+1 I would add that a lot of high frequency stat arb is reliant on bid/ask spread capture to maintain an edge, with the other leg(s) serving as little more than a liquid dirty hedge. (Spread capture probably trumping the mean reversion aspect on shorter timeframes). Quote a price on anything, so long as the spread is wide, and there's a pseudo-replication to lay off onto. I like to think of it as generalized market making, with balls.
Yep, that's another way to do it and to think about it. Point is, with the more prominent firms nobody wants to take "flat price directional risk" if they can help it - which means some sort of arbitrage or spread or some flavor of relative value strategy ( hedged or unhedged ). The combinations and permutations are are almost limitless - creative souls do really well in this environment. And making markets to capture the bid-ask spread based upon the sizing and price of your analog hedge is definitely a very good strategy.
Do explain why. Wide sounds arbitrary. How is that determined ? 1 or 2 sigmas from a normal spread ? Also, what is pseudo-replication exactly in this context ?
Deterministic arbitrage does not exist....at least not in the book of investment / trading terms: http://www.investopedia.com/search/default.aspx?q=arbitrage#axzz1iKeTln50 stat arbitrage conversion arbitrage currency arbitrage merger arbitrage regulatory arbitrage fixed income arbitrage
Gents, you will not find the most currently exploitable arbitrage strategies in a text book or through an internet search. What correlated and worked a few months ago ( or weeks or days for that matter ) might be completely irrelevant these days. There exists legions of quants who are employed to do nothing else but to ferret out these opportunities. I have never found an exploitable arbitrage edge or idea in a trading book. Seriously.
I don't know if you're referring to me or not, but I was simply pointing out that "statistical arbitrage" and "deterministic arbitrage" are specific terms in finance. There seemed to be a lot of confusion in this thread. I have no interest in discussing edges.... Why would I want to do that?
Agreed .. but there can be a fair bit of "arithmetic" involved in selecting and pricing off an analogue(s).