OPTIMAL NET EXPOSURE


In this section we calculate numerically the optimal net market exposure as a function of the market return, volatility, alpha and the long short correlation.


The following chart shows sensitivity analysis for the optimal net market exposure as a function of the long short correlation and market return given the alpha of 10% and the market volatility of 15% as in the previous example.


Fig 6: Sensitivity Analysis - Optimal net market exposure as a function of the long short correlation and the market return


 


Main findings are:<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />

  • The greater the market return the greater the optimal net market exposure

  • For a zero market return, the optimal net market exposure is zero. For a positive market return, the optimal net exposure is positive and decreasing as correlation increases. For a negative market return, the optimal net exposure is negative and increasing as correlation increases.

CONCLUSION


We derived and analyzed the theoretical result for the information ratio as a function of the net market exposure and other main parameters. We also derived the optimal net exposure as a function of the long short correlation and the market return. Our analysis has the following important implications on managing a long short investment strategy:



  • When the long and short portfolios are highly correlated (e.g. as in a pairs based statistical arbitrage book), any systematic market exposure is likely to lead to detrimental results in terms of the risk return trade-off. A market neutral approach is also going to be preferable in case of a low market return expectation and/or strong long short alpha.

  • As the long short correlation decreases, for example as a consequence of lifting sector neutrality constraints, the existence of some systematic exposure to the market may be rewarded. In particular, the greater the markets return expectation, the greater the net market exposure should be. However a systematic net market exposure also leads to high sensitivity to the market return forecast error. Our overall conclusion is that even in this case, a significant systematic market exposure is not likely to be beneficial unless alpha is weak, in which case a more direct exposure to the market, e.g. via index funds, is likely to be more desirable.

  • The market return delivered within the long short strategy could be enhanced by tactical decisions: for example, by successfully selecting when to be long and short of the market. In that case, the optimal exposure results presented in this paper will provide us with guidance on the overall magnitude of market exposure.

THE APPENDIX


We construct a long short strategy by overlaying a directional strategy over the market neutral strategy. We start from conventional definitions of return and return volatilities for the market neutral and the market separately and then derive the combined (long short) returns and volatilities as well as the information ratio.

Rls = Rmn + kRm


Where:
R ls denotes the long short return
R mn denotes the market neutral return and
R m denotes the market return
k denotes the Net market exposure and k>0 implies net long position, whilst
k<0 implies net short position



We can express the long short strategy information ratio IR=Rls / óls in terms of the Net


market exposure, market neutral as well as market returns, volatilities and the correlation


between the market and the market neutral strategy.



IRls =Rls /
óls = (Rmn +kRm) / (ó2mn  + k2ó2m + 2ñkómnóm)1/2


Where:
ó mn             denotes the market neutral volatility
ó m        denotes the market volatility
ñ          denotes the correlation between the market and the market neutral strategy

It is helpful to decompose the market neutral return and its volatility into the long and the short side returns volatilities and the correlation.


Rmn = Rl - Rs
ó 2 mn = ó12  + ós2 - 2 ñls ól ós


Where:
Rl, Rs        denote the long-side and the short-side return respectively,
ól, ós     denote the long-side and the short-side return volatilities and
ñls         denotes the long-side and the short-side return correlation



So the information ratio becomes


IRls = (Rl - Rs + kRm) / [(1+2ñkóm) ó12+ (1+2ñkóm) ós2-2ñlsólós + ks2ó2m- 4ññls kólósóm]1/2


 


The returns from the long and the short side can be further decomposed as follows:


Rl = Rm + ál
Rs= Rm - ás



Where:


ál , ás        denote long and short alphas (measures of long-side out-performance and short-side under-performance against the market)


So that the information ratio finally becomes


IRls = (ál + ás + kRm) / [(1+2ñkóm) ó12+ (1+2ñkóm) ós2-2ñlsólós + ks2ó 2 m- 4ññls kólósóm]1/2


The effects of all nine parameters under realistic assumptions are summarized in the table below:










































Parameter


Effect on information ratio when increasing


Effect on information ratio when decreasing


ál


Positive


Negative


ás


Positive


Negative


ám


Negative


Positive


ól


Negative


Positive


ós


Negative


Positive


ñls


Positive


Negative


ñ


Negative


Positive


Rm


Positive


Negative


k


Can be either


Can be either


 


The net market exposure can have either positive or negative effect as it features both in the numerator as well as the denominator of the above formula. In order to investigate the impact of taking net long / short position, it is helpful to make some realistic assumptions


Which simplify the above formula:


If ál = ás = á / 2, ól = ós= óm and ñ = ()3 we can condense alphas and volatilities so that the information ratio can be simplified. 


These assumptions are rather realistic and robust in terms of how they affect the information ratio


For additional information, please contact: Melissa Hill,
Sabre Fund Management Ltd, Windsor House, 55 St James's Street, London SW1A 1LA
Tel: +44 (0)20 7316 2801 Fax: +44(0)20 7316 0180
Email: melissa.hill@sabrefund.com Website: www.sabrefund.com


Subscribe to free daily newsletter
latestjobs
VP/SVP Credit Quant- NY- Investment Bank

Sun, 28 Dec 2014 00:00:00 GMT

Quantitative Research | Equity | New York

Sat, 27 Dec 2014 00:00:00 GMT

SVP Model Validation

Sat, 27 Dec 2014 00:00:00 GMT

events
2 weeks 1 day from now - New Orleans
3 weeks 3 days from now - Boston
3 weeks 3 days from now - New York
4 weeks 12 hours from now - New York
specialreports