Universal portfolios. (English) Zbl 0900.90052

Summary: We exhibit an algorithm for portfolio selection that asymptotically outperforms the best stock in the market. Let \({\mathbf x}_i=(x_{i1},x_{i2},\dots,x_{im})^t\) denote the performance of the stock market on day \(i\), where \(x_{ij}\) is the factor by which the \(j\)th stock increases on day \(i\). Let \({\mathbf b}_i=(b_{i1},b_{i2},\dots,b_{im})^t\), \(b_{ij}\geq 0\), \(\sum_j b_{ij}=1\), denote the proportion \(b_{ij}\) of wealth invested in the \(j\)th stock on day \(i\). Then \(S_n=\prod^n_{i=1} {\mathbf b}^t_i {\mathbf x}_i\) is the factor by which wealth is increased in \(n\) trading days. Consider as a goal the wealth \(S^*_n=\max_{\mathbf b} \prod^n_{i=1}{\mathbf b}^t{\mathbf x}_i\) that can be achieved by the best constant rebalanced portfolio chosen after the stock outcomes are revealed. It can be shown that \(S^*_n\) exceeds the best stock, the Dow Jones average, and the value line index at time \(n\). In fact, \(S^*_n\) usually exceeds these quantities by an exponential factor. Let \({\mathbf x}_1, {\mathbf x}_2,\dots,\) be an arbitrary sequence of market vectors. It will be shown that the nonanticipating sequence of portfolios \(\widehat{\mathbf b}_k=\int{\mathbf b}\prod^{k-1}_{i=1} {\mathbf b}^t{\mathbf x}_i d{\mathbf b}/\int\prod^{k-1}_{i=1}{\mathbf b}^t{\mathbf x}_i d{\mathbf b}\) yields wealth \(\widehat{S}_n=\prod^n_{k=1}\widehat{\mathbf b}^t_k {\mathbf x}_k\) such that \((1/n)\ln(S^*_n/\widehat{S}_n)\to 0\), for every bounded sequence \({\mathbf x}_1{\mathbf x}_2,\dots,\) and, under mild conditions, achieves \[ \widehat{S}_n\sim\;\frac{S^*_n(m-1)!(2\pi/n)^{(m-1)/2}}{|J_n|^{1/2}} , \] where \(J_n\) is an \((m-1)\times (m-1)\) sensitivity matrix. Thus this portfolio strategy has the same exponential rate of growth as the apparently unachievable \(S^*_n\).


91G10 Portfolio theory
Full Text: DOI


[1] Algoet P., Ann. Probab 16 pp 876– (1988)
[2] Arrow K., Essays in the Theory of Risk Bearing (1974) · Zbl 0215.58602
[3] DOI: 10.1109/18.21241 · Zbl 0662.90023
[4] Bell R., Math, Oper. Res 5 pp 161– (1980)
[5] Bell R., Management Sci 34 pp 724– (1980)
[6] Blackwell D., Proc. Int. Congress Math pp 336– (1956)
[7] Blackwell D., Pacific J. Math 6 pp 1– (1956) · Zbl 0074.34403
[8] Breiman L., Fourth Berkeley Symp. on Mathematical Statistics and Probability 1 pp 65– (1961)
[9] DOI: 10.1109/TIT.1984.1056869 · Zbl 0541.90007
[10] DOI: 10.1016/0196-8858(86)90029-1 · Zbl 0604.90018
[11] DOI: 10.1109/TIT.1978.1055912 · Zbl 0382.94013
[12] DOI: 10.2307/1426692 · Zbl 0456.90100
[13] Hakansson N., Portfolio Theory, 25 Years After: Essays in Honor of Harry Markowitz 11 pp 169– (1979)
[14] Hannan J. F., Ann. Math. Stat 16 pp 37– (1955)
[15] Kelly J. L., Bell Systems Tech. J pp 917– (1956)
[16] DOI: 10.2307/2975974
[17] DOI: 10.2307/2326680
[18] DOI: 10.1016/0304-405X(74)90009-9 · Zbl 1131.91345
[19] DOI: 10.1086/295078
[20] Robbins H., Proc. Second Berkeley Symp. on Mathematical Statistics and Probability (1951)
[21] DOI: 10.2307/2329779
[22] DOI: 10.2307/1926559
[23] DOI: 10.1016/0378-4266(79)90023-2
[24] Sharpe W. F., Management Sci 9 pp 277– (1963)
[25] Thorp E., Business Econ. Statist. Proc. Amer Stat. Assoc pp 215– (1971)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.