[R-SIG-Finance] HF strategy style change detection based on txn data
sunduck at gmail.com
Fri Feb 17 14:51:46 CET 2012
thanks a lot for your help!!
I think it's a very promising idea to combine factor analysis (to be
more precisely latent factor analysis) + regime shifts. The latent
factor (-s) I guess could capture the intrinsic dymanics of a strategy
at any desirable frequency and then be analysed for structural breaks.
I don't think it's somehow possible to apply standard factor analysis
as there are no readily available market factors for high frequency
strategies at any desirable frequency, because all these strategies
are very different in their nature and have very different intraday
statistical properties to be able to be comparable with some factor
As of the trading decisions analysis it's assumed (intrinsically
implied) that all the decisions made inside the strategies are already
optimal because they have all the desired statistical properties
needed to be included in the fund of funds portfolio and are in the
responsibility of the 3rd-party asset managers.
Thanks once time again Brian!
2012/2/17 Brian G. Peterson <brian at braverock.com>:
> vid Ardia has written extensivel
More information about the R-SIG-Finance