I have my opinions on how to create a successful automated trading strategy. My experiences form my opinions.
But I'd like to once again invite you to share your opinions and experiences. I say once again, because back on my blog a few months ago there was a lively discussing on this topic, but now that the forum is in-place this is a much better way to share our ideas and theories on the subject.
So here is my list of initial questions. Please share your thoughts. Some of these questions are generalized. Some are specific. I think all are important.
---
How many strategies have you created?
Of those, how many would you classify as profitable?
Of the ones you classified as profitable, how do you define profitability? How long of a period did you trade it cash before deciding it was profitable?
How many strategies have you created in which you initially thought they were very profitable, but later discovered they were not (ie: curve fitted).
Do you prefer to trade discretionary, mechanical, or automated?
How many trades do you require in a backtest in order to feel the results are not curve fitted? (ie: 500 trades)
How much historical data do you require in order to feel the results are not curve fitted? (ie: 3 months, or 3 years, etc)
Percentage wise, how much of a draw down is acceptable when compared to total net profit? (ie: if net profit is 1,000 and draw down is 100 that would be 10%)
What other criteria do your strategies need to meet in order for you to feel they are not curve fitted?
Do your strategies rely on traditional indicators such as Moving Averages, Stochastics, CCI, Bollinger Bands, etc?
Do your strategies rely on price action such as higher highs, lower lows, double tops/bottoms, pivots, day highs/lows, opening range breakouts, etc?
Do your strategies have pre-defined stops or targets (ie: 8 ticks)? Or do they use a variable dynamic such as ATR or Standard Deviation? Or do you have no pre-defined limit and you exit a trade solely based on something from item #10 or #11 telling you to do so?
---
Thanks for your input!!
Mike
Need help? 1) Stop changing things. No new indicators. No new charts. No new methods. Be consistent with what is in front of you first. 2) Start a journal and post to it every day with the trades you made. It will show your strengths and weaknesses. 3) Set goals for yourself that you can reach every day. Make them about how you trade, and not about how much money you make. 4) Accept responsibility for your actions. Stop looking elsewhere to explain away poor performance. You must look within. 5) Have a question? Create a new thread so the community can help.
If you want to support our community, become an Elite Member.
Well now I thought it would only be fair to answer my own questions:
Hundreds.
Two or three.
I define it as a reasonable number of trades (ie 2-5 per day), a draw down of 10% or less of net profit, a high "common sense" factor (no weird custom-curve fitted values), as long of a backtesting period as I can technically manage, and at least a few weeks of cash trading that resembles the historical backtesting results.
Over a hundred.
Automated.
At minimum 2-5 trades a day, preferably the backtest period will allow for hundreds of trades. At minimum I need at least 200 trades in the backtest to feel it is not curve fitted, but I have had strategies with as high as 600 trades turn out to still be curve fitted.
At minimum 3 months. Since I deal with Zen Fire, and I deal with tick data and NinjaTrader, it usually is not possible to do more than 3 months. NT7 better fix this or I'm leaving Ninja. I would feel much better with at least 36 months of backtesting data.
I like to keep it at 10% or less. I always double the backtest figure when I tell myself what the real-world situation might bring, so the backtest figure needs to be 10% or less.
Common sense. Usually if a value is way outside of what you would consider normal, it means it is curve fitted. I also want to see the executions on a chart and not have the reaction of "ugh" because I would have never taken such trades discretionary.
Yes, I use Bollinger Bands, a couple of Moving Averages here and there, sometimes a CCI or %R. But mainly I use custom stuff I've written.
Not so much. It's not to say I haven't tried incorporating such things, but I could never improve the performance of the strategy by filtering based on this. And I have not yet been able to create an indicator that can "block off" poor-performing price action in general, like what I can see visually when I look at a chart. If I could do this then I am sure it would help, but there is no substitute for the human brain.
Normally I use a pre-defined stop and a pre-defined "high" target. But I would say that the majority of my exits occur based on a condition other than a target-met or a stop hit. I've also tried using things like ATR or StdDev or other indicators for trailing stops or to determine volatility of the market and set stops accordingly.
I really look forward to reading some other responses. In only took a few minutes to reply, so I hope you will take the time!
Mike
Need help? 1) Stop changing things. No new indicators. No new charts. No new methods. Be consistent with what is in front of you first. 2) Start a journal and post to it every day with the trades you made. It will show your strengths and weaknesses. 3) Set goals for yourself that you can reach every day. Make them about how you trade, and not about how much money you make. 4) Accept responsibility for your actions. Stop looking elsewhere to explain away poor performance. You must look within. 5) Have a question? Create a new thread so the community can help.
If you want to support our community, become an Elite Member.
The following 2 users say Thank You to Big Mike for this post:
4. 1 but I added a newsfilter/newscalendar when it is not allowed to trade. So it is constantly in development
5. Automated: Bad for your eye but I live longer. Discretionary is better to train your eye but not for my heart :-)
6.don’t know
7. depends on the TF the strategies is working on. 6 month in average
8. the MAX DD shouldn’t be higher then 10%-20% but the average shouldn’t be higher then 5%. But I haven’t really thought about that yet. When it looks good I take it :-)
9.Common sense
10.+ 11. both
12. habe both. Different exits depends on the marketmove.
My strategies have an internal backtesting feature on MT4 which allows it to learn and change there parameters to not primary get the max profit but the safest profit. The max profit could be a pike in a field of loosers but the safest is in the middle of a field of winners parameters. If the solution is better not to trade then it won’t trade. Even for days until the market has changed to a “safer” place.
But that is on MT4. I will see how far I get with NT
The following user says Thank You to siscop for this post:
> 200 trades, but the number of trades is not the main point imho. 200 trades in very different market conditions (high volat, low volat, trending, choppy, low volumes, ...) is better than 1000 in a calm-slow-trendy market.
depends of the underlying contract/timeframe. 10 years of daily bars for swing strategies on stocks, 6 to 9 months for small TF.
10%
common sense.
no more traditional indicators (99.9% of my automated stuff failed with these)
yes
both
counter trend
Last edited by sam028; September 6th, 2009 at 02:49 PM.
The following user says Thank You to sam028 for this post:
I think that discretionary trading is the only way to help us and can be usefull
500+ trades
at least 3 months
I think that the strategies must be profitable on long term and can have any type of drawdown
make small wins consistently
Some yes, some rely on payed indicators, but automated doesn't work also. the problem are the exits, if the trade goes against you, you need to get out with very small losses and automating that is very difficult
Not rely, but must have that information to decide trade or not trade. It's better to not trade if something is not right.
The stops and targets must not be fixed! The market is always changing, you need to have stops and targets that move. Sometimes the market wants to give you more, sometimes not!
The following user says Thank You to pedroxp for this post:
13. Speaking to your successful strategies, do they trade with the trend, counter trend, or both?
My answer is I always focus on trades with the trend on my strategies. In discretionary I may take a few CT trades but with strategies I always purposely write them to exclude CT trades as much as possible.
Mike
Need help? 1) Stop changing things. No new indicators. No new charts. No new methods. Be consistent with what is in front of you first. 2) Start a journal and post to it every day with the trades you made. It will show your strengths and weaknesses. 3) Set goals for yourself that you can reach every day. Make them about how you trade, and not about how much money you make. 4) Accept responsibility for your actions. Stop looking elsewhere to explain away poor performance. You must look within. 5) Have a question? Create a new thread so the community can help.
If you want to support our community, become an Elite Member.
To create a successful automated trading strategy, one must program it to take advantage of what a computer can do better than a human instead of trying to make the computer trade like a human.
A computer is faster and quicker than a human.
A computer can handle more data than a human.
A computer can do more things simultaneously than a human.
Creating automated trading strategies based on SQUIGGLY LINE indicators is laughable at best. PRICE, VOLUME, TIME are the only parameters that you should be concerned with. For example, RANGE is price over time. Same is true for volatility.
The following user says Thank You to TheRumpledOne for this post:
To create a successful automated trading strategy, one must program it to take advantage of what a computer can do better than a human instead of trying to make the computer trade like a human.
A computer is faster and quicker than a human.
A computer can handle more data than a human.
A computer can do more things simultaneously than a human.
Creating automated trading strategies based on SQUIGGLY LINE indicators is laughable at best. PRICE, VOLUME, TIME are the only parameters that you should be concerned with. For example, RANGE is price over time. Same is true for volatility.
So can you answer 1..13 for us?
Mike
Need help? 1) Stop changing things. No new indicators. No new charts. No new methods. Be consistent with what is in front of you first. 2) Start a journal and post to it every day with the trades you made. It will show your strengths and weaknesses. 3) Set goals for yourself that you can reach every day. Make them about how you trade, and not about how much money you make. 4) Accept responsibility for your actions. Stop looking elsewhere to explain away poor performance. You must look within. 5) Have a question? Create a new thread so the community can help.
If you want to support our community, become an Elite Member.
13. sometimes both, sometimes neither... unfortunately in ninja can't do neither since can't submit both bids/asks at same time... but vers 7 should fix this monkey
Linear Mode
