Quote from darkhorse:
Pretty much... though it's not that, say, a directional strategy "won't pass muster" with a portfolio margin system.
Most of the time, a directional strategy will fit within the parameters of a portfolio margin system just fine, because directional traders simply don't need that much excess leverage to make their strategies work.
Every once in a while, though, a directional trader may see an attractive inflection point where the addition of concentrated leverage is a cool option to have.
If you are trading cash, you have this option to "take it up" as much as you need to (within sane reason), whether it be your own (margined) cash or the allocator's cash. If you are relying on portfolio margin to provide a synthetic form of additional leverage, though, such may not be available to you.
You also inadvertently hit on a hidden downside of market neutral type strategies - when embedded leverage is built into the structure of the methodology, the possibility of things going bad can be very bad.
There is a meaningful risk difference between generating 20% returns with a modest or low leveraged strategy, versus generating 20% returns with a highly leveraged strategy, where the assumption of low risk hinges on the relationship between two assets remaining stable. Most of the time that assumption is correct, but when it fails you can wind up like this: http://i.imgur.com/rpzVa.jpg
So the "paradox" here goes into even deeper theoretical waters, re, which strategies are truly "risky" and which are not, e.g., a robust strategy with more drawdowns but a reliable long-term profile and low reliance on stable relationship assumptions, or a low drawdown strategy that relies on levered asset relationships, that looks good 99 months out of 100 but every so often, "oh shit" etc...
Yes, indeed. Although, this assumes that the "oh shit" moment actually happens with these types of leveraged, market neutral strategies. If it doesn't, then I would certainly prefer the strategy with very small drawdowns and steady returns.
I hear what you are saying about the breakdown in the asset relationship hurting you. This would be mitigated substantially by having several hundred, perhaps thousands of these relationships you are betting on, such that any one of them going against you has a minimal impact.
Of course, one could point out that all the spreads could go against you at the same time. However, if all these spreads are uncorrelated, then the chance of this happening is extremely, extremely low. Now, I can already hear the peanut gallery grumbling about LTCM or the quant meltdown and how correlations all go to extremes under certain distressed market conditions and that you will get taken to the cleaners eventually.
I don't discount this scenario, but I will point out that - provided the portfolio is structured correctly - you can minimize your exposure to this scenario greatly. Certainly, you have a much higher probability of getting blown out on a directional trade that hits you during a freak market scenario (such as the flash crash) then you do on 1000 uncorrelated spreads going against you at the same time.