Truth in Backtesting (Part 11)

In my last installment of this blog mini-series, I discussed… well honestly, I don’t know.  That was well over two years ago!  I am happy to have found this link, though, because now I have some items to add to it.

I just finished reading Building Winning Algorithmic Trading Systems by Kevin Davey (2014).  For anyone looking to get into algorithmic trading system development, I would definitely recommend this book.  I’ve read a few books on the subject and this gave me a few new ideas that I had not read before.

According to Davey, incubation is the phrase of trading system development where I should wait and watch system performance before I start trading with live money. 

One reason Davey doesn’t feel the need to incubate with any real money (even in small size) is because he is careful to avoid certain backtesting pitfalls he feels would falsely exaggerate live-trading performance. The first such pitfall to avoid involves buy (sell) orders that fill at the low (high) of a bar.  This rarely occurs in real life, Davey says.  Unscrupulous system vendors and naive developers produce strategies that frequently include this.

Davey also makes a point to avoid exotic technical analysis templates like Renko, Kase, and Point-and-Figure charts.  “The way these are built from history,” he writes, backtesting fills often cannot be believed.”

Either of these pitfalls could make a trading system perform better in development then they are likely to trade in real time.  They are therefore good vehicles for misrepresentation and that is why I have categorized this under optionScam.com.

I will conclude with the next post.