My system results... what do you think?

[jmcguilly]

Quote from maos:


that drawdown translates to less than 30% of capital traded at that time, and for me that is acceptable.

The problem with developing good trading systems is that we all use past historical data. Markets can and do frequently change and past assumptions and methods either become less profitable in the future or don't work at all. Old style scalping methods when decimilazation happened is one example.

Many times when I have developed and traded systems with real $ the 'drawdown period' is usually when you first start trading it. I have stopped trading systems in the past due to the drawdown and lost money where if I had continued the system would have been profitable. It's easy to say you will TRADE THE SYSTEM but if you have a couple of big losing trades in a row it is HARD.

Also keep in mind that if you lose 30% right off the bat you need to make 40% just to get even. It's very important, in my opinion, to have systems with very low drawdown. I've found better success at having a range of systems and understanding which system works in what type of market environment. Constantly check the real results of your systems and trade the ones that are working in TODAY's Market!

The most profitable system you develop may NOT be the best system for you to trade. I find systems that are less profitable but have fewer consecutive losing trades or higher consecutive winning trades far easier than ones that may win around 50% or less but have high risk/reward ratios. Having 3 to 5 losing trades in a row is easy to understand by looking at a backtest report but is not too fun to deal with using YOUR hard earned money.

 

[Samson77]

Quote from maos:

Samson,

be rest assured I am not selling anything.

I just wanted some opinions because I'm planning to begin trading it next week...

I was just teasing you but please be careful trading this I have seen several systems with similar results blow out bad in live trading.

 

[maos]

Quote from jmcguilly:


Many times when I have developed and traded systems with real $ the 'drawdown period' is usually when you first start trading it. I have stopped trading systems in the past due to the drawdown and lost money where if I had continued the system would have been profitable. It's easy to say you will TRADE THE SYSTEM but if you have a couple of big losing trades in a row it is HARD.

jmcguilly,

I completely agree with you that going through a drawdown (especially a 30% one) must be like hell itself.

But I truly belive if you start trading the system YOU developed from scratch, if you know why is it going through a drawdown at any given time, it just might be a little easier to bear through it.

You could say : reduce your position size, you will make less profit, but the drawdown will be smaller and thus ensure you stick long enough to reap those profits.
But the thing to consider is that I have only 30,000$ to invest, 150+ watchlist to monitor, and easily 10-20 signals to trade at one time, which I have to take to ensure equity curve smoothness.

All that translates to less than 100share position in more than 70% cases at the beginning until my account grows, and I do not want to trade odd-lots, because I'm afraid that my executions would go to hell.

So, I think that 30% drawdown wait-out for me , at this stage, might be inevitable...

 

[man]

sharpe ratio of 1.3 and profit factor of 1.5 are relatively low IMHO, since the 150 stock universe is probably subject to several ex-post factors. i would recommend you paper trade it until you have forty trades and compare the two distributions of results.

some poster mentioned slippage of 1.7% - i admire everybody who can build a trading system with average trading length below two months on that transaction cost. hats off.

peace

 

[ssternlight]

Looking at the equity curve it looks as if you might not have taken all the trades -- i.e. run out of capital. You might check the number of trades by using a constant position sizing to see how many trades you are missing and how the trade performance varies as a result.

Another idea is to set the simulator to take the worst trades first to see the grim reaper scenario.

It would be interesting to start testing in ' 03 to see more recent performance -- and bypass the big up and down of previous years. This will probably yield a more accurate idea of current expected returns.

Finally, start the system at the beginning of this year and look at performance now.

I suggest the above as starting live trading when a drawdown begins is always a possiblity. Your ability to hold through it is frequently related to the size of the drawdown...and of course your available capital and ability to withstand the pucker factor.

Regards,

Sam

 

[vikana]

I think your system looks promising. Given the length of the trades, there is one final test I'd do. If you can buy and sell and the worse possible price of of the day and stay profitable, I'd put money on it.

E.g (long). buy at the high of the entry bar and sell at the low of the exit bar. This eliminates the need to worry about slip etc and gives you a realistic worst case scenario.

Edit: One final note. Just trade 100 shares per position for a month or so. You'll then know how close to simulation you are. I always do this.

 

[maos]

Thank you all for your suggestions, they are very much appreciated...

vikana, I will try your test, I haven't thought of that myself and it's the kind I like the best...worst case scenario.

 

[maos]

I did the test vikana suggested (long only)...not good

take a look, tell me what you think...

 

[maos]

equity curve...

 

[ssternlight]

Maos,

The equity curve suggests that you are running out of capital with this system.

That means you are not seeing all the trades it generates. Since you don't have any idea as to the intraday sequence from which trades get picked up it might be the case that early trades -- the one's you get in reality -- underperform the full trade set. This happens frequently with dipbuyers.

Consequently, you can't place complete reliance on these results. I'd suggest you test with infinite capital -- $1M should do it -- and see what the total package looks like. Then get some intraday data and test off of that using a time filter to more finely hone the approach.