Dear all,I have been recently very interested in statistical arbitrage. But at the same time I have heard about other related terms such as quantitative trading and algorithmic trading, which make me a little bit confused. Do these three terms involve simiar techniques but have different focuses? If I want to learn some basics of statistical arbitrage and quantitative trading by self-taught, could anyone please recommend some good references or websites? Thanks very much in advance.Cheers,

This is a very good question. Anybody cares to answer?

To my mind:statistical arbitrage = models responsible for generating orders which seek to capture alpha based on a real or *perceived* asset mispricing. typically generate a large number of small orders.algorithmic trading = models responsible for executing orders to reach some execution target such as minmal market impact. e.g. VWAP, TWAP, volume participation.quantitative trading = any trading style based on the use of computer models to generate orders, so I would include the typical momentum based models used by CTAs which aren't based on an asset mispricing.I agree it has got a little confusing with each term being used generically.

I agree with StatTrader and it can be rather confusing to those coming into the industry. Most shops that deal in the statarb world keep their info proprietary so there won't be an abundance of info out there for you to study but there is a book by Andrew Pole that may be a decent start. Best of luck!

Stat arb ought to have a statistical/probabilistic aspect to it as well! (As against pure, model-independent, arb.)P

Nice to be specific StatTrader; as "Programmatic", "Stat arb", "model-driven", "black box"; all seem to be generic terms for Algo trading that are commonly used by traders/heads of desk.

A "traditional" definition for algo trading is at: http://www.automatedtrader.net/Algorithmic_Trading.xhtm

Good luck finding publicly available up to date information on Statistical Arbitrage, the details are highly proprietary.The following statement by Prof. Andrew Lo can serve as a general definition: "StatArb refers to highly technical short-term mean-reversion strategies involving large numbers of securities (hundreds to thousands, depending on the amount of risk capital), very short holding periods (measured in days to seconds), and substantial computational, trading, and IT infrastructure".So there are two elements: (1) the basic money making idea is to exploit some kind of short term over-reaction effect (mean-reversion in prices) in a portfolio of a large number of securities, (2) the implementation is thorough computerized trading. Because of the second element StatArb is one example of Algorithmic Trading, but only one example; I could program my computer to buy/sell stocks based on the phases of the Moon for example, that would be algorithmic trading but not StartArb.BTW the Wikipedia entry on Statistical Arbitrage, from which I stole this quote, is not bad in my opinion. (Always keeping in mind that the specifics are proprietary, it does give some useful info).

Dear commonweal & circumfly (excuse me if i'm a little late ), I'm making my master thesis on that argument, maybe a formal definition of stat arb could help you.In their paper "Testing Market Efficiency using Statistical Arbitrage with Applications to Momentum and Value Strategies " Hogan, Jarrow, Teo and Warachka define statistical arbitrage in this way:"[...] statistical arbitrage is a zero initial cost, self-financing trading strategy (x(t) : t ≥0) with cumulative discounted value v(t) [...] statistical arbitrage satisfies four conditions (i) it is a zero initial cost (v(0) = 0)self-financing trading strategy, that in the limit has (ii) positive expected discounted profits, (iii) a probability of a loss converging to zero, and (iv) a time-averaged variance converging to zero if the probability of a loss does not become zero in finite time."It's interesting to see that no one could guarantee that a strategy that today we call stat arb, with the implicit assumption that on the long term the return will be non-negative, tomorrow will not be called a speculation, with the implicit assumption that on the long term the return could be negative . So the term "arbitrage" is misleading.Two classical stat arb trading strategies are Momentum trading and Pair Trading. There are many reference on those arguments on the web.An user of the forum, msperlin, wrote some good papers on "Pair Trading", and a free downloadable m-script.For more details i recommend you to search the willmott forum with those keywords, you will find some very interesting opinions .