I think it would be helpful to have a demarcation, as it were, between holding periods in research studies.

Arguably (perhaps very arguably), there exist repeatable & tradeable phenomena whose holding periods stretch from microseconds to a period approaching 2 days. I would posit that a graphical representation of 'Forecastability' on the vertical axis and Holding Period' on the abscissa would present as a rapidly declining exponential function of some sort.

One would also put forward that if we then started again from a holding period of thirty days or so then once more the chart would show an upward slant (though not exponential)

The puzzle is the part in the middle between about two days and a month. In terms of 'Forecastability' versus 'Holding Period', we are left with a 'U' - ish shape with an extended flat period in the middle.

One further suggests that the Stock Index Futures, Currencies (major not EM), major commodities and Long End Interest rate futures belong at the shorter end of forecasting reliability whereas physical stocks, short end rates and EM ccy's belong at the other end.

It's kind of like a Forecastability Curve (in the same vein as a yield curve). Just like a yield curve there are liquidity preferences et. al.

Something to ponder.





Speak your mind


Resources & Links