We continue to analyze the algorithms that are coming into the Quantopian Open, looking at the leaderboard, and looking at how past winners are performing. All of that analysis is informing our work on the hedge fund. We now know more about what we want in the hedge fund. The information runs the other way, too. The June rule changes we are announcing today are intended to make the contest leaderboard look more like the hedge fund selection methods. If you haven't read the fund's information page, I strongly suggest you do so now.
While the contest is limited to one winner per month, the hedge fund is not. We're going to need dozens of algorithms in the hedge fund, and anyone making it through the low beta filter is going to be well-positioned for those rewards. That's an important concept to keep in mind when writing your algorithm.
1) To be a top-ranked algorithm, your entry must have low market exposure. Shortly, we will apply a filter based on your entry's beta to SPY. All of the entries that pass the filter will bubble up to the top of the leaderboard, ranked by overall contest score, followed by all of the entries that don't pass the filter, ranked by overall contest score. Your beta to SPY must be between .3 and -.3. Entries with low beta will get a blue badge next to their name on the leaderboard.
Beta calculations can be noisy, so we have chosen a specific beta calculation that smooths out the noise. We are computing your beta to SPY over a trailing 1-year period at the end of each month for a year, and then averaging those results. In practice, that means computing your beta to SPY for the year previous to April 30, 2015, for the year previous to March 31, 2015, for the year previous to February 28, 2015, etc. until we have 12 computations, and then averaging them. That number isn't in your backtest results yet, but we'll give you a way to check it using research for now.
This change obviously is going to make a big difference on the leaderboard. Of the almost 600 entries in the June contest, only about 150 of them have low enough beta to pass the filter. There are some algorithms with high contest scores overall that will be moved from the top 10 all the way down to 150. For some people, this is going to be frustrating. We regret that very much. We want everyone to enjoy the competition, not be frustrated by it. Despite that downside, the contest is getting better every time we iterate on it. That means that for others, the rule change brings a new opportunity to crack the top of the leaderboard and win the chance to manage $100,000.
And, even if you don't win the contest, there are many more opportunities with the hedge fund.
2) The backtest period for this contest is the two years that end on April 30, 2015. Going forward, future contests will have an updated backtest window as well. The last day of the backtest is determined by the last day for submissions for the previous contest.
3) Algorithms that don't have any trades in their paper trading yet will automatically get a consistency score of .5. Our consistency calculation can't return a result in this case, so we need to pick a number. We chose the fairly punitive value of .5 because we don't want these non-trading algorithms to be highly ranked.
4) Algorithms don't have any trades in their paper trading will be ranked last in the beta scoring rank. Again, this is because we don't want these non-trading algorithms to be highly ranked.
*With apologies to Meghan Trainor for the subject line