mult.upgrade: Portfolio selection using multiplicative upgrades
In logopt: Log optimal (universal) portfolios

Description Usage Arguments Details Value Author(s) References

Implement the portfolio selction defined by Hemlbold, Schapire, Singer and Warmuth. The algorihtm is extremely simple but the function defined here is not able to exactly reproduced the published results.

1	mult.upgrade(x, eta, method)

`x`	Time series of relative prices
`eta`	Parameter defining the learning rate
`method`	The optimization method, currently only "EG" supported, "exact" is planned, but with low priority given the original paper assertion that the two approaches were very close in performance.

Straightforward implementation of the multiplicative update algorithm. The complexity is linear in the number of stocks and in the number of days making it a very fast algorithm, but with a less favorable theoretical bound than Cover's Universal algorithm.

Return the time series of portfolio weights.

Marc Delvaux

David P. Helmbold, Robert E. Schapire, Yoram Singer and Manfreg K. Warmuth. On-line portfolio selection using multiplicative updates. Mathematical Finance, 8(4): 325-347, October 1998.

logopt documentation built on May 2, 2019, 5:49 p.m.