I created a system, what next!?

Hi all,
Ok i have devised a system that works well, the drawdowns are minimal and the losses are managed.It has short losing streaks and is very comfortable.
My questions are:
I have backtested it 9 months is this a long enough period to make a proper judgement of the system?

What are the best statistical analyses to apply to the data and why.

Thanks in advance and happy trading!

I would spend another 9 months on forward testing. Backtesting offers quite a limited outlook of how the strategy will perform in real world.

Backtesting is always inaccurate from my experience. I have spent months, nay years, backtesting systems on paper only to find that they did not work as expected in forward testing. Why? Mostly for practical reasons such as market times and psychological constraints such as failing to understand real time experience. Lower time frames are particularly difficult to test because of slippage and spreads.

Note that above I said "as expected". So I will state confidently that I feel backtesting is essential and it also helps understand the difference between expectation and reality.

Furthermore, it's important to look at market conditions - stuff like the carry trade between 2000 and 2008. Anyone with more than one brain cell and a bundle of money could have ridden that train and fallen off it in 2008. So will your strategy work in all market conditions?

I'm sorry I didn't answer your questions directly, even though you probably got my points.

You don't mention time frames or what pair? I think 9 months is a long enough time to back test any strategy based on H1 or lower time frames, but maybe not on Eur/Chf, to pose an extreme example.

For statistical analysis the first priority is to calculate expectancy and then mold a risk model based on number of trades in time and available capital. Quite a lot to learn about in that sentence.

Forward testing with a small account beats any backtest! 9 months is plenty, but there are never any guarantees in trading.

As others have suggested, you may want to try out some forward testing, back testing can only do so much as it is based on old data and may not reflect current market scenario..

I highly recommend to test it on a lot of pairs, and time periods, maybe from the 70's to now. (Euro crisis is good for check the strats). As smikester mentioned, you should test it in all market conditions. I'm using only one indi for my "holy grail" strategy, but I have to change the parameters for the CHF pairs... IN SUM: Test it fully, analyze it, than test with demo, and live (low amount of mony first of course), and gooooooo burn the market

Good luck!

How many trades the system has closed is far more important than the actual duration of the backtest. 20 closed trades is not giving a reliable idea of how good the system is, while 200 is pretty solid.

Hi all,
Thanks for the replies!
I have began forward testing yesterday and plan to test around another 6 months and if it is still proving well i will take it live.

I am using Eur/Usd only at present as i know this market inside out, the strat is based around 15 min tf.
I have 183 trade sample set at present.

Thanks very much for the help and if anyone can shed any extra light on statistical analysis it would be greatly appreciated.
I have plotted the equity curve so far and is a steady slope.

thanks in advance

I would suggest you go live right away as part of your testing instead of using a demo account, but with a broker that allows you to trade micro or nano lots (10 cents/pip or 1 cent/pip). Demo accounts sometimes act differently than live accounts and often will not show you the true performance of your trading system.

I would definitely second this post, it makes more sense to start testing on live accounts but nano ones..

When I back test a system I first test the last 3 months then go to 6 months if 3 months looked good win/loss and draw down, next I go to 1 year repeat same process if all looks good test on 5yr time frame repeat again. Next I start forward testing.

This way I have multiple types of trading environments that my system have been tested in.

Minotaur,

You need to read this blog. It's an absolute must-read for anybody contemplating the use of a completely mechanical system, whether automated or manual. These guys knew what they were about; soultrader was something of a legend when he used to post here at FF.

There's also some outstanding info on system development here and here.

Good luck.
David

Thanks Hanover,
Informative stuff,
My system is fairly well factored around what i like to call the unknown element.

I have taken the principle of the market being completely unknown and designed a system around it , In fact i have spent months/years designing a system around losing and loss management it is a result of that.

About back testing i have tested by hand the amount of data available on 15min tf, but unfortunately my charts on mt4 only go back this far, if you could point me in the direction of longer data i will happily test as many years as possible!

Unfortunately i dont have the starting capital available to begin a live account even if it is micro/nano. I am in the process of saving now.

I have been through the process of blowing account after account and i even suffered a HUGE blowout (the last time i traded live) and i will not commit any funds that i should have for my family.

Slowly slowly catchee monkey!

I'm in no rush to be full time trader just enjoying the process!

This is gold, hanover!

Gold!