Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
New rules and modifications for the march contest?

Are there going to be on new rules for the march contest?
I might be useful to get the as soon as possible?

Besides a ton of suggestions from the other posts here are some:

The back-testing dates ought to be the same for all participants say if one submit on day 28 and the other on day 22. It should not do 28 -2 years or 22-2 years.

So back-testing could say take 2 random years. (same for all algos)

1 response

There are several edge cases in dividends and spinoffs that are susceptible.
-There is a known issue of 25% or greater special dividends not handled. If you go short a special dividend stock you get a free 25% plus drop. Solution: Ban short positions in those stocks like the current leveraged ETF ban until fixed.
-I don't think spinoffs are handled properly, ex try Agilent's Keysight spinoff for shareholders of record Oct 22, 14. Again you short the stock and get a free pop equal to the spinoff value without getting a short position in the spinoff in your account like you should. Solution: Again ban short positions in spin-off stocks during backtest and paper if possible.
-In general, if you can change your portfolio mid-contest you can manipulate the time between ex date and the date money is actually taken from the account. You can just load up a bunch of short positions that go ex in the weeks before the contest is decided but don't have a paid date until after the contest is decided. In an extreme case you could go through nearly the whole universe of dividend paying stocks that month, shorting them just overnight before the ex date then covering the next day. This wont impact real money performance, because again you put whatever stocks you want back in for the real money segment and of course the paper money dividends won't impact the real money positions laid on after the contest is decided. Solution: None except to have someone look at algorithms. As long as the final algorithm can be different from the contest or backtest algorithm, there is almost no end to these kind of games.