Building Mean-Reverting Baskets with Stationary Portfolios
Summary
The document discusses how to extend pairs trading to baskets of more than two securities. Its central idea is to seek portfolio weights that produce a stationary portfolio series, then trade deviations as they revert. It points to general portfolio methods and frames principal component analysis as one possible way to inspect combinations: lower-variance principal portfolios may be more stationary candidates than higher-variance ones.
That screening intuition comes with a trade-off. A low-variance portfolio may have less movement available to trade, which can limit profit opportunities even if it appears more stable. The responses also place basket construction within statistical arbitrage and suggest considering economic or statistical factors, such as industries, size, or fundamentals. The document offers conceptual guidance and references to further reading rather than a specified estimation procedure, empirical results, or a complete trading system. Stationarity alone does not establish profitability, and the text does not address transaction costs, stability out of sample, or risk controls.
Key ideas
- A mean-reverting basket can be sought by finding portfolio weights that create a stationary return or price series.
- Principal component portfolios with lower variance may be candidates for greater stationarity.
- Lower variance can also mean less opportunity to trade deviations profitably.
- Basket approaches fit within statistical arbitrage and can use statistical or economically defined factors.
- The document gives high-level guidance but no full construction algorithm or performance evidence.
Tags
Full text
# How to build a mean reverting basket? # How to build a mean reverting basket? I have been playing with mean reverting pairs, but seems that most of the low hanging fruit (ie pairs) have been squeezed already. I would like to start with mean reverting baskets (>2 securities) in order to find unexplored, juicier strategies. Please can you guide me recommending books, articles, etc., on how to build such mean reversing baskets? ## Answer by John (score 11, accepted) https://quant.stackexchange.com/a/3464 There are multiple approaches that you could consider. The basic idea across all of them is that you want to find a portfolio that is stationary. In the two-asset case, it is well known how to accomplish this. This paper by Marcelo Perlin describes one approach: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=952782 but I am not particularly inclined to use this. Alternately, see Section 4 of this paper by Attilio Meucci: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1404905 which provides a more general framework. One way to think about it is that if you arrange the principal portfolios (through PCA) by variance, then the lowest variance ones would be more stationary than the highest variance ones. However, you have to offset the fact that since they have less variance, there is less opportunity to make profitable trades. ## Answer by Tal Fishman (score 6) https://quant.stackexchange.com/a/3466 Pairs trading is just one type of statistical arbitrage (check out references on wikipedia page). It sounds like you are talking about trading "factors" against each other. Factors could be industries, size, fundamentals, or purely statistical. Start with Ed Thorp's Wilmott articles on statistical arbitrage. Then read Attilio Meucci's Review. An example stat arb strategy is studied by Avellaneda and Lee.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.