Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Is there a risk management layer planned for Hedge Fund ?

There was (and is still on) lot of debate on the recent change in contest rules.

My thougt is Low Beta / Low drawdown / Positive returns / low corelations
were already good matrix to compare the algos.

Adding more rules in the contest will draw the croud towards more corelated
strategies.

Instead of adding more and more rules to the contest I feel, Q must propose a
proper risk management layer for the fund. Which should filter the order triggers
from algos and increase the confidence of investors.

This layer should even have over ruling right to put an algo in paper
treading mode if there are erratic tradings or high drawdown occuring.

1 response

I can understand why people are worried this is a slippery slope situation where we may get to the point where there are so many rules in place that all strategies are about the same.

At the current point we're at, however, the rules are in place to support the idea that Quantopian is a hedge fund, so asking that people provide strategies that are hedged is not that strange feeling. They want hedged strategies that act on their own, away from the market movements. The previous problem was that people were able to still very easily over-fit strategies by selecting certain strong performers.

This is their version of "risk management," making sure people are not utilizing "all in" strategies with historical bias. Strategies that support hedging at all times are far more likely to perform similarly in the future, or at least be hedged...and less likely to be over fit from bias.

What you're suggesting here is really something that people ought to be building into their strategies as a form of protection and security. It would be interesting to see some sort of framework coming from Quantopian for people to implement this specifically though with various parameters, that might be really cool.

I've been thinking up a simple system to control drawdown, for example. A sort of "stop loss" for the entire strategy. I think the hard part is that all strategies are a bit different and people's needs are all different. I can envision possibly some ways there could be a risk management layer automatically applied to strategies, but I can also see how that might get messy fast.