Nice to read something on this site from someone apparently knowing what he is talking about. I just wonder have you ever come across evidence that firm like DE Shaw or Renaissance ever at all engaged in anything remotely definable as high frequency trading? I have not come across such evidence yet and thus refrain from including them in the hft camp.
Sounds good. I'll PM you later today about the research. To my knowledge, and im far from an insider, both firms probably have evolved as the definition of HFT has changed, I would conjecture neither is heavily into automated market making, for example. D.E. Shaw said "it depends on your definition" and claim to not see themselves as an HFT firm, RenTec is notoriously mysterious, but I've seen anecdotal evidence just on the web of them being involved in HFT such as Simons research into high frequency financial signal processing. These are both large multi strategy groups, so you can bet they're going to at least be involved in automated execution though, which is a competitive market in itself. You can also kinda infer from their job postings that they want people with the *nix and C++ competency...that isn't necessarily proof of HFT, but that kind of developer costs money. Again, I have NO inside knowledge on this matter, just telling you what I've gathered...mostly aggregating a few Google results I've seen haha.
Well, in order to have a successful stat arb strategy that reverts back to an established mean on a consistent basis, you are obviously dealing with arbitrage between highly correlated products. And therein lies the rub. It's easy to establish and test these various combinations - the hard part is that it becomes a pure ECN execution speed game in a very crowded arena. While I agree with Mr. Chan's premise on the face of it, in practice this is a super-competitive trading strategy when implemented on a high frequency basis. And my hunch is that the bigger names have access to order flows or a front-running arrangement that smaller firms and independents do not. IMHO, if you tried to do this without the infrastructure and arrangements ($$$) you would be getting hung on legs and getting picked off constantly. Just my opinion.
Bone, I think you meant to say highly cointegrated products. Correlation isn't a requirement for mean reversion.
I had something more on-topic in mind: Beat the Market http://www.amazon.com/Kelly-Capital...0K6_1_2?s=books&ie=UTF8&qid=1414446875&sr=1-2 Math professor does well in the world of trading! Who'da thunk it?
Yes, you are correct and thanks for catching that. You require cointegration to ensure that the divergence will be temporary and the pair will indeed revert back to the mean.
I think Ernie Chan is one of the few "real deal" who would write books. Most of the trading books are just trash and try to earn your money by selling you a trash book.
What leads you to believe that? I have only read one of his books (Automated Trading, I think is what it was called)... And in no way was I impressed with his insights. None of his narratives went anywhere beyond common sense. Not saying he is a sham, but his books certainly aren't enough to suggest he is successful beyond selling books.