The question is as mechanical as it gets; I'm not asking any strategy-related questions.
I'm sure it differs from firm to firm, but literally, for a given hedge fund, is it one guy calling a shot, a team of analysts chattering about it, and one guy pushing a button? What are some notable firms that rely on one central decision-maker? What about some that are a team of analysts? And those that have a team of analysts with proprietary algorithms helping them? Pretty much was interested in if anybody had interesting anecdotes about what the trading firms of real hedge funds actually operated like.
Structurally, logistically, what are interesting anecdotes about what happens before some grunt plugs in the numbers and hits BUY/SELL?
Most hedge funds centralise their execution regardless of how many decision makers there are. Would be crazy if one desk was selling whilst another was buying. There are some other efficiencies involved as well.
Same at banks - when I was working on interest rate options desk we had to do our cash trades with the cash desk, swap hedges with the swaps desk etc. Sometimes we did our own futures trading as that was on screens (I'm old enough that swaps weren't on the screen yet), for large trades we sent it to the futures desk. Each desk then pools the net risk from internal trading and goes out to hedge it.
The exception are multistrat 'umbrella' funds with lots of people running their own book with p&l cut, particularly with quicker strategies where you don't want to wait for an internal team to do the trade for you, more like a prop shop than a hedge fund.
Execution is often a mixture of 'grunt', prop algos and outsourcing to banks ('do this trade today your benchmark is VWAP'). And the banks in turn of course will use a mixture of 'grunt's and algos (assuming they don't just match the orders on their desk internally or in their private dark pool). I'd say in equities that around 90% of trades go to algos, one way or the other (directly or via banks). In futures 75%. These are % by # of trades, by volume the human share will be higher - humans still tend to do the larger trades especially in futures. More complex trading is also more likely to be done by 'grunts'.
Clearly smaller funds are more likely to outsource their trades or use 'grunts' rather than build their own algos.
Sometimes it's hard to tell the difference between humans and algos. A 'grunt' trader might be given a large order, say 500 US 10 year futures contracts. They probably aren't going to do that entirely manually. They're more likely to use a simple iceberg algo if the market is holding steady, gradually feeding in clips of 50 lots to the algo which in turn is probably only showing 1 or 2 lots in the market. Then if the market starts to run away from them they might switch to just dumping it manually. Is that a human, or an algo, executing that order?
Bank algos and internal algos have different incentives. Internal algos are always built to provide best execution, outright. Bank algos are there to provide best execution against a benchmark, like VWAP. These aren't always the same.
Hope this helps.
GAT
