Average MFE/MAE snippet

Viewing 12 posts - 1 through 12 (of 12 total)

Author

Posts
04/07/2021 at 6:10 PM #166439 quote

zilliq
Participant

Master

Hi Guys,

Does someone of you have an average MAE/MFE snippet as indicated on the backtests results ?

Thanks and have a nice day

04/08/2021 at 8:16 AM #166456 quote

nonetheless
Participant

Master

no idea how you would code that, but you can always drag the results into Excel if you just want to know the averages.

Or do you want to use that data as part of your strategy?

04/08/2021 at 8:19 AM #166457 quote

Nicolas
Keymaster

Master

Some examples about MAE/MFE calculation:

MAE MFE code/indicator

MAE and MFE calculator.

MAE maximisation

nonetheless thanked this post

04/08/2021 at 8:31 AM #166463 quote

zilliq
Participant

Master

Thanks @Nicolas

But

Link1 : Doesn’t work

Link2 : It’s an indicator not in proorder

Link3 : I will try

@nonetheless

Yes I want to try a filter based on

Have a nice day Guys

04/08/2021 at 8:37 AM #166464 quote

zilliq
Participant

Master

The link3 with your code @Nicolas seems not to work

Any ideas ?

2021-04-08_09h35_54.jpg

04/08/2021 at 8:58 AM #166467 quote

Nicolas
Keymaster

Master

It is a calculation in points, not in currency, I do not understand the comparison?

04/08/2021 at 9:31 AM #166475 quote

zilliq
Participant

Master

It’s in EUR/USD so 1 pip=10 USD easier to calculate and verify

The problem seems to be when PRT calculate MFE

As you see on the picture

The abs perf (In this strategy) is 300 USD (Stop loss) and as you see on the strategyprofit curve its never higher of 300 USD (on this trade)

But in the backtest result we have a MFE of 336 USD ???

2021-04-08_10h21_20.jpg

04/08/2021 at 9:39 AM #166480 quote

zilliq
Participant

Master

@Nicolas

To try use a strategy on EUR/USD, put a stop/target at 300 USD

Add

graph (((high-positionprice)*pointvalue)*countofposition)/pipsize

You will have a “band” between -300/+300 but a MFE higher ??

Don’t understand why, or it’s a bug

04/08/2021 at 9:57 AM #166483 quote
Nicolas
Keymaster

Master
So
```
 (((high-positionprice)*pointvalue)*countofposition)/pipsize
```
was the calculation you were looking for to compute the MFE for a long order, right?

For the 336USD MFE, please share the code so I can replicate and find out why, thanks.
04/08/2021 at 11:41 AM #166516 quote

zilliq
Participant

Master

I will send you, but first, very often on a backtest MFE is a Zero

A bug I see very often

2021-04-08_12h40_05.jpg

04/08/2021 at 12:01 PM #166519 quote
zilliq
Participant

Master
Try with this simple code for example:
```
defparam cumulateorders=false
 
//dummy strategy
if rsi[14] crosses over 50 then
buy 100000 contracts at market
 
endif
 
set stop ploss 25
SET TARGET PPROFIT tg

 
//increase the array index
if not longonmarket and longonmarket[1] then
index=index+1
endif
 
//compute the MFE
if longonmarket then
mfe = max(mfe,high-tradeprice)
$mfeArray[index] = mfe
endif
 
//at least 10 orders to allow the average calculation
if isset($mfeArray[2]) then
sum=0
for i = 0 to index do
sum=sum+$mfeArray[i]
next
avg = sum/index
endif
 
graph avg
```
On backtest results MFe and MAE are generally Zero

And when you have MFE it’s over the Target …

Seems to have different bugs

2021-04-08_13h00_04.jpg
04/08/2021 at 1:15 PM #166526 quote

zilliq
Participant

Master

Another, problem, with the same code above

With tg=41 for example

If we add all MFE and divide py 8=number of trades, the results (Average MFE is 287.5)

BUT if you see teh backtests results it gives 258.75 …

2021-04-08_14h12_19.jpg 2021-04-08_14h13_31.jpg
Author

Posts