How To Develop Winning Automated Trading Systems | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Top 100 best Trading Systems: post good Trading Systems here 62 replies

Automated or semi automated trading? 3 replies

do automated trading systems work or... 21 replies

Automated Trading Systems 18 replies

Systems, Systems and Systems... 1 reply

Attachments: How To Develop Winning Automated Trading Systems
Exit Attachments

How To Develop Winning Automated Trading Systems

  • #1
  • First Post: Jul 22, 2017 10:33pm Jul 22, 2017 10:33pm
Hello,

I having been discretionary/manually trading unsuccessfully for 3 years via my own price action trading strategy. I took a year off trading to think about why my trading business is failing.

I come to the conclusion: I need program and back test my own trading ideas/strategy for odds of winning based on historical backtesting data. I need to have confidence that the strategy/system has produced money in the past and worth trading in the future with real cash.

So all capital is on reserve as I proceed to developing a fully automated trading system. I only want to test my own trading ideas from years of watching price action on the charts.

So for the past few months I have been programming a few strategies in NinjaTrader. I do have a programming back ground.

I click the back test button and the results on a few strategies look good. Good meaning, profit made from 4 years of historical back testing.

Now I do not know what to do next. I struggle deeply with trust of the back testing historical data. I honestly do not know how test the system for robustness, strength, and odds of winning.

Questions Please:

1. I need a step by step process on to
1. How To Properly Test a Strategy.
2. How to Decide if the strategy is worth going live or how to evaluate a system performance.
3. How to walk forward
4. When and How to Optimize a Strategy.

Do you know if any book, Ebook, recommended training course on properly designing and testing strategies?

For instance, I may come up with 100 trading strategies to program and test, I need a check list to see which one go live and which one goes in trash.

Thank you kindly. I really appreciate any help I can get.
  • #2
  • Jul 22, 2017 11:30pm Jul 22, 2017 11:30pm
The back testing is a rude simulation of what happened. In back testing you do not have the real spreads and this is very important if your strategies are more day trading. Of course a strategy who produced profit in back testing have a greater odds to produce in real. Test your strategy in demo account for a couple of months and after open a real account. But fully automated strategy I believe it can't work in the long term.
  • #3
  • Jul 23, 2017 12:19am Jul 23, 2017 12:19am
Thank you saklas for your response

Quote
In back testing you do not have the real spreads and this is very important if your strategies are more day trading

My strategies will be for smaller time frame charts or intraday trading, so granularity or price movement inside the bar (3 min to 5 min) will be important. For this type of resolution I plan to use NinjaTrader 8 High Order Fill resolution when back testing intra-day strategies on the small time charts. In addition, I plan manual back test, market replay, and run the back testing simulator for 200 trades and compare the results of each phase and compare accuracy with the market and manual back test report as the governing results to measure against the back testing simulator results.

Thanks
  • #4
  • Jul 23, 2017 1:37am Jul 23, 2017 1:37am
Quoting goodoboy
. I need a step by step process on to 1. How To Properly Test a Strategy. 2. How to Decide if the strategy is worth going live or how to evaluate a system performance. 3. How to walk forward 4. When and How to Optimize a Strategy. Do you know if any book, Ebook, recommended training course on properly designing and testing strategies? For instance, I may come up with 100 trading strategies to program and test, I need a check list to see which one go live and which one goes in trash. Thank you kindly. I really appreciate any help I can get.
Hi goodoboy

Have you considered doing Kevin Davey's strategy workshop? His preference is TradeStation but the processes he uses are applicable to any robust testing platform such as NinjaTrader, TradingBlox, Amibroker etc.

More info here

...and more about his processes here and here.

Have a read of his book first and listen to the podcast. If that appeals to you, then seriously consider the workshop.

Cheers

C
  • #5
  • Jul 23, 2017 2:30am Jul 23, 2017 2:30am
Back test is used to fit your trading system to a designed ideal trading conditions (some price movement scenario in intra-day case). It is the same as in-sample test in statistics analysis or curve fit. Minimum data sets need 12 cases (data) to meet statistics significant condition.

You can save 12 screenshots of 12 intraday charts of 5 minutes with all lines, indicators, etc. Then, you run yours new trading algo on this 12 screenshot cases to see if your algo work or not on MT4. Then, you find problems. Then, you modify your trading system. Then, run it again, etc. ...... long, repeat process. You focus only on one trading system in order to get only profitable algo, which is enough for back test.

