wilmott.com

For a trader which are the most important quantitative studies to see in a equity time series?? garch?hurst?partial autocorr, autocorr?alpha?TA?

I would have thought state space models like kalman techniques, especially as they cope with irregular data.

kalman, ora EKF or UKF or particle filter?they use Kalman to a sort of moving average to predict one step ahead?hodrick prescott filter is used? and hilbert filter?

Am sure they all have use, it would depend on the problem your working on.You had any experience yourself ?

I'm working in a function for a trader and he want to see and control the more significative quantitative indicator, I have experience with filtering and econometrics models, but I don't know what is the best practice. I want to create a function with all top quantitative analisys but I have no idea what is the most usefull indicator for prop-trading!any other idea?

I'm not an expert so only the most basic advice . You need to compile as much information on the characteristics of your data, is it vaguely normal, lognormal, completely, skewed ? , kustosis heavy. Is is highly irregular with regards to arrival of data ?. Any lagged relationships ? (check for cointegrated system) . You can use things like detrended fluctuation analysis (non stationary data) , hurst exponent (stationary), to see if your data is trend following or mean reverting. Once you know your data set it will become alot clearer and possibly a specific filter / method will (hopefully) suggest itself.

tnx a lot!!!

any other idea?