Automated System Development

[rogerha]

Morning All,

As you will appreciate by my 'newbie' tag I am just that, been trading/learning for around 2 years. Like many traders have experienced, I have and still occasionally have trouble pulling the trigger. Therefore I have attempted to automate a trading system. Entirely my own creation using some reasonably well known indicators, I have learnt to program in E-Signal. Of course the beauty of learning to program is that you can optimise your system automatically over a period of 120 days (the maximum E-Signal can provide).

I have attached the E-Signal Backtest file for comment and perusal. Currently it is running to provide the optimum combiation of Profit and Minimal drawdown, more profit is available out of this system but (not suprisingly) at greater risk.

I would be very keen to receive any comments or advice on this, as I say, I represent no company and neither do I profess to be the best trader in the world. All that I have done is learn to program in order to overcome the biggest trading problem I personally have.

The system is not operational as yet. Also please ignore the 'Contracts' tab in Trades. It is a fudge to get E-Signal to represent a single Euro FX Futures contract.

This is based on the last 120 days (I realise that probably isn't enough) but I would like to gauge its performance to see how it compares to some of the more successful traders on the forum.

Thanks for your time.

 

[rogerha]

Should have mentioned, this is based on trading a single Euro FX Futures Contract on a $10k account.

 

[TWI]

Forgive me if I have misinterpreted your post but I think the fact you use the word "optimise" in the context you do suggests this is a little curve fitted to the last 120days of data. Optimisation is a wonderful tool for finding great historical trading systems that become real-time disasters. Keep watching your system in real-time and see if the results you get from here on emulate those of your historical data. If you want a demo of this more quickly, optimise over the first 60 days of data and then test it over the full 120 days. Will be interested to hear the result.

 

[rogerha]

Well I have been running the system (live Paper Trading) for 3 months actually, and yes the real-time trading matches the patterns shown in the Analyser.

The real truth is, I only just got the hang of the Analyzer tweak to get it to represent my real trading results (if that makes sense). Chicken and Egg, but the real trading came first. The idea of the 'optimisation' was to tweak the configurable parameters to increase the system potential.

Does that make sense ? I think what I am trying to say, I have used the backtesting to enhance a working system.

I agree though, the real proof is in the pudding, and the next 120 days will be key. But we all have to start somewhere

 

[cd173]

Hi rogerha

Have you tried trading these settings in real time?
You may find it gives somewhat different results than the backtest.
IMHO paper or real time trading is the only real way of finding out if it's going to be profitable when you pull the trigger.

Cheers
C

 

[TWI]

Why do you consider the drawdown is so much greater for the short than the long trades in your system?

 

[rogerha]

Twalker,

I must confess I cannot see a specific 'reason' why that is the case. Its not as though the Euro has been trending generally up or down for the entire period or anything. There is no long/short bias applied to this system so its nothing artifically creating that scenario.

 

[rdstagg]

System Development

Without knowing how the system is based its a little difficult to pinpoint any areas for inspection.
Some general comments therefore:-

1. 120 days is a v small sample of time - unless you're trading 3 - 4 times a day of course!

2. Maybe the shorts have a different flavour to longs - the moves may be quicker and sharper rather than the longs which take more time. Shorts tend to be born out of fear/panic whereas longs are fed on greed.
Try optimising parameters for longs and shorts separately.

3. be very careful of curve fitting

Have fun!

 

[rogerha]

Appreciate the advice RDStagg. You seem to understand, I don't think by any means that I have a definitive working system until of course I start trading it live.

 

[rdstagg]

Platform?

What platform do you use?
You might want to spell out the strategy rules - in my humble opinion you will get more response (good & bad) and you may find someone has already invented a better mousetrap. At the very least it will leap you forward - I guarantee it.

I have learnt over the years of trading that even if you give someone the holy grail they still have to actually TRADE it.
A system is merely the start.

Hope this helps.

 

[fowkesp]

rogerha,

One thought - try putting some realistic figures in there for slippage and commission, unless your system already accounts for these. If you have ignored S&L, then your results will be considerably different, as S&L will consume a significant proportion of your average winning trade value.