Tomorrow we're closing out the February edition of the Quantopian Open. When the market closes, we'll sharpen our pencils, crunch the numbers, and see who won. The leaderboard is full of excitement. In the last few days Grant passed Kyu, and has closed in on Schäfer, with only 2/3 of a point separation. It's still anybody's game - if the market moves on Friday, the whole top 10 might get re-shuffled.
We're on target to start trading the winner's algorithm on Monday with $100,000 of real money. Make sure you check back on Monday - we're going to have a nifty new screen where you can see the winner's prize money tracked minute-to-minute as his algorithm does the work.
On Monday, we're kicking off the March edition of the Quantopian Open. Everyone who has placed an entry so far, and everyone who places their entry before market open (9:30 EST) on Monday, will be competing to manage another $100,000. This edition of the contest will end on March 31, after 22 trading days. The March contest is running with the same rule set and judging criteria as the February one did. One big improvement for March is that soon we will permit the corporate fundamental data in your contest entries.
I encourage everyone to get an algo running in the contest! It costs you nothing, and I assure you that you will learn from watching your algorithm run against live data. Backtesting and paper trading are very different in practice, and you need experience doing both in order to become a good algorithmic trader.
If you don't manage to get an algorithm into the contest by Monday, all is not lost. Everyone who enters an algorithm after Monday's deadline will be entered into the April edition of the contest.
We plan on making a scoring change in the April edition. As several people here in the Quantopian community have noted, it's possible to "curve-fit" your backtest for a good backtest score, and that doesn't work for making a good algorithm in the long run. To prevent that strategy from being effective, and even more importantly to encourage good algorithm writing, we're going to add a new scoring factor. We're going to compare the algorithm's backtest performance to the paper trading performance. The bigger the difference between the two tests, the bigger the penalty. Sometime during March we're going to start publishing that penalty factor and how we're computing it in the daily leaderboard CSV. We look forward to your feedback on the scoring change and will use that feedback to shape the new rule for April.
