MFI Stats

MFI Stats

Good morning everyone.  It is Thanksgiving morning and I have decided to update some of my most mind-blowing MFI stats.  It is easy to get down on MFI as in total it has certainly been hot and cold. But people who read my posts know that MFI stocks that pay a 2.6% dividend or more have greatly outperformed.  I have put together a few tables to dig into that further.  First, here is a table many should be familiar with:

Category	Value
Total	193,541
Total Russell 3K	213,600
New	160,622
Dogs	223,281
Dividend	414,624

This of course shows the overall performance of MFI since 2006 if you started with $100,000 amd spread it equally over 12 monthly tracking portfolios (which take the 50 stocks from the screen greater than $100m market cap).

It does not take a rocket scientist to see in total MFI is trailing just being in the Russell 3000. Picking stocks new to the screen in past year and picking stocks on the screen a year ago that did poorly (dogs) has been a marginal approach.  BUT, picking stocks with a 2.6% yield or higher has been an excellent approach.  This is a subset of the 50 stocks, usually say 15 to 20.  I have been using this approach in my MFI Formula portfolio... to limited success thus far.

A Second Dimension

What happens if we take the table above and add a second dimension?  Does anything else matter?  Not that the dividend result by itself isn't outstanding.  I did look at market cap.  I have simply split the stocks (recall market cap greater than $100m) into three buckets, small, medium and large.

Right now, this is highly simplistic. Under $600m is small. Over 1,500m is big. In between is medium.  Someday, I will either index this (as $600m in 2006 is different than $600m in 2016) or use some sort of ranking (17 smallest of the 50 are small, 17 largest are large and rest are medium).

Either way, the table is fascinating.

Market Cap	0	1	Total
Big	5%	26%	12%
Medium	4%	16%	7%
Small	6%	8%	7%
Total	5%	17%	9%

Look at that.  First, just by picking market caps over $1,500m from the $100m screen improves your results significantly.  Now look at the 0 and 1 columns. These are dividend flag, with a "1" meaning dividend is 2.6% or greater.  Again, in total 17% per year just with this slice (which corresponds to first table I showed).  But if you just go greater than $600m (exclude small),  you move to the 20 to 21% per year range. Whoa.  And that is my new "Formula" approach, which I started two years ago and has been marginally better than R3K. Here is table of my Formula approach results:

Date	Differential	Value	R3K Value
10/1/2014	0.00%	100,000	100,000
11/1/2014	1.63%	102,288	100,658
12/1/2014	0.30%	101,375	101,075
1/1/2015	-0.33%	100,664	100,995
2/1/2015	1.67%	101,281	99,611
3/1/2015	3.62%	106,073	102,454
4/1/2015	5.09%	106,781	101,695
5/1/2015	5.78%	107,859	102,077
6/1/2015	5.08%	108,213	103,134
7/1/2015	4.82%	106,423	101,604
8/1/2015	3.71%	107,112	103,404
9/1/2015	4.38%	101,523	97,139
10/1/2015	6.39%	100,703	94,312
11/1/2015	2.93%	104,835	101,907
12/1/2015	-3.48%	99,443	102,926
1/1/2016	-3.36%	96,830	100,012
2/1/2016	-0.72%	93,797	94,367
3/1/2016	4.07%	98,588	94,367
4/1/2016	0.34%	102,371	101,548
5/1/2016	-1.91%	99,774	101,200
6/1/2016	-0.42%	103,145	103,568
7/1/2016	-0.19%	103,022	103,216
8/1/2016	2.56%	109,419	106,857
9/1/2016	-0.34%	106,704	107,048
10/1/2016	-0.85%	106,377	107,230
11/1/2016	0.42%	105,421	105,003
Current	1.26%	111,327	110,069

So you see a very slight lead, an 11% gain versus 10%. Certainly not what the history since 2006 has been (that works out to a shade over 6% annual IRR, not the 20% from above).

Now it is early days.  Two years is not long enough to test this approach.  I am sticking with it.  One could argue that this approach worked well in a low interest rate environment as larger cap dividend stocks are more attractive when treasury rates are so low.  That is a very valid thought.  Although I have a couple counters.  If that is true,  why has it not worked these past two years? Rates were at their lowest. Also, then it would not have worked well before 2008.  We should look at that.

I do think there is a grain of truth in that thought, but the differential is so great that I think there is still substantial juice in the dividend subset.  

Dividend Stocks By Year

Let us try another slice of the data. Let us just pick stocks that have the dividend flag.  Let us then look at a matrix by year of purchase and s/m/l.  Now when one starts slicing and dicing data, you can end up with crumbs... so some of these data points can be largely influenced by a couple of very good or very poor stocks.

