Stragedy break down:

A

amigasearch

Hi. I am new to systems development, i started because I have ideas I would like to make mechanical. I have posted before with some questions, and now, i have others. I hope I can get other system developers opinions/experience.

I have a programmed system, which preforms very well for all of 2003 (backtest). 65/35 winners to losers, $12 avg. trade, 10 percents max DD and max DD avg. is around 4 percent.
I factor in commsisions, and the like. Trades on S&P emini, and does avg. 10 roundtrips a day (2000 trades a year).

This system, I know I can improve (adjusting stops and the like), but that is another thread.

The problem - 2002 backtest result is poor, very poor. dont have exact figures in front of me, but it is all negative (I know that max DD is 56 percent, and avg DD is high also). No profit at all.
2001 data holds up - it does not neccesarily go into the red, but it is not profitable at all. I would not trade it though.

I know, that at the least, I would have to identify what type of market my system seems to work in. The problem is, I have no experience and dont know which path to take to determine what type of market 01, 02, and 03 was.

I like the system, but of course, if I start to trade it, market can "turn into" (like a werewolf!) 02 and my system crashes.

I can always program something to identify what type of market I am in, and then, create systems for each type of market I identify, having that system "kick in" accordingly. Is this feasible thinking?

Ok - now, am i approaching this the right way? I mean should I even consider a system that is nice, but just for this year?
Help is needed.... Thanks to all.
 
Thanks for that. That looks like my equity curve (of my system!) except that my gains for 2003 exceeds all my loss from prior bear years.

Here is the reason why I have not looked at long term charts yet - #1 I would not know what to do with it(I will explain) and #2 my app. only does intraday ( i am in process of programming daily / weekly).

So, back to #1. Now that I know ( and I suspected this, just did not look) that 2001 and 2002 were "bad" years, please tell me if this is feasible strategy:

1. Create a system that would work well in 2001 / 2002 market.
2. Have program identify if market looks like 01, 02, or 03.
3. Have program trade according to result....

There would be a look back period on determining type of market.

Does any of this put me on a positive path, or is my thinking poor? Thanks.
 
Your system does well in up markets and not well in down markets. Preferable would be a system that does well 1. in both markets; or at least 2. a system that does well in one market and treads water in the other market.

If you had traded live through the last few years, you might've been hanging on with white knuckles, sweat and that crunching feeling in your guts. Would you be able to live like that? After two years of poor performance, you might not have been too optomistic that your system would reap back the losses in the third year. How would you know?

It appears you may be able to use your particular system only during up markets. Perhaps a sentiment type indicator or longer term trend type indicators may be applicable to help determine the market's overall direction before taking any signals?
 
I understand. I might just scrap this system, or try to improve it to at least preform decent over all markets. This is my ideal goal, and reaching it I believe will be hard.
Any one else have systems in place that dynamically change with market sentiment.....
 
Quote from amigasearch:

Hi. I am new to systems development, i started because I have ideas I would like to make mechanical. I have posted before with some questions, and now, i have others. I hope I can get other system developers opinions/experience.

I have a programmed system, which preforms very well for all of 2003 (backtest). 65/35 winners to losers, $12 avg. trade, 10 percents max DD and max DD avg. is around 4 percent.
I factor in commsisions, and the like. Trades on S&P emini, and does avg. 10 roundtrips a day (2000 trades a year).

This system, I know I can improve (adjusting stops and the like), but that is another thread.

The problem - 2002 backtest result is poor, very poor. dont have exact figures in front of me, but it is all negative (I know that max DD is 56 percent, and avg DD is high also). No profit at all.
2001 data holds up - it does not neccesarily go into the red, but it is not profitable at all. I would not trade it though.

I know, that at the least, I would have to identify what type of market my system seems to work in. The problem is, I have no experience and dont know which path to take to determine what type of market 01, 02, and 03 was.

I like the system, but of course, if I start to trade it, market can "turn into" (like a werewolf!) 02 and my system crashes.

I can always program something to identify what type of market I am in, and then, create systems for each type of market I identify, having that system "kick in" accordingly. Is this feasible thinking?

Ok - now, am i approaching this the right way? I mean should I even consider a system that is nice, but just for this year?
Help is needed.... Thanks to all.
More...

1. win rate too high
2. not enough data
3. too many trades
4. win loss ratio too low
 
Trying to develop systems for short term trading is futile in my opinion. The ideal time-frame for systems development is midrange/longer termtrend following. Examine your own need to trade short term and see if you could gradualte/move up from that.
 
Quote from amigasearch:



I like the system, but of course, if I start to trade it, market can "turn into" (like a werewolf!) 02 and my system crashes.
More...

Problem is here.

Never favor, it creates a bias.
 
You must log in or register to reply here.
Back
Top