Fire and Fortitude in Algorithmic Trading (Part 1)

Today I want to review where I’ve been (starting here) in this discussion of algorithmic trading and then move forward.

I have three main critiques of technical analysis (TA). First, it’s not always objective (as discussed here and here). Second, it’s rarely backed by supporting evidence. Finally, it seems really difficult to find viable trading strategies.

With regard to critique #3, I feel like I want to climb to the nearest mountaintop and yell “nothing works, folks!” An infinite number of possible strategies await to be tested, though, as mentioned in [1]. If I have 40 different indicators then I can test them on 20 different markets, throw in daily + 4h + hourly + 5m time frames, and voila: 3,200 strategies. What I have tested in my first three months is tiny in comparison.

Seldom have I experienced failure like I have in my brief stint as a quant. One by one, I saw hallmarks of popular TA fall by the wayside: momentum, moving averages, Bollinger Bands, RSI, Williams %R, trend-following, mean-reversion, etc. None of these have worked sufficiently well over a long period of time, and many are losers. I went through books with each setup presented accompanied by [deceptive?] claims of live-trading success on different markets. None of these worked either. Having immersed myself in the financial domain for a while, I’m not surprised to see an indicator or two not work very well. I have read books, articles, and sat through countless presentations these indicators over the years, though, and failure to find a single viable strategy, is quite the shock.

My discipline and commitment have eroded over time. I started by testing one strategy per day. As I learned the process better, I increased to two, five, and eventually 10/day. I started to run lean of strategy ideas. It’s very important to have this in strong supply because once my time had to be devoted to finding strategies, my daily testing decreased and I felt less productive as a result. Up until that point, I was able to achieve my daily goal, but as I watched the tally of total strategies tested continue to mount, my frustration and fatigue got real.

What I experienced is to be expected according to algorithmic trading guru KD. KD is a vendor of algorithmic trading education. He has my respect (until he doesn’t, and I’ll let you know if that happens). He says we can expect to find one viable strategy for every 100 – 200 strategies tested. He also says to break for a few weeks if you start to get frustrated and impatient.

In other words, those who undertake this trading approach will need to settle in for a very long haul…