Issue with the In-sample Sharpe
-
Hi there,
I submitted a strategy named Sun73_Q15_Futures_2a #1356718 and I got the following message: “Not eligible send to contest. In-Sample Sharpe must be larger than 1”.
However, when building such strategy in JupyterLab the in-sample Sharpe was above 1.
Indeed, by checking at the graph below I noticed the reported in-sample sharpe is 0.92, but the sharpe ratio considering the same (in-sample) period is 1.96 (which is close to the one I obtained in JupyterLab).
Is this issue related to the ongoing Futures data revision? How can I proceed, since new strategies will probably have the same issue?
Many thanks!
-
@support
Happens for crypto too actually.
And the Sharpe Ratio I get in the Quantiacs online Jupyter notebook is 2.4
But the strategy is rejected when I submit it.
Pls see screenshots below.
-
@sun-73 thanks for the detailed report, we are checking
-
@sheikh thanks, but what is the second plot referring to? If the strategy is rejected, there should be no plot. Is it another strategy?
In this case, we see 0.93 in the global statistics, and 1.18 in the "sliced" statistics. Are you selecting with the slice the exact IS period? The difference is tiny, as bitcoin is very volatile and a few days can induce a huge difference in mean returns.
-
@support
Sorry if I was not clear.
So when I run the strategy in the online Jupyter, I get SR = 2.4
I then submit the strategy to the crypto competition.
It is accepted to be sent for submission. 'Strategy sent successfully'
But after the 'filter' checks have finished running (btw this took more than a day to complete), it is then 'Not eligible to send to contest because the IS Sharpe must be larger than 1'.
See screenshot below for that from my account.
When I click on the chart thumbnail under 'Preview', the chart plot I get is the one I sent you previously above that you are asking about.
Hope that's clearer
It is possible that this might be related to an imported function that I am using which I pip installed in the Jupyter init module but the live system might not be catching. (I emailed the code to you).
-
@support
Ok, I submitted it again and it got through to the crypto competition. Looks like it was able to catch the imported library this time.
I still don't understand why there are two different SR calculations (1.72 and 2.41) at the time of submission though? That's quite puzzling. Is it really catching the imported function?
Anyway, I'm just glad it got into the competition.
Thanks.
-
@sheikh thanks a lot for the details, let us check the issue. The results have to be the same within a reasonable tolerance, we need to see what is happening.
-
-
@support Thank you once again. You guys are the best!
