How to independently verify any quant-based system without using or disclosing any proprietary information
One of the biggest problems for prop shops, firms and hedge funds is independently verifying the results of quant-based systems and portfolio managers. Firms need to verify to invest but developers and pm’s don’t want to divulge their proprietary algorithms, their alpha, for verification.
Here is a really fast and easy way to independently verify quantitative results (think hours instead of days, weeks or months):
Turn the alpha into a data stream and send it to the interested party to be tested.
Since any quantitative strategy will have a statistically valid algorithm(s) as the core of it’s alpha, that algorithm can be turned into a data stream which can be independently tested for statistical significance without disclosing any proprietary information to the tester.
For example, my out-of-sample results (Sharpe 5.7) are based on an algorithm called Edge. The amount of alpha in Edge can be independently verified like this:
Edge is a statistical algorithm
The end product of the algorithm is a data stream, a time series just like ES, US, etc. The data stream runs from 2006-present, 10 minute bars
To verify Edge, run quant tests against the Edge data stream by verifying the trades in the underlying contract are statistically significant
Steps:
Test without Edge first to establish a baseline: buy when yesterday’s close was lower than the prior day’s close and the market is down today
Exit at 14:00 pm ct today (1st test)
Exit at 14:00 pm ct tomorrow (2nd test)
Now, add Edge. Run the same tests with the additional parameter that Edge > +2 for entry
The second round of tests should be statistically significant with better results than the first if there is any alpha in the Edge data stream
If the results are better and significant, there is verification that the Edge data stream contains a statistically significant amount of alpha
(You can try this yourself, an Intraday Edge data stream is here: <https://www.dropbox.com/s/i3fgdwk1mzehkyx/msFinancials_intraday_ES.txt?dl=0> )
It can be used to test for Overnight, Intraday and 1 thru 5 day alpha against ES
To vet further, testers can run arbitrary tests against an alpha stream – that’s kind of a cool term – they can even move on to more sophisticated tests like Monte Carlo simulations
Conclusion:
Using the steps above, a firm can verify any quant-based system or strategy – often within hours instead of days or months – and developers and portfolio managers can safely submit their work without exposing their proprietary information.
—h
Henry Carstens
503-701-5741
carstens@verticalsolutions.com
www.verticalsolutions.com
PS Lets call the steps above ‘Alpha Stream Verification’ It can also be used to rack and stack submissions to your firm, e.g., Alpha Streams that I can test and verify in hours go to the top of the pile improving both quality and quantity of accepted submissions and throughput for the firm