Year	big	medium	small	Total
2006	47%	44%	7%	34%
2007	15%	-10%	-29%	-12%
2008	22%	-8%	-9%	-4%
2009	32%	53%	60%	51%
2010	31%	13%	11%	18%
2011	22%	16%	-8%	9%
2012	53%	30%	25%	41%
2013	41%	19%	5%	27%
2014	12%	-2%	8%	9%
2015	2%	-10%	3%	1%
2016	2%	26%	27%	15%
Grand Total	26%	16%	8%	17%

Now this is a fascinating table.  Every reader of my blog should think about what this shows.

First, look before we had ultra low rates, 2006-2008.  The average Big is 28%. The average Medium is 9%. The average B/M is 18%. The R3K over those purchase years was 5.3%.  Definitely worked before low treasury rates.

Now the frustrating part is that in 2015 and 2016 it stopped working for large.  Like you could draw a line.  That goes a long way to explaining why I started this approach in OCtober 2014 and my results have been just "meh".

So of course the question to answer is are two "meh" years enough to convince an investor that this approach no longer works or at least no longer provides a substantial advantage?

Everyone has to answer that question for themselves.  Here is a listing of the individual stocks in the medium/big category for 2015 purchase year:

Big Stocks From 2015

Date	Stock	Percent Change	Mkt Cap
12/31/15	BBY	60.3%	10,501
11/27/15	BBY	56.2%	10,851
10/30/15	BBY	16.0%	12,059
7/30/15	BBY	9.0%	11,346
11/27/15	BKE	-14.8%	1,589
10/30/15	BKE	-36.3%	1,720
9/25/15	BKE	-31.2%	1,817
8/28/15	BKE	-32.5%	2,067
4/24/15	BKE	-29.9%	2,308
2/27/15	BKE	-32.9%	2,433
1/30/15	BKE	-40.5%	2,457
11/27/15	CA	15.8%	12,247
10/30/15	CA	14.8%	12,035
9/25/15	CA	25.1%	11,929
8/28/15	CA	29.4%	11,885
7/30/15	CA	22.8%	12,827
6/26/15	CA	13.7%	13,017
5/29/15	CA	9.2%	13,261
2/27/15	CA	-6.4%	14,254
1/30/15	CA	-1.8%	13,281
12/31/15	CALM	-12.1%	2,247
11/27/15	CALM	-27.2%	2,714
10/30/15	CALM	-26.9%	2,593
7/30/15	COH	44.4%	8,642
1/30/15	COH	3.6%	10,250
12/31/15	CSCO	13.2%	138,002
11/27/15	CSCO	12.5%	138,840
10/30/15	CSCO	9.7%	146,702
9/25/15	CSCO	26.1%	132,337
8/28/15	CSCO	24.7%	131,638
7/30/15	CSCO	12.3%	144,132
6/26/15	CSCO	5.2%	144,030
5/29/15	CSCO	1.9%	149,276
4/24/15	CSCO	0.9%	147,184
3/27/15	CSCO	6.3%	138,553
2/27/15	CSCO	-7.7%	150,708
1/30/15	CSCO	-6.9%	133,565
12/31/15	GME	-3.5%	2,941
11/27/15	GME	-24.1%	3,795
10/30/15	GME	-45.6%	4,879
9/25/15	GME	-29.4%	4,342
7/30/15	GME	-31.3%	4,944
6/26/15	GME	-35.8%	4,743
5/29/15	GME	-31.0%	4,675
4/24/15	GME	-16.3%	4,384
3/27/15	GME	-18.0%	4,130
2/27/15	GME	-14.3%	4,041
1/30/15	GME	-22.9%	3,853
12/31/15	HPQ	29.3%	21,359
11/27/15	HPQ	22.7%	22,572
9/25/15	HPQ	53.1%	45,043
6/26/15	HRB	-21.4%	8,418
10/30/15	KING	20.3%	4,709
9/25/15	KING	34.3%	4,218
8/28/15	KING	33.3%	4,250
7/30/15	KING	16.4%	4,878
6/26/15	KING	27.8%	4,440
5/29/15	KING	22.1%	4,648
4/24/15	KING	7.4%	5,396
3/27/15	KING	15.9%	5,000
2/27/15	KING	21.2%	5,084
1/30/15	KING	44.4%	4,245
3/27/15	MSFT	35.9%	336,692
2/27/15	MSFT	20.2%	360,359
1/30/15	MSFT	40.