I was hoping someone could shed some quick insight on this topic as I'm having a little difficulty wrapping my head around it at the moment. Suppose I'm trading a set of pairs and if the spread reaches some threshold then a trade is executed. Currently, I have my program set up such that once a condition is true then a target position is set. There are only 3 states buy spread, sell spread, and do nothing. I'd like to accumulate a position if either the buy or sell condition is true. So the target position I have set isn't really a target position, but rather the maximum position I'd like to take on. So from an automated execution perspective, let's say I've coded up a basic TWAP algo - does it make sense to just create TWAPs with long time horizons then allow the OMS to cancel the order, and only partially fill, if the spread state changes? Now to the much more complicated case, I'm working on a price dependent execution algorithm. Basically the algorithm will buy/sell more aggressively based on spread conditions. This is layered on top of a base algo (e.g. VWAP/TWAP). Does it actually make sense to initiate my scheduler for the algos in the morning despite spread conditions? Then allow the internal logic to determine trade size (if 0 then do nothing?). Thanks
i read your first question 3-4 times, and have no idea what you're asking. while your 2nd question assumes the one you're seeking an answer from knows your goals. irregardless, the answer to either of them is you need to run each of your variants live and pick the one that best expresses the effect (reduced slip, participation, etc) you're trying to solve for. there's no "quick answer" for trying to predict how a proprietary execution algo might work.