I was traying to make an EA for triangular arbitrage ... and end up with this one 20 Sharpe Ratio

 

what do you think about this:


I'm doing it for other days .... but think I've hit hard with this one...

I'm now testing with other symbols and random days (its all from a random day).

This result is for 1 h time frame for 1 day

 
While it looks interesting, only 27 trades is a sample too small to draw any conclusion..
 
Rafael Grecco #:
While it looks interesting, only 27 trades is a sample too small to draw any conclusion..

yep ... that's why I'm working on doing more days ...

what number of trades could be a good sample?

 
Extremly high sharp ratios are often appear on a small sample of deals if they occasionally happen to be successfull. In most cases, it's statistically insignificant. And will disappear after increasing number of deals.
 
Kristian Kafarov #:
Extremly high sharp ratios are often appear on a small sample of deals if they occasionally happen to be successfull. In most cases, it's statistically insignificant. And will disappear after increasing number of deals.

yes ... you are correct ... but finding still good results ... at this moment I cant do more than 1 month ...  (results have changed because I changed the EA (a lot) but still very good results ... and without fine tunning).


 
Javier Santiago Gaston De Iriarte Cabrera #:

yep ... that's why I'm working on doing more days ...

what number of trades could be a good sample?

Like a thousand 😆 at least that's what a physics math prof used to say: "A thousand is a solid start in statistics" He was from the department of physical chemistry where they do a lot of quantum mechanics so I guess they know it
 
Tobias Johannes Zimmer #:
Like a thousand 😆 at least that's what a physics math prof used to say: "A thousand is a solid start in statistics" He was from the department of physical chemistry where they do a lot of quantum mechanics so I guess they know it

well if so ... 228 trades are around a quarter of a thousand ... so a quarter of 21 should be around 5 Sharpe Ratio (at a bad scenario with 0 Sharpe for the next 775 trades) ... that's still amazing ... (I just finished and article about this). ...

ok... you are right ... I should do more testing's ... (but it takes a whole day or two of calculations to make those tests ... ) I think its still a good strategy... you can all test as you want when you read this article. I don't need to spend one week testing it to know its a good strategy.

(don't miscount the closing orders because it has a logic for that and changing inputs changes results)

 
Javier Santiago Gaston De Iriarte Cabrera #:

well if so ... 228 trades are around a quarter of a thousand ... so a quarter of 21 should be around 5 Sharpe Ratio (at a bad scenario with 0 Sharpe for the next 775 trades) ... that's still amazing ... (I just finished and article about this). ...

ok... you are right ... I should do more testing's ... (but it takes a whole day or two of calculations to make those tests ... ) I think its still a good strategy... you can all test as you want when you read this article. I don't need to spend one week testing it to know its a good strategy.

(don't miscount the closing orders because it has a logic for that and changing inputs changes results)

Can't help you much with all that but the sentence was so absurd that it stuck with me: "A thousand is a number that statisticians like." I guess it should be taken with a grain of salt like most rules of thumb. Trying it on different symbols would be my best guess.
 
Tobias Johannes Zimmer #:
Can't help you much with all that but the sentence was so absurd that it stuck with me: "A thousand is a number that statisticians like." I guess it should be taken with a grain of salt like most rules of thumb. Trying it on different symbols would be my best guess.

yep... thats what I'm doing now ... I will only try in one more symbol ... and if results are so spectacular ... we have one winner.

I'm pretty forward on this ... because the strategy seems good (uses predictions) ... may be the EA is not good enough ... but its just an article to show what you cand do with this strategy ... (its more the strategy than the bot ... but without a simple bot you can't show the strat...

I'm not selling a complete bot ... its a bot that works ... and gives good results ... but it must be finished ... you can use it ... but carefully (for example ...  I set close all orders ... and not close by tickets).

 
Javier Santiago Gaston De Iriarte Cabrera #:

yep... thats what I'm doing now ... I will only try in one more symbol ... and if results are so spectacular ... we have one winner.

I'm pretty forward on this ... because the strategy seems good (uses predictions) ... may be the EA is not good enough ... but its just an article to show what you cand do with this strategy ... (its more the strategy than the bot ... but without a simple bot you can't show the strat...

I'm not selling a complete bot ... its a bot that works ... and gives good results ... but it must be finished ... you can use it ... but carefully (for example ...  I set close all orders ... and not close by tickets).

I heard if the triangular arbitration concept, seems to be a real problem in sports betting. Didn't think much of it trading-wise but I might look at the article
 
Tobias Johannes Zimmer #:
I heard if the triangular arbitration concept, seems to be a real problem in sports betting. Didn't think much of it trading-wise but I might look at the article

I don't know about sports betting ... but I suppose the problem is the same (spread) ... this really works with zero spreads account ... and in sports, I suppose the difference between the total sum of the odds is the commission of sport bookies (as the spread in brokers).