macrocephalopod

can someone smarter than me explain how the big pod shop model (Millennium, Citadel etc.) generally works pretty well? The whole idea of yanking the portfolio after some non-outlier drawdown in a relatively short period of time shouldn't really add any value imo

Few things to consider, 1. good PMs will manage to the dd limit so it’s not simply a case of getting cut if the strategy has a bad period 2. you can obviously use discretion if a mean reverting trade takes a loss but clearly has higher E(pnl) from here,

3. If you think you have some selection edge in hiring, then cutting the pods that lose money and keeping the ones that make money can actually be a pretty good way to refine that edge!

4. you are less concerned with losing a few million $ on a large number of pods, what worries you is missing a pod that can generate hundreds of millions per year (so as well as aggressively cutting losers, you aggressively scale winners to capacity)

5. Depending on how smart/centralised you are you can try to juice your returns by hedging common risk (eg equity factors) and applying leverage, afaik Citadel does this aggressively, Millennium bit so much

6. You are taking netting risk because you need to pay the winners but can’t claw back from the losers. This requires you to keep drawdowns small, and there are some other ways to mitigate it for example by passing on larger execution and financing fees to the pods

than what you are seeing (eg pod A wants to be long and pod B short, you can charge both of them 25bps in financing and clip the spread). Creates a buffer that partially pays for your netting risk.

MLP also has really good set ups to trade a broad array of products cheaply and has internal crossing so potential offset for t costs. Pod models like citadel are far less appealing bc you’re an input into citadel securities and are executed “low urgency” by default