That’s a pretty good test of the structure of a price curve, and actually informs us why price curves are fractal (if you think about it a while).
Basically it shows that there’s generally a swing big enough to trigger a stop loss, more frequently than there’s a straight shot upward (or downward) to the take profit level.
That implies the market must have a fairly specific curve shape. Even more so, I’ll bet that it’s true across just about all time frames… even more closely specifying a type of curve. That solution is the fractal curve we see… or rather, the fractal curve we see is one member of the family of possible solutions.
It’s not perfect of course ~ there are news events and fundamentals and things like that… but basically the price curve is driven toward this shape, through the combined actions of all the market participants trying to exploit an edge.
Essentially every predictable, stable technical edge will eventually be hammered into nothing anyhow, by trading algorithms in a finite amount of time. Which is not to say that it’s unprofitable… rather, it’s not likely profitable forever. It’s entirely possible to craft an algorithm to seek out any and all advantages by itself, simply by giving it a wide variety of fundamental waveforms and allowing it to ‘weight’ each wave component until there is as close a match to the solution curve as you want.
Or to say that in a far simpler way: if there’s an obvious steady rising line, every trader is going to trade profitably against it until the trend stops rising. Rinse and repeat for falling trends, obvious patterns, and so on.
It is of course possible to ‘overfit’ also ~ that is, create a function that perfectly mirrors the past but is a poor predictor of the future. I’m sure some coders that create such algorithms can get into that in a bit more detail; you obviously don’t want to overfit a predictive algorithm because once you do, it starts to become more and more useless.
So where is the window for success? Certainly there are opportunities.
As best I can tell (and I am no expert!) they have to be in the following solution spaces:
a) based on market fundamentals over finite time scales (e.g. dollar trending this week, that sort of thing ~ catch it before the trend fades)
b) based on market factors, for instance, if there generally aren’t enough market participants during asian session to break through support or resistance areas established prior, that’s exploitable (of course I just made that up, but things ‘like that’ could lend an edge)
c) based on weakly arbitraged timescales (say no bank would waste its valuable time trying to scalp every last crumb of value out of the market, but those crumbs are still left on the table and worth a lot to your average scalper)
d) based on peculiarities in how the market movers (banks) trade ~ for instance the use of institutional levels, and being able to spot patterns of behaviour
e) probably other stuff like the above (again, remember I’m no expert trader!)
Ironically, the stupidest strategies such as say, buy X, wait a while, sell X… buy X, wait a while, sell X… actually can work while a market trends upwards. It might even work for months or years.
So is it the strategy… or is it knowing how good of a strategy you’ve got, and when to use it? That’s a big question right there.
