparing Video For Download...