Then, you use/open demo account for further forward test on live. It is the same as out-sample test in statistics analysis. Due to your trading system is ideally designed on pre-conditions of a specific trading scenario; so, when such a scenario come/merge out, let your system to run. In such way to test if your trading system work or not? If not work, then, go back to back test, repeat process.

Doing test is simple, but, it cost huge time.
  • #6
  • Jul 23, 2017 3:07am Jul 23, 2017 3:07am
Quoting goodoboy
Now I do not know what to do next. I struggle deeply with trust of the back testing historical data. I honestly do not know how test the system for robustness, strength, and odds of winning. Questions Please: 1. I need a step by step process on to 1. How To Properly Test a Strategy. 2. How to Decide if the strategy is worth going live or how to evaluate a system performance. 3. How to walk forward 4. When and How to Optimize a Strategy. Do you know if any book, Ebook, recommended training course on properly designing and testing strategies? For...
Hi, you think to much men-))
its simple, just forward test, real time, on demo, or real mini account.
If system will survive in forwoard test for period 1 year, then your back test was right, and you can trade this system (with right MM) on real account and good capital.
i haved the same situation, and on bakctest system gived good rezultate, ea work on 7 pairs, test of every pair from 2014 show 100 win ratio, this mean no stop loss hit.. How can i bealive in that? its creazy..
So from april this year, i started forward test in live account, not demo-is not the same.
And what you think? well i haved 4 sl hits on that period until today(trading 6 pairs)..not to much if not think about every 10% loss on each sl, and making in the same time not more thant 30% per month, this is the difference between back and live test. I realise that automated system need my personal asistance in every trade, and i take the last decision to take or not the stop order what was place by ea..
And just after this, i stop to loose money, and i steel test this ea in real time.
Attached Image (click to enlarge)
Click to Enlarge Name: ff.jpg Size: 145 KB
  • #7
  • Jul 23, 2017 6:45am Jul 23, 2017 6:45am
Hmmm ... following the BT: in three (!) years more than a damn Million you got using 1k only - time to think about buying a Company like Amazon, FB or Microsoft ... or what ever you want .....
  • #8
  • Jul 23, 2017 8:55am Jul 23, 2017 8:55am
Quoting goodoboy
Hello, I having been discretionary/manually trading unsuccessfully for 3 years via my own price action trading strategy. I took a year off trading to think about why my trading business is failing. I come to the conclusion: I need program and back test my own trading ideas/strategy for odds of winning based on historical backtesting data.

Of course.

If you want to trade with a positive expectancy, you have to be able to backtest a range of methods reliably and efficiently, just to judge which one(s) to forward test before alighting on one and making it "your method".

Otherwise how would you ever know that what you choose has a genuine edge at all?

No criticism intended - please forgive me for commenting that it's really a shame to trade unsuccessfully for 3 years and then think for another year to learn this, when so many beginners' trading books actually explain it pretty clearly in the first chapter or two.

I'm guessing that you tried to learn to trade through online "information", forums and websites, rather than through reading textbooks?


Quoting goodoboy
1. I need a step by step process on to 1. How To Properly Test a Strategy. 2. How to Decide if the strategy is worth going live or how to evaluate a system performance. 3. How to walk forward 4. When and How to Optimize a Strategy. Do you know if any book, Ebook, recommended training course on properly designing and testing strategies?

Yes - no problem at all: just try Van Tharp's first book "Trade Your Way to Financial Freedom", or Tushar Chande's book "Beyond Technical Analysis" first, and then follow it up with Michael Harris's book "Profitability and Systematic Trading". All will be revealed! And good luck.

(By the way - although I agree completely with your comment that you need to program and backtest your own trading ideas/strategy based on its performance over historical data as a starting-point, I don't accept for a moment that that's a reason for actually wanting to trade using an automated system, and that isn't something I'd touch, myself. But that's another matter altogether, of course.)
  • #9
  • Jul 23, 2017 10:06am Jul 23, 2017 10:06am
Backtesting as much as possible will help give you confidence in your strategy, and also give you an idea of what profit and drawdown you can expect if you were to trade live.

The next step is forward testing. You can use a demo account if you wish, but my preference is to use a real account but trade micro-lots. In my experience, demo accounts do not always accurately reproduce the behaviour of live accounts, and so a low-risk live trade test is better than a demo.

