Forums  > Trading  > Methods for compensating for the computational price of updating an online model for market making or mid / high frequency trading  
     
Page 1 of 1
Display using:  

d00dlejumpquant


Total Posts: 7
Joined: Jul 2018
 
Posted: 2019-09-30 18:40
Hello NP people, this post probably sounds a bit too broad, naive, and and maybe a bit silly--and that'd be because I'm a silly naive high school student who knows a little math that enjoys spending time tweaking trading models without really knowing what I'm doing!

This question can be abstracted to something more general, which is basically what the title of this post is. Say you have some predictive online model for market making or HFT (and to avoid someone pointing this out; no I am not grouping market making and HFT together and saying they're the same thing, I know that is not the case. My objective is just to consider the strategies that are somewhat latency constrained). In my more specific example, say I start with a Gaussian distribution over the "true value" of something I'm market making with. In my example, I have a method for updating the p.d.f. over the asset I'm trading based on a classification of "informed" order flow. It is somewhat based on the Glosten-Milgrom Model if people are familiar with that, however really I am curious about a broader approach people have for dealing with this problem versus just my example.

Am I correct that this is more latency-constrained for the most part than anything else? The first step would be testing how computationally expensive it is to update the distribution obviously, but I'm curious about any preliminary insight anyone can provide. Thanks!

just here for phun
Previous Thread :: Next Thread 
Page 1 of 1