Systems that don't work anymore?

T

travis

I have been doing some testing on a system that exploits the short covering that happens in the last 5 minutes of trading before the 4 pm close.

It works, really well, for the past few years since 1999, but from 2004 on it just breaks even.

Do you figure it's because so many people know about it and use automated systems to trade this method, or is it because the markets have been different in the last two years?

If it's the latter, then I can still trade because it may start working again. On the other hand I don't see how the markets have been different in the past two years. At least not in a sense of falling versus rising, because the system works in 1999, 2000, 2001, 2002 and 2003.

The only reason I can see is that too many people have learned about this and have eliminated that inefficiency.
 
a lot of systems have gone sour in this low-volatile, highly competitive environment, maybe there hasnt been enough volatility to make it worthwhile from 2004 onwards, just look at a 10yr graph of the ATR of any stock to see what i mean.
 
It sounds like you are too far deep into the numbers part of the game.

When you are testing a system, think about what that system is actually doing. Why does it work? Who is participating in your trades?

In your case, you might be dealing with a situation where a large imbalance of buying before the close happened (such as a bull market, when people are buying for the next day's jump) or as you say, short covering (such as a bear market, when people are covering from the long short all day).

That may not be the case in the past few months/years. Everything is cyclical, and you really have 2 choices:

1. Go with the flow, win big in the markets mentioned first, breakeven in the later markets
2. Develop a more diversified "system" of strategies, where you are always just winning small.

The later is preferred of course, if you are to raise capital or similar. Cash flow is always preferred to be stable -- something that is the trickiest part of this business.

Good luck with your testing!
 
Equity curve attached, in response to opmtrader.

To jonnysharp instead I have to say - I have indeed checked out the volatility of the last two years, and he's right in saying that there was a sharp decrease.

I used True Range divided by the index value, applied on a daily time frame. Well, exactly when my system breaks even we have had an average daily True Range of 1% to 2% of the index value, whereas from 1999 to 2002, when the system does well, the daily True Range was much higher, from 2% to 4%, with many peaks of about 6%.

Now I would like to ask him or anyone - when will the true range increase, and why are we having such a low daily true range?
 

Attachments

The average range today relative to historical data is normal.

The ATR of 1999 to 2000 is abnormal.
It takes sometime to revert back to normal thus 2001 to 2002 ATR is still too high.

Use data during the old Bush time to compare against the current period, you will find that there are a lot of similarities.
 
You must log in or register to reply here.
Back
Top