Purple Cow Short Interest Metrics

Seth Godin talks about the need to be remarkable, and even bizarre, to grab the attention of the right people. After all you are unlikely to notice a cow on the road unless it is purple in color. I have talked about short interest before in another post.

Let's change the topic to the color purple.

Internal Struggle 
I'm still not sure what is more important, the days to cover or the percent short interest. In the last post, I compromised and combined the two metrics. The following is my attempt, on a small data sample and with no back testing, to expand on the idea.

If you multiply the percent short interest * days to cover you are actually double counting the first metric.

a)(Volume Short / Float) = percent short interest

b)(Volume Short / Float) * (Float / Average Volume) = Volume Short / Average Volume = Days to Cover

So to emphasis each metric {percent short interest & days to cover} I will attempt to establish the mean, standard deviation and variance of  (i)the candidate's  industry %short interest sample & (ii) candidate's industry days to cover ratio sample.


Percent Short Interest
This short sample is a breakdown of one industry by percent short interest. I generated it in www.finviz.com. If I sort by  percent short interest column you can see that there are three types of  records, with large day to cover ratios. I derived the average, calculated the standard deviation and mapped the percent short interest to a bell curve using standardization.

Click on the Data Table below to see the results. Lets do a comparison to the days to cover; On a bell curve the first 6 lines are above the first standard deviation, with 4 not applying to the usual definition of short interest ratio. "Investopedia indicates a bullish guideline for the days to cover when it is greater than eight days to cover."

Clickable Data Table

Shares Float	Float Short%	Days To Cover	Avg Vol(K)	standardization
37.47	0.1526	5.56	1029.41	1.713946415
27.06	0.1468	30.18	131.66	1.6065934
19.81	0.1413	6.86	408.23	1.504793127
72.17	0.1345	15.01	646.65	1.378930971
37.39	0.1340	4.74	1056.48	1.369676401
81.8	0.1253	4.86	2107.32	1.208646878
60.76	0.0865	8.26	636.29	0.490492225
39.5	0.0565	2.1	1063.13	-0.06478199
4.86	0.0516	19.26	13.02	-0.15547678
9.53	0.0495	9.69	48.62	-0.19434598
9.98	0.0442	10.47	42.1	-0.29244442
18.13	0.0418	2.44	311.36	-0.33686636
39.18	0.0414	4.33	374.9	-0.34427001
48.15	0.0375	5.84	309.1	-0.41645566
7.18	0.0347	25.43	9.8	-0.46828126
289.2	0.0193	1.68	3313.68	-0.75332202
386.71	0.0110	1.3	3280.47	-0.90694789
17.5	0.0091	6.64	24.1	-0.94211525
6.72	0.0064	1.02	42.33	-0.99208993
27.04	0.0025	1.15	57.94	-1.06427558
6.59	0.0010	0.61	10.68	-1.09203929
5.96	0.0009	0.58	8.94	-1.09389021
8.29	0.0004	0.78	4.61	-1.10314478
average	0.06
stDev	0.05402736
variance	0.002918956
skew	0.694462922

Days to Cover (Short Interest Ratio) Data Table
Now let's generate an excel data chart of  days to cover data. Get a list of industry competitors from ValueLine that your equity candidate belongs with. Or use www.barchart.com to get competitors in that industry.  Next, get historic days to cover data from the last 12 months.

The candidate averaged around 4 days to cover over the 12 months, and had a standard deviation of about 1.5 days. In other words, my range of 2 standard deviations gave me {+1 to +7 days}.

The industry standard deviation was different. My sample industry had 11 companies, ranging in all sizes and shapes. The bell curve skewed dramatically to the right and gave a mean of ~6 and standard deviation of 5. This would result in a range of {-4 to +16 } for 2 standard deviations.

There were several months where competitors had the 2 and 3 standard deviations barriers breached on the ratio data table. My initial assessment of the candidate's days to cover was done using the "rule of thumb" proposed above by Investopedia. After including the data of it's competitors, the candidates ratio was near the average days to cover and not really shorted by the bears dramatically!

Conclusion
My conclusion is a simple one. Make sure that both metrics exceed the normal standard deviation before acting on them. When you see that the days to cover ratio exceeds the normal limits of the bell curve check to see if the %short interest is in this higher standard deviations also.

A Purple Cow or in other words, what is remarkable about this?
The 12 month historical days to cover ratio data showed that this particular industry grew more and more bearish as we approached the present environment. Can I conclude that the last 6 weeks downturn have affected the short sellers? That may be stretching the limits of a simple empirical experiment. What should be noted is that the days to cover ratio is a dynamic indicator that changes with the external environment. That what appears to be a high number for one candidate may be normal for that industry, at this current time.