• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering. Learn more Join!
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job. Learn more Join!
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models. Learn more Join!

Systematic Trading - there are secrets to the trade

Hello Quantnet friends,

This is Donny again, your friendly quantitative developer in Singapore. It's been a year in my hedge fund and the ride has been great. I've learnt many things one of which is that systematic trading isn't as easy as I thought it would be.

Here are my common observations of strategies using the usual indicators (MA, MACD, RSI, ADX, Stochastics, Resistance and Support Levels)

1. As advertised online, such strategies work great for 3 months in a year. However, when I backtest it for the whole year, it results in losses for the other 9 months resulting in an overall lost in the year.

2. There isn't a scientific way of specifying the parameters of an indicator other than shifting them back and forth and figure out which values give the best PnL. In other words, optimizing.

3. (Probably already known) Systematic trading does not take into account information of the asset other than price. Since indicators are only ONE step away from the price, it seems like this sort of trading is a matter of fitting a trade based on patterns.

With half a year repeatedly facing 1, 2 and 3, I've come to the conclusion that all the big boys making 10%+ a year returns - the Rentecs, the DE Shaws and the SACs - actually HAVE a secret algorithm that consistently make money. This algorithm is not formed from the basis of technical indicators but from the pillars of advance statistics and math. And why should it work? Well, my limited understanding would say that there is actually some development of a model that seeks to explain the market and its future behavior. Technical indicators just repaint the market based on what happened historically.

My guess is that this secret algorithm will not be found in any of those Wiley Trading or Wiley Finance books. Instead, they are found in academia in the form of a paper entitled something like "Optimal Stopping Time for Levy Processes" which took probably 3 months to go from theory to programming to rigorous backtesting finally to implementation.

What's your opinion on this? I value your feedback because in all seriousness, I need to choose wisely how I should spend my time developing a strategy - Option 1: Read half a dozen books from Wiley Trading avaiable on Amazon.com or Option 2: Use my stochastic calculus background and try to tear up some as difficult as this.


More to follow if response to this thread is good.
nice, but as you've said, technical indicators are useful for past references only.

there's no secret algo. all they have is a bunch of financial engineers to build lots and lots of models. that's what happened to the loss at jpmorgan on 1 of those mis-specified model.

taking things into account such as the factor models of inflation, expected unemployment/jobs reports, market corrections and other 1,000 dynamic variables that could be possibly correlated, models make much better predictors than tech indicators. And of course, one might need to take into account the possibility of a recession once in a while.