alg_CWMR: Confidence Weighted Mean Reversion Algorithm (CWMR)

In ngloe/olpsR: Algorithms and functions for On-line Portfolio Selection

Description

computes the Confidence Weighted Mean Reversion algorithm by Li et al. 2013

Usage

alg_CWMR(returns, phi, epsilon)

Arguments

returns	Matrix of price relatives, i.e. the ratio of the closing (opening) price today and the day before (use function get_price_relatives to calculate from asset prices).
phi	confidence parameter (typical values are 1.28, 1.64, 1.95, 2.57 corresponding to a confidence level of 80%, 90%, 95%, 99%)
epsilon	sensitivity parameter (typically \in [0,1])

Details

Li et al. provide different versions of their CWMR algorithm. The implemented version is deterministic CWMR-Var. Also CWMR requires a normalization step to ensure that the portfolio weights satisfy the assumptions of on-line portfolio selection (no negative weights). It is implemented as a simplex projection according to Duchi et al. 2008 (see also projsplx).

Value

Object of class OLP containing

Alg	Name of the Algorithm
Names	vector of asset names in the portfolio
Weights	calculated portfolio weights as a vector
Wealth	wealth achieved by the portfolio as a vector
mu	exponential growth rate
APY	annual percantage yield (252 trading days)
sigma	standard deviation of exponential growth rate
ASTDV	annualized standard deviation (252 trading days)
MDD	maximum draw down (downside risk)
SR	Sharpe ratio
CR	Calmar ratio

see also print.OLP, plot.OLP

Note

The print method for OLP objects prints only a short summary.

References

Li, B.; Hoi, S. C. H.; Zhao, P. & Gopalkrishnan, V. Confidence Weighted Mean Reversion Strategy for Online Portfolio Selection, ACM, 2013

Duchi, J.; Shalev-Shwartz, S.; Singer, Y. & Chandra, T. Efficient projections onto the l 1-ball for learning in high dimensions, Proceedings of the 25th international conference on Machine learning, 2008

Examples

# load data
data(NYSE)
# select stocks
x = cbind(kinar=NYSE$kinar, iroqu=NYSE$iroqu)

# calculate CWMR algorithm
CWMR = alg_CWMR(x, phi=1.96, epsilon=0.5); CWMR
plot(CWMR)

ngloe/olpsR documentation built on May 23, 2019, 4:42 p.m.