Forex Software — New metric -- "Consistency"

Re: New metric -- "Consistency"

2017-09-23T14:27:12Z

This works really well -- thank you, footon.

Re: New metric -- "Consistency"

2017-09-23T10:05:18Z

Learning the ropes, updated (optimized) version below.

Re: New metric -- "Consistency"

2017-09-22T19:22:36Z

Ah, footon -- I am in heaven. Thank you!!!

By the way, at first it didn't load. There is a checkbox under Control Panel / Custom Code called 'Load custom account statistics at startup' that I had to check.

This is a super feature I didn't know about. So, now I can try-out new statistics if something interesting comes to mind.

Again -- thanks very much.

Re: New metric -- "Consistency"

2017-09-22T19:07:37Z

Try this, Steve!

Put it in User Files\Code

EDIT:
New version below.

Re: New metric -- "Consistency"

2017-09-22T18:58:46Z

No -- that is not correct. In your example the 'Max Stagnation' would probably be large and since it is in the denominator, then it would probably overcome the value in the numerator. Also, since your strategy was good enough to make it into your portfolio in the first place, then it probably isn't so bad that it executes numerous losing trades in a row. However, I agree, there could be scenarios where this simple ratio might not apply. My intent was to point out that a couple of simple stats that we already have available may additional value when applied as a ratio.

"Max stagnation” is defined as the number of consecutive days where the strategy has not made any profits. This could be due to not trading or to losing trades.

So -- if your strategy encounters a region of data that it is not well trained to survive, then it either will not trade or it will trade and lose. That could result in a larger "Max stagnation" value (and smaller Consistency metric).

It's the *ratio* of 'Execution Orders' / 'Max stagnation' that I am referring to. In my experience it sort of mirrors SQN. However, two strategies with similar SQN can often times be differentiated by having different 'Consistency' factors. I would then choose the one with the higher Consistency factor.

In my earlier post I attached an image -- which helps you see what I mean. If you have a counter example, attaching an image would help me see what you mean.

Browse through some of your strategies in FSB and do a quick 'Consistency' calculation in your head. I'll bet 25 cents that your better strategies have a larger Consistency factor.

Re: New metric -- "Consistency"

2017-09-22T17:02:54Z

Here is a scenario

If I have a breakout EA that does very badly in ranging times. It traded more frequently (i.e more executed trades that were bad trades due to false entries) as compared to another EA that executed lesser trades during ranging to avoid false entries.

If I were to use your calculation whereby Executed Trades/Max Stagnation, then it seem to me that the 1st EA that made lots of "wrong" trades is more consistent then the 2nd EA (which is better in handling ranging time by avoiding multiple error entries)?

New metric -- "Consistency"

2017-09-22T16:42:05Z

I would like to propose a new metric -- actually, it's a ratio of 2 older metrics:

Consistency = "Executed orders" / "Max stagnation"

The larger the ratio, the better -- I'll explain...

A large 'Max stagnation' is bad -- since it indicates either (a) the strategy is not trading, or (b) it is losing. The number of 'Executed Orders' is neither good or bad. However, the ratio of 'Executed orders' / 'Max stagnation' provides a glimpse into a strategy's trading behavior -- and I'll refer to that as 'Consistency'. When comparing two strategies, the one with the larger 'Consistency' metric does a better job of trading.

This metric is easy enough to estimate in your head when viewing each strategy's 'Account Statistics'. If others find it valuable, then perhaps it could be promoted and a 'Consistency' metric added to the list.

I've attached an image to show what I mean. The strategy on the left has decent stats, but the Consistency metric is poor. This is due to the fact it trades poorly when the data is ranging. The strategy on the right has a good Consistency metric -- and you can see it trades well over the 3 types of data -- trend-up, trend-down and ranging.