wscrp: Weighted Successive Constant Rebalanced Portfolio
In logopt: Log optimal (universal) portfolios

Description Usage Arguments Details Value Author(s) References

Calculate current portfolio weights as a weighted average of the best CRP until now and the previous weights

1	wscrp(x, from = 1, by = 1, alpha = 0.99, fast.only = TRUE)

`x`	Time series of relative prices
`from`	Start index, can be different from 1 to avoid the very instable first periods.
`by`	Step in time to recalculate best CRP
`alpha`	The weight between old and new, should be in range [0,1], closer to 1 means more filtering
`fast.only`	Only use the fast but slightly inaccurate version of CRP optimization

At each time index ti = from + i * by, calculate the weights for the best CRP up to but not including ti, call that crp_i, calculate the weigthed sum of (1-alpha) * crp_i + alpha * w_i-1 = w_i, use w_i as portfolio weights until the next optimization point. Return the time series of the corresponding portfolio wealth. w_0 is the uniform weight vector.

Return the time series of portfolio wealths.

Marc Delvaux

Gaivoronski, A and Stella, F (2000): Nonstationary Optimization Approach for Finding Universal Portfolios. Published in: Annals of Operations Research , Vol. 100, (2000): pp. 165-188.

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