According to the Sharpe ratio - page 2

 
Sprut112:
Maybe, but when the best signals have 0.03 it's not very good.

It's just a number, it doesn't guarantee or give anything. A system can steadily profit for 20 years with a Sharpe of 0.03 and fail tomorrow with a Sharpe of more than 1. It also depends on how you calculate it. My robot's sharpe ratio was over 12 when I was counting on my balance. The same robot calculated using equity will have a low coefficient.

 
Maxim Romanov:

It's just a number, it doesn't guarantee or give anything. A system can steadily profit for 20 years with a Sharpe of 0.03 and fail tomorrow with a Sharpe of more than 1. It also depends on how you calculate it. My robot's sharpe ratio was over 12 when I was counting on my balance. The same robot calculated by equity will have a low coefficient.

So I got excited about my 4. (using fxbook), it's more detailed
 
Sprut112:
An excellent indicator >1.
Going through the top ten robot signals, I couldn't find anything even close. The reliability range is 0.03 to 0.3. That's not very impressive.

Not so long ago, there was a discussion about the Sharpe Ratio in the machine learning thread. The Sharpe Ratio that appears in economic reports and reports of financial companies is almost certainly annualized on daily returnees, in metatrader it is so to say "self-baked", calculated from trades and not normalized by the root of the quantity, be careful with it.

 
Sprut112:
So I got excited early with my 4. (using fxbook), it's more detailed.

The methodology for calculating the Sharpe Ratio varies considerably from author to author.

In particular, many sources do not take the average of individual deals, but the average of periods. In addition, the average of deals, days, weeks, months may be taken, and then - the average of these values.

4 is something very high, as a rule, 2 is already considered an infrequent good indicator. I suspect it's just a matter of different scoring methodology.

 
Georgiy Merts:

The methodology for calculating the Sharpe Ratio varies considerably from author to author.

In particular, in many sources the average is taken not by individual deals, but by periods. In addition, the average of deals, days, weeks, months may be taken, and then the average of these values is taken.

4 is something very high, as a rule, 2 is already considered an infrequent good indicator. I suspect that it is just a matter of different calculation methods.

Probably not enough deals my robot made, I think that's the reason
 
Грааль:

Not so long ago in the machine learning thread, there was an argument about the Sharpe Ratio. The Sharpe Ratio that appears in economic reports and financial company reports is almost certainly annualized on daily returnees, in Metatrader it is so to speak "self-baked", calculated from trades and not normalized by the root of the quantity, be careful with it.

Yes, if you count sharpe in companies also by equity) it will tend to 0 at all.
 
I tried to calculate the report but it gives an error 504. Can anyone tell me how to get rid of it?
 

The Sharpe coefficient can be calculated like this:

You can google all the parameters

 
Коэффициент Шарпа (Sharpe Ratio)
Коэффициент Шарпа (Sharpe Ratio)
  • 2017.05.19
  • www.mql5.com
полезен, использовать его нужно много и часто вреден, избавиться от него нужно как можно быстрее бессмыслица, часто вводящая в заблуждение...
 
Sprut112:
A perfect score is >1.
Going through the top ten robot signals, I couldn't find anything even close. The reliability range is 0.03 to 0.3. Not very impressive.

Maybe not on the best ones, but on signals with an honourable rating of the 2nd or 3rd thousand you can find it :) I have a robot trading between 1 and 1.09.