Relationships worth to be analyzed in order book data

agerigk · May 16th, 2010, 10:59 am

Hi guys,I managed to get access to a very nice level 2 order book database for my master's thesis and now I am thinking about which relationships in order book data are worth looking at.Apart from some obvious things (like a correlation between the bid-ask spreads of the 10 quotes) what kind of relationships are interesting to look at?It would be great to get some input.If it is of interest: I am analyzing the data with a D-vine pair copula construction.

Alan · May 16th, 2010, 5:06 pm

Well, the obvious topic is to see if you can detect symptoms of the so-called flash-crash cascadethat we saw on May 6. When heavy selling is triggering a bunch of stop-loss orders, then how wouldyou detect this in the order book?. Were there cascade symptoms prior to May 6? How doesthe book change when the NYSE goes into slow mode? A zillion obvious questions ...

sidmaestro · May 24th, 2010, 5:34 am

An interesting study that I have been working on (but left unfinished) is to track algo'scan you please suggest some references to this in the form of books or papers or any web resources ? this sounds extremely interesting.also since you have experience on this can you also suggest what kind of algos would be easiest to study in this manner. thanks in advance

sidmaestro · May 24th, 2010, 7:18 am

dear outrun sir,although i know this is too much to ask from a menial noob quant, i would love to do anything associated with this and would be extremely happy if you could let me contribute in any small riskless way. i am a master's degree holder in mathematics and now more than just a beginner in quant finance.thanking yousincerely