If a System Looks to Good To Be True, Then It Is!

by Mark Williams on February 20, 2009

So, I have been on a good many trading message boards, and on these boards, I always run across some douchebag (for lack of a better term to use) who proclaims to have the holy grail of a system and that everyone should use it, blah blah blah.

And usually when this happens, I either ask for backtested results, or for them to show the people in the board the system.

Then of course, if I see the following, the red flags come up:

  • If I see an outrageously complicated beast of a system
  • Backtested results that look to good to be true
  • Stochastics

I have yet to have a MAJORLY complicated system work well in the long run.

EVERY overly complicated system that I have tested has fallen flat on it’s face at best, or blown up the person’s account at worst (overly complicated meaning greater than 7 indicators or indicators of indicators of indicators).

You’d think that the more indicators that you add, the better the system should do, but I guess because of all the conflicting signals that one can get with said system, that is not the case.

Now, there are complicated systems that do work, but this leads us to…

Backtesting results.

Now, if the person has backtested results that look to good to be true, I usually suspect a phenomenon known as curve fitting. Curve fitting is the accidental over-optimization of a system for a small sample of data.

For example, let’s say that I test a system on the Standard and Poor’s Index SDPR fund (Ticker: SPY) only from the dates of Aug 2003 to Aug 2008. During this period, I notice that from January 2004 to November 2006, I had a huge drawdown because of a massive amount of volitilaty during this period.

So, to compensate for this, I add a lot of volitilaty filters and that drawdown disappears and my results look AWESOME!

So then, I put the system online, and commence to lose my ass!

Why’d this happen? Well, I compensated for the drawdown for one period in the past, however I did not increase my sample size (add a few more years like at least 10 or so, or change dates to further time in the past)  therefore, I nver did test how robust my system truly was.

So, because of my faulty backtest, I was lulled into a false sense of security and did not realize that my system would work very, very, very well…If it were Aug 2003 to Aug 2008 😉

Finally, if the system contains Stochastics, MACD, multiple EMAs and SMAs, and the system gives off great results, shenanigans are being thrown. The reason being is due to another phenomenom called “Trader Effects.”

I will go over “Trader Effects” in more detail in another post, but in a nut shell, trader effects are caused by a system, or set of indicators, being extremely well known to the point where EVERYBODY (and brokerage houses) begin to use to them. Because of this, very smart and scrupolous traders/Hedge funds will either front run orders, OR WORSE YET,  purposely take a massive position on the other side of your trade whipsawing you out and giving them profit.

Furthermore, and this tip may hopefully save you THOUSANDS, unless you have read this too late, the MAJORITY of trading systems and trading classes sold teach these techniques. Not to say that Stochastics, MACD, and co will not work, but because everyone and their mother learns how to use these indicators, how can you find major edge from them?

bookmark bookmark bookmark bookmark bookmark bookmark bookmark bookmark bookmark bookmark bookmark bookmark bookmark bookmark bookmark

{ 0 comments… add one now }

Leave a Comment

You can use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

This site is using OpenAvatar based on

Disclaimer | Privacy Policy © 2007-2008 YouSuckAtTrading.Com All Rights Reserved.