I was trying to figure out the variance between my trading system stated APR on live trading and my actual results in live trading. To me, I thought I should be getting similar APR results when going live. This past year, I have been using a a main dataset. However, I could not figure out why this dataset was getting much better results on my trading systems compared to other datasets such as the larger popular indexes such as S&P 500, Russel 1000, Russel 2000, etc. I did notice that I would sometimes get lower priced asset signals and I would just ignore those. I did not think too much of this, knowing that I will eventually add a minimum price and minimum volume filter. I only like to trade stocks with higher volume and not like to trade lower priced assets. Anyway, once I put these filters in place, my dataset results with my strategies has dropped the APR dramatically. So it makes sense that if I don't take all my mechanical trading signals, my results will vary from actual results. This is not even considering the variance you get from past results don't equal promise of future results, slippage, etc. I guess one more piece to the puzzle... Thanks, Larry
This seems completely obvious to me. Never done any sort of random filtering such as this for decades. If you develop a system, you have to follow it as closely as possible. Any untested and subjective biases are not allowed.
transaction costs are a huge factor with low priced stocks. they will appear more profitable if you are not accounting this properly
D08, It was not obvious to me, but in hindsight it should have been. These are brand new strategies so live trading with $$$ shows flaws in my system logic that I did not account for. For my analysis on my trading system, much of the profits are from much lower asset prices that I am unwilling to trade on. Only after I saw the deviation, did I figure this out. Thank you, Larry
This is quite typical, same reason why lower market cap equities produce the bulk of profits due to more volatility, even if the liquidity filter is constant. Simply test the system as you would trade it, don't add any qualifiers in production.