Time Series Analysis' Application in Trading

[suby 10]

I was wondering who here use's time series analysis in their trading and if you could kind of provide some examples with its applications...

 

Suby

[BlueHorseshoe 10]

I was wondering who here use's time series analysis in their trading and if you could kind of provide some examples with its applications...

 

Suby

 

Pretty much all financial price data is time series data (eg closing prices), so virtually any analysis you might perform will loosely fit this description.

 

I'm guessing, however, that you're talking about techniques derived from signal processing and information theory?

 

If so, then the MESA software website has numerous free articles by John Ehlers discussing these kinds of approach. Though I am sceptical of the validity of the approaches he recommends, he is extremely helpful in providing explicit code for things like fisher transforms, kalman filters, fourier analysis, least mean squares etc. The M.R. Bryant articles at the breakout futures website are equally helpful, and a little more diverse. The 'quant forum' is Wilmott, but you'd better be good at maths or it's all double-dutch. Ernie P Chan's site is useful, as is the archived Brett Steenbarger blog for simple ideas on data-mining applications. You might also find yourself dipping into neural nets, markov models, and genetic algorithms - all very exotic-sounding, but ultimately all just tools to do a job.

 

Before even considering any kind of trading approach though, or selecting the time series analysis tools most suitable for it, you will want to uncover vary basic and general tendencies of price behavior in the markets you wish to trade. You'll need to determine things like the serial dependence of the time series, whether probability distribution frequency of the data conforms to a stochastic model and with what degrees of skew, kurtosis etc . . . Unless you either have a solid ground in statistics already, or find it sufficiently interesting to learn as you go, this mighn't be the best approach for you. On the other hand, it can be great if you like to be very quantitative about things. For instance, most experienced traders know that the Euro tends to exhibit more "follow through" or momentum than the ES; being able to describe it and measure it in precise terms of autocorrelation of the time series can be far more useful though . . .

 

If you want any more help with resources then let me know.

 

BlueHorseshoe