Back to Community
Quantopian open. Stability Factor. How to fix it.

My numerous warnings that stability factor used in Quantopian open scoring system promote algos witch are doing nothing or consistently loosing still did not get any response from Quantopian staff.
Look here,and here, and here, and here, and here.

If you look at formula you may find out that it is well known coefficient of determination or R-squared of the equity trajectory through time
and does not take into consideration if the equity trending up or down or flat.
My recomendation:
To make stability factor reliable multiply coefficient of determination by sign of anyone of
1 Omega Ratio -1
2 CAGR
3 Sortino Ratio
4 Sharpe Ratio
Then only stability factor will do it's job and may be used as separate ranking factor.

5 responses

Why would you want to introduce collinearity to one of the few independent ranking factors??

Simon ,
It is nice to hear that you are interested in this subject again.
As I know Quantopian open scoring system do not include nether Omega Ratio nether CARG.

If CARG is positive Omega Ratio -1, Sortino Ratio, Sharpe Ratio should be positive ether in this case if coefficient of determination is high then Highest Stability factor will point for good algo (stability of gain).
If CARG is negative Omega Ratio -1, Sortino Ratio, Sharpe Ratio should be negative ether in this case if coefficient of determination is high then Stability factor will be negative and will point for bad algo (stability of loosing).

You're right, I forgot - I literally don't care about this at all. It's meaningless minutiae, like arguing with the owner of a carnival game that his balloons are under-inflated. Just drop it; nobody cares, your time would be better spent improving your aim, and as a bonus, I wouldn't have to read about this so-called 'issue' every four weeks.

To whom, especially from Quantopian staff, the process of choosing mangers for 10B hedge fund is not meaningless minutiae, you are welcome to discuss.

To my surprise I find out that there were not announced changes in Quantopian open stability factor calculations.
Before, till October prize(contest 9 ) stability was calculated as coefficient of determination and was in range 0:1.
Starting February 2016 prize (contest 8) and so on stability factor is in range -1:+1.
**What the changes was made?
To whom I should say thank you?

Stability factor now in contest scoring doing its job right.**

https://drive.google.com/file/d/0B10uueaFeSYvTHJMYkQyeW5QR1U/view