I believe participating in the contest is not necessary for an allocation. My impression is that the Quantopian team is fairly accessible via email if you wish to solicit an allocation. In addition, my impression is that if you run a backtest with fantastic stats Quantopian has some internal mechanism that will put you on their radar. They may reach out to you six months later if the OOS has held up.

As for the future, my opinion is that Quantopian should transition algorithms away from the current low-level portfolio ordering paradigm to one of generation of higher-level signals, which Quantopian would then evaluate, combine, balance, and maximize behind the scenes. This would give Quantopian precise control over overall risk exposure, turnover, etc. The optimizer was a step in this direction. I think they should take it one further. They could for example, use a higher-frequency signal to time the intraday entry and exit of positions generated by a lower-frequency signal, thus exploiting alpha from both signals without necessarily increasing turnover. Furthermore -- combining opposing signals means effectively getting a zero-slippage, zero-cost fill, and the remaining capital can be maximized towards the remaining, more highly predictive positions. Lots of advantages in going higher-level. I think a lot of algorithm authors are already combining signals in order to generate their trades to great effect. So, the advantages of Quantopian doing it themselves, with more signals. more sophistication, and more granularity, should be immediately apparent. In my mind this is an avenue by which they could start accepting a broader range of strategies, such as Warren's.