Applying Normalised Volume

Iâ€™ve never been able to get much out of volume analysis, but then again Iâ€™ve not really given it too much serious attention. Logically, if the size of up days and down days has some predictive ability, the volume behind such moves must glean some important data?

There are a gazillion ways that volume could be examined, but I have chosen to take a look at the predictive ability of volume when used in combination with my favourite weapon of choice, the DVB. Specifically I wanted to look at returns from normalised volume moves.

So here goes:

Normalise Volume (NV)

Take the 10 period percent rank of the volume on the SPY.

Quartiles:

Long signals on the DVB (DVB below 0.5)

NVÂ Â Â Â Â Â Â Â | Â Â Â Â Â Â Â Â Â Â Â Â Average daily return.
0-0.25Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â 0.09%
0.25-0.5Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â 0.05%
0.5-0.75Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â 0.13%
0.75-1.00Â Â Â Â Â Â Â Â Â Â Â Â Â 0.02%

Short signals on the DVB (DVB above 0.5)

NVÂ Â Â Â Â Â Â Â | Â Â Â Â Â Â Â Â Â Â Â Â Average daily return.
0-0.25Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â 0.05%
0.25-0.5Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â 0.02%
0.5-0.75Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â 0.08%
0.75-1.00Â Â Â Â Â Â Â Â Â Â Â Â Â 0.04%

Specifically this is for example greater than or equal to 0.5 and less than or equal to 0.75. So there will be some overlap between categories.

Take away:

Despite much criticism of low volume days, these days in the lowest quartile appear to have at average to above average returns.

The best daily returns come when volume is above average but without being too extreme. Just trading DVB when the volume is in this normalised sweet spot returns some fairly consistent results with the max draw down reduced just 8.30%.