How long you forward-test your strategy before using significant money will depend on the mechanics of your actual strategy, and you can, of course, increase your trade size gradually as you gain more confidence.
  • #10
  • Jul 23, 2017 11:57am Jul 23, 2017 11:57am
Quoting Quickly
I'm guessing that you tried to learn to trade through online "information", forums and websites, rather than through reading textbooks?

Many folk posting in forums have tried to learn online, I guess.

This is partly why there's a great discrepancy between the proportions of winning and losing accounts held at brokerages when their figures are published, and the general forum perception that about 95% are losing and never become profitable. It relates back mostly to how people learnt in the first place.

In this internet age, people often don't quite appreciate how significant it is that traditional textbooks have quality control and peer review before publication, whereas internet information isn't vouched for at all in any similar way, and is self-reinforcing and self-perpetuating so that it's common for misguided approaches to become the majority view more and more, as they're so often repeated.

Anyway you suggested good books to Goodoboy.
  • #11
  • Jul 23, 2017 1:18pm Jul 23, 2017 1:18pm
Quoting Quickly
{quote} Of course. If you want to trade with a positive expectancy, you have to be able to backtest a range of methods reliably and efficiently, just to judge which one(s) to forward test before alighting on one and making it "your method". Otherwise how would you ever know that what you choose has a genuine edge at all? No criticism intended - please forgive me for commenting that it's really a shame to trade unsuccessfully for 3 years and then think for another year to learn this, when so many beginners' trading books actually explain it pretty...

Thank you Quickly,

You are correct, I tried manually trade via my own un-proven trading method that I did not back test and traded with no discipline and consistency. I also lost money follow an experienced trader on the website trader room. All reason for non-sucess, is my fault entirely. No excuses. Luckily, I only lost about $7000 over the few years doing it this way. But all is not bad, I did study some chart time.

I am still learning Quickly, I do not know it all and never will or want to. All I know is, I want to pursue the automated trading route as it fits my personality "now" and for myself my odds of success is higher on this approach. My trading business cost is now $0 per month. And it will stay this way until I find a proven trading strategy(s) I can automated trade with reliability. And I know what to do if the system starts to fail and so forth.

That's why I am taking my time to learn and read. Learn my platform, learn the quality of my back testing historical data, learn to trust the back test results. And have fun while doing it.
  • #12
  • Jul 23, 2017 1:22pm Jul 23, 2017 1:22pm
Quoting KarlKraus
{quote} It relates back mostly to how people learnt in the first place. In this internet age, people often don't quite appreciate how significant it is that traditional textbooks have quality control and peer review before publication, whereas internet information isn't vouched for at all in any similar way, and is self-reinforcing and self-perpetuating so that it's common for misguided approaches to become the majority view more and more, as they're so often repeated.
Thanks KarlKraus,

You are 100% correct. I can definitely state that my first experience in trading set the tone.
  • #13
  • Jul 23, 2017 1:38pm Jul 23, 2017 1:38pm
Quoting Copernicus
{quote} Hi goodoboy Have you considered doing Kevin Davey's strategy workshop? His preference is TradeStation but the processes he uses are applicable to any robust testing platform such as NinjaTrader, TradingBlox, Amibroker etc. More info here ...and more about his processes here and here....
Thank you Copernicus for the reference. I will look into Kevin's book (and few other books) as starting point as well as listen to podcasts as driving to and from work.
  • #14
  • Jul 24, 2017 2:23am Jul 24, 2017 2:23am
Quoting goodoboy
Do you know if any book, Ebook, recommended training course on properly designing and testing strategies?
You'll find plenty of good material in Dr Daniel Fernandez' blog.
  • #15
  • Jul 24, 2017 2:34am Jul 24, 2017 2:34am
Quoting shum
{quote} Hi, you think to much men-)) its simple, just forward test, real time, on demo, or real mini account. If system will survive in forwoard test for period 1 year, then your back test was right, and you can trade this system (with right MM) on real account and good capital. i haved the same situation, and on bakctest system gived good rezultate, ea work on 7 pairs, test of every pair from 2014 show 100 win ratio, this mean no stop loss hit.. How can i bealive in that? its creazy.. So from april this year, i started forward test in live account,...
Good MM, even with 60% win trade with 1:3 R:R still give you good return
Risk comes from not knowing what you're doing.
  • #16
  • Last Post: Jul 24, 2017 9:30am Jul 24, 2017 9:30am
Quoting hanover
{quote} You'll find plenty of good material in Dr Daniel Fernandez' blog.
Thank you hanover.
0 traders viewing now