1 (edited by h.neurath 2015-09-27 19:34:42)

Topic: Comaprison of M5 Strategies

Within the Stats Table you can find a selection of M5 strategies on different markets that have been created and analyzed by using FSB software. Partly the strategies are adaptations to the forex market that have been proofed to be successful in equity markets already. While others are almost completely random developed strategies by using the Generator setting a single precondition only.
I do not want to discuss the strategies in detail form the logic side of view but on the stats resulted.
In addition let me add that I suggest that non of the strategies shown (maybe except of one) will be profitable over a longer period (3-5 years) of time.
Why do I think so? The strategies are more or less simple and even although profitable strategies could be simple and therefore the simplicity should not be a criterion for exclusion, to make a simple idea steady running profitable efforts the need for a more complex program code and makes just the simple ideas more and more elaborate.
Second the strategies are based mainly on indicator logics. But it is common known and accepted that there is no indicator or a combination of indicators that is able to describe the market (the price) correctly. A Holy Grail does not exist.
From the strategy point of view it is always favorable to combine an (or more) indicator condition(s) with a direct price event(s) that will have two advantages: first the strategy entry/exit logic will be closer to the market price because it will be triggered by a certain price condition and second the indicator logic could work as a filter (trend, momentum etc. filters) which could make the entry/exit logic distinct more selective to the appropriate market environment the strategy was designed for.
In that perspective it would be preferable to have more possibilities to add or combine a direct price action or a structural or pattern condition to the strategies. In my opinion the software does underrepresent such conditions. For example a strategy that is build on an price/time chart and a price chart (Renko/Three Line Break) or combining a candle pattern with volume or momentum indicators, triangle pattern, in-time break outs etc. But to discuss the FSB software in general is not intention of the writing.
Let´s have look on the table. The time frame of the strategies is M5 generally. The Net Balance is spanning from 2700-4500 Euro setting a starting capital of 1000 Euro for all strategies.
Net Balance is not really such an important figure as one might imagine because it doesn´t tells what has to be accepted to come to a high value for Net Balance, There are other figures that will give more information on the quality of the value for Net Balance. Annualized profit (%) is a good figure to compare different Strategies. For that USDJPY M5_2 Slop… shows the lowest value around 2600 Euro while EURUSD M5_1 HeikenAsh-… shows the highest value with around 5297 Euro. To give the profit for the quarter is widespread also.  Profit per day is not a real figure with a high probability that the profit will be below or above the figure.

http://www11.pic-upload.de/27.09.15/o3ct1hlgedi.png

Please consider using one of the Strategies the Stats indicate that you should be able to compete at the Robbins World Cup Trading Championship. And even with GBPUSD M5_1… and EURUSD M5_1 HeikenAsh-… you would have been a serious competitor for Dr. Andrea Unger.

Do you really believe that? I hope you do not. Furthermore I have created and seen strategies that showed an annualized profit distinct above 10.000 Euro using a starting capital of 1000 Euro only.
Not only on the M5 time frame (with very limited back testing history) but also for higher time frames. But I guess most of these strategies will fail.
System quality number: I do not know how the figure is calculated exactly. But a value below 2 or even 3 can be regarded as a falsification criterion for the strategy.
Sharp ratio. What does the difference between 0.25 and 0.48 really tells us?
An important figure that tells us something about the selectivity of the Strategy is the max. equity drawdown %. Here I like figures below 20% and around 13% for EURUSD M5_1 HHL-3MAC… is excellent.

Selectivity of the Strategy is also reflected by the win/loss ratio. The more selective a strategy is the higher will become the win/loss ratio. I personally seek for Strategies that show a win/loss ratio above 0.65 and above 0.7 is excellent.
But that is something that has to do with psychological acceptance. There are Strategies that have a low win/loss ratio because of trying to catch a longer trend move and that implicates the acceptance of multiple losing trades.
Normally the higher the win/loss ratio the lower will be the average profit/loss ratio. That could be also seen if you compare USDJPY M5_2 Slop-CCI… and EURUSD M5_1 HeikenAsh-…
A further important figure for me is time in position. A selective Strategy should have a value below 30%. A value below 20% is excellent.
A Strategy that is showing a time in position above 50%  has to suspected usually. Often there is a higher part of the characteristic of a buy and hold strategy flanked by an asymmetric SL and Take profit /Exit Limit. A distinct change of market volatility will collapse such a Strategy.

Re: Comaprison of M5 Strategies

There are other softwares that your results can be loaded into for examination, for example Quant Analyzer.

Myself, I may have confidence in a 5 minute strategy for a week or so before it has to be recalculated. No way for a month.

A daily system I would use perhaps for a month or so.

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

3 (edited by h.neurath 2015-10-03 22:23:48)

Re: Comaprison of M5 Strategies

Hello Blaiserboy,

Just a week for a Strategy on 5min time frame is extremely short.
What is the ordinary testing time frame for your 5min Strategy before you go live?
It depends also on the value for "time in position". I prefer a low value, best below 20%. Because it means that the Strategy is more selective as mentioned above already. Advantage is a short tie up of capital.
Such a Strategy will be in the market just around 10 weeks a year. Therefore a Quarter or even a half year live Testing on a Demoaccount is a minimum testing time frame.

Re: Comaprison of M5 Strategies

I do not often use a 5 minute time frame. I try to stick to 4 hour and daily.

I mentioned a week.. meaning the time between optimizations for a 5 minute EA

At this time I am working on a 2 minute ea and optimizing from Jan 1 2011 to today... just as an experiment on Mt5

I used 5 minute on FSBPro and am seeing if I can make the good things happen on a 2 minute chart.

I do understand all of your theory, it sounds great, I like to see actual performance as the market tomorrow will be different from yesterday and that is why I optimize often.

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....