Hodrick Prescott

Overview



The hodrick prescott filter is a filter for extracting a signal from noise. It does a fairly good job at something the signal and presenting what looks a like a trend.

Trade Signals



The hodrick filter can be used to try to determine the direction of the current trend in order to extrapolate a future price path.

Computational Issues



The filter works through a process of linear algebra by creating a series of matrices and executing

Time Frames



One issues that can be difficult to deal with when using the Hodrick filter, is that it can be hard to specify a time frame. When using moving averages, one can easily specify a long or short time frame by setting the number of days to use in the moving average. The Hodrick Prescott filter does have an input parameter that dictates the amount of smoothing to apply, however, it is not effective at capturing longer term trends.

One strategy to deal with this issue is to artificially create a longer term framework by averaging the points in the data series. For instance, we could collapse every 10 points into a single point by taking the average of those original points. Then, we can apply the hodrick filter to this averaged series. Even though the filter is still detecting a shorter term trend, each point represents 4 weeks, so the short term trend actually represents a longer time frame. 

let hd = await import('/lib/time-series/filter/hodrick-prescott/v1.0.0/hodrick-prescott.srs.mjs');
let data = [{price:100}, {price:101},{price:102},{price:100},{price:99},{price:100},{price:98},{price:100},{price:102},];

let data2 = $list(data).mapAppend(p=>p.price,{
  hodrick:data=>hd.hodrickPrescott(data, 10)
}).items;
Try it!


Contents