One of the earliest trading programs I wrote (circa sophomore year of high school), traded on momentum. It acted quiet naively which is a proper reflection of my mind at the time. If the slope of the price for an asset was greater than and less than some prespecified values the asset was bought until it was lower or greater than a terminal slope value; The inverse applied for shorting. My program then searched for the best combinations of entry and exit slopes over a range of different historic windows.
I was quite excited at first when I found some bots that did extremely well. Then I had the insight to try testing the bot on a different time period to confirm they did just as well; They did not. I am actually proud that I recognized the need for out of sample testing. I think at the time the math taught in my connections class (read: dorky kids) was trig/precalculus. I had no knowledge of statistics outside of averages.
The point to take away from this experience was my sixteen year old brain’s expectation of herding behaviour; It is instinctual. Behaviour finance expounds on this concept. At my young age I believed that if a stock was moving up, people would pile on; If a stock was moving down, people would abandon ship. Eventually, the fundamentalists would correct very large errors, but in the short term, I thought the technical traders set movements.
Currently, I believe that momentum is a huge factor for a certain class of trader; This class is not sophisticated. It may be possible to exploit their actions, or their likely actions, but since they probably dominate only over the short term, their is too much noise relative to signal.
Some people (in this case momentum traders) behave in a predictable way. This capacity for prediction does not mean profit. Predicting the movement of one fish in a school does not mean you know the school’s trajectory — although it might give a hint that is occasionally correct.
