I currently use a strategy that appears to have modest performance when tested historically over the last 12 years.
- I would like to get a feedback on my stop loss policy (limit portfolio loss to -6% and individual stock loss to -8 to -10%). When I analyzed the historic returns for portfolio losses exceeding -6%, it happened only 4% of the time in the last 12 years. I am not sure that keeping the weekly portfolio losses to less than-6% for only 4% of the time will impact the future performance of the system that much. In fact it improved the returns and reduced the risks on historical testing.
- Some of my friends who are more knowledgeable and more experienced than me are advising me not to be in the stocks for the time being because of the current and anticipated market conditions. My preliminary analysis of the historical return of the combined strategy tells me that I may lose as much as 2/3 of my total return over a 10 year period if I opt to be in the market only when the conditions are ideal for stocks (based on broad market index performance). Should I be in the market all the time trading stocks given the historical system performance- both in trending and trading markets?
- Finally I am reading an excellent book by Leonard Zacks (the founder and CEO of Zacks Research). This book is titled The Handbook of Equity Market Anomalies. On page 303 of the book Len Zacks discusses about Data Traps in Backtesting. Specifically he mentions three potential biases in back testing which could affect the future performance of a system which is perceived to be successful on backtesting.
- In-sample versus out-of-sample performance
- Survivor bias and
- Restatement bias
How do we avoid the biases? He gives an answer for avoiding the first bias (sampling). What about the other two?
Sorry for a long communication. Any help is appreciated.