Maoning Wang 2016-05-30

From cslt Wiki
Jump to: navigation, search

last week:

1. understanding Kelly Criterion: most online portfolio selection algorithms introduced by LiBin can be interpreted as applications of Kelly Criterion under reletively simple assumptions.

2.understanding basic perception algorithms: but finally I find LiBin only uses the concept of loss function like \vec{x}\dot\vec{b}-\epsilon...

3.test the lib OLPS:

(1)the best algorithm may be "Online Moving Average Reversion" as shown by their data

(2)I encounter problems of function 'fetch' in Matlab!

this week:

1.test OLPS with data selected by ourselves

2.consider why "Online Moving Average Reversion" performs better