More Magic Formula Statistics

Statistics 101

Nice Sunday morning here in the Nutmeg state.  My New Year's diet is going okay so far (thanks for asking).  I started the NY at 186.2 pounds and weighed in this morning at 184.0.  I did want to eat some comfort food during the dreary NFL playoff games last night, but resisted.

I continue the blog I started yesterday looking at returns by varying variables in my MFI Tracking Database, which essentially has 50 MFI stocks years by month going back to January 2006.  Each month is a pull from the official MFI website of top 50 stocks over $100m in market cap.

Results by Number of Letters in Ticker

I know this one seems like a total farce, but I ran it once before and it was actually a pretty good predictor.

Letters in Ticker	Avg Percent Change	Stock Count
2	15.7%	142
3	11.7%	1,523
4	1.8%	2,434
5	5.1%	18
7	-53.6%	37
8	-66.1%	13
Grand Total	5.2%	4,167

So, I actually find that a bit stunning.  Of course 7 or 8 letters probably means it is traded on pink sheets or OB. Four letters are generally Nasdaq stocks, though that seems to be changing a bit (think FB).  I wondered if the poor performance by 4 letter tickers was China driven, so table below excludes China, Taiwan and South Africa:

Letters in Ticker	Avg Percent Change	Stock Count
2	15.7%	142
3	13.1%	1,485
4	2.4%	2,340
5	5.1%	18
7	-46.2%	27
8	-66.1%	13
Grand Total	6.3%	4,025

Hmm. That isn't it.  Interesting, isn't it? Hard to think it is just "luck" as the spread is large and it is a lot of stocks.  How about if we also filter out small stocks and just go greater than 700m?

 

Letters in Ticker	Avg Percent Change	Stock Count
2	14.0%	122
3	11.4%	1,020
4	0.6%	809
5	10.3%	11
8	-74.2%	11
Grand Total	6.7%	1,973

Do not know about you, but two or three letters are looking better to me.

Buy Stock Purchase Month

I generally start my portfolios near the end of the month (say December 31st).  Here are the results by purchase month to see if there is seasonality (though not sure why there would be as we hold for a year).

Row Labels	Average of Percent Change
1	3.6%
2	11.0%
3	8.1%
4	5.3%
5	7.4%
6	12.3%
7	-2.7%
8	1.5%
9	3.7%
10	9.1%
11	0.8%
12	7.5%
Grand Total	5.2%

Hmm, again more spread than I would expect.  Let us pull out China, Taiwan and South Africa.

Row Labels	Average of Percent Change
1	4.6%
2	12.3%
3	9.5%
4	6.7%
5	8.9%
6	14.3%
7	-1.8%
8	2.1%
9	4.8%
10	10.1%
11	1.5%
12	8.0%
Grand Total	6.3%

Looks like June and February are the best times to start. Sadly, I can not do by dividend yield as I only stared capturing that in January 2010.  If people have an idea how I can backfill (I only flagged greater or smaller that 2.6% before then), let me know.