New research explores whether 130/30 is actually "optimal"
Aug 1st, 2007 | Filed under: 130/30In late June, Citigroup released a comprehensive 36 page quantitative analysis of 1X0/X0 strategies. We found that the report went beyond similar studies and reached a number of useful conclusions about the optimal X in 1X0/X0. The paper’s lead author, London-based Citigroup MD Manolis Liodakis has written this summary exclusively for AllAboutAlpha.com. We have divided it into two postings. The second half will be posted tomorrow.
Special to AllAboutAlpha.com by: Manolis Liodakis, PhD, Managing Director and Head of Global Quantitative Equity Research, Citigroup Investment Research
Fund managers are always looking for ways to improve their performance. The success and growth of the hedge fund industry has highlighted the potential benefits from shorting stocks and investing with leverage.Short extensions are also an increasingly popular method of performance enhancement for fund managers. Their main benefit comes from enabling negative plays to be made on stocks, even when their weight in the benchmark is small. Unfortunately, most managers face some sort of active risk constraint such as a tracking error target. (Adding new long and short positions to an existing portfolio usually increases tracking error unless it is controlled in some way.)
Motivations for the use of short extension strategies
Long-only portfolios are constrained in the negative plays that they can make on small-cap stocks. Managers can make a positive play on a stock by overweighting it in relation to the benchmark. They can also make negative plays by under-weighting a stock they don’t like. With these negative plays however there is an obvious limit to the size of the play they can make; it is not possible to under-weight a stock by more than its weight in the benchmark. The distribution of weights in large and small-cap stocks varies considerably between different types of benchmark.
To continue reading this article please login (at the right) or click here to learn more about accessing our archives.
A very key issue that has a bearing on level of X in 1X0/X0 is how good or bad is the stock picking model/ability/alpha. Two alternate ways to look at this issue are:
(a) The level of optimal X in 1X0/X0 depends highly upon the Information Coefficient (or the Information Ratio) of the active strategy. More specifically, in law man’s terms the level of correlation of the active weights with the future returns. If the manager has very good alpha stock picking models, then a higher X is preferred.
(b) Another alternate way to look at this is using the beta (which is different from the tracking error point of view). Most serious alpha stock picking models end up having negative beta with the general market. If we imagine the 1X0/X0 as made up of 100 & X0/X0, then the negative correlation can help the 1X0/X0 achieve a higher Sharpe, since negative correlation will reduce the combined volatility. The level of beta of the stock picking alpha process is an alternate way of deciding the X in 1X0/X0. Higher the negative correlation, the higher level of X we can take.
[…] While active managers (particularly quant managers) may never actually run out of active short ideas, it is quite possible that at some point their longs may have greater alpha potential than their shorts. In fact, short extension enthusiasts may recall that this was central to the arguments made by Citibank’s Manolis Liodakis in a posting here last summer.   […]