Does Any Technical Analysis Work? (Part 1)

I went over my second critique of technical analysis (TA) here. My third criticism is that finding a viable TA-based strategy is extremely difficult.

As discussed in this second paragraph, my goal with trading system development is to find viable strategies: not necessarily profitable ones. I will go into more detail about what makes for a viable trading strategy in future posts.

As discussed in the second paragraph here, in order to test strategies I need objective, codable guidelines. Many, but not all, TA-based strategies are this way.

I borrow from Kevin Davey’s approach and define a strategy as a particular code and/or particular time frame and/or particular market. If I take three different codes, for example, and test them on five different markets and two different time frames, then I have studied 3 * 5 * 2 = 30 different strategies.

I have tested nearly 300 strategies on three different markets. I would guess this to be a lot relative to the average trader because most people don’t learn to code, learn to test, or learn to develop strategies. As I discussed in the linked post from the first paragraph, most traders don’t show or provide any data.

Nevertheless, 300 strategies just barely qualifies me as a newbie when it comes to algorithmic trading. I’m not talking about the options work I have been doing for the last several years. I’m talking about my work on futures markets. Thus far, I’ve tested strategies on US equities, on gold, and on crude oil. What has taken me months would take a seasoned quant days (hours?). I am still learning how to code, still finding strategies, and still trying to figure out how I plan to organize all this.

My initial impression thus far is that finding viable trading strategies is really, really difficult.

Many things about the financial industry and trading domain do not surprise me. I’m not surprised by the fraudulent claims. I’m not surprised by the statistical manipulation. The whole debate about whether successful traders actually exist implies not only that something works but also that someone has discovered it. I’ve questioned whether a better way exists somewhere.

In discussing all these topics, I’ve been talking around the issue of viability.

With regard to viability itself, what does surprise me is to fail time after time after time—nearly 300 in all—at my attempt to find it. I have not found any strategies that meet all my criteria. I have not found any strategies that have compelled me to start trading them straight away.

I will continue this discussion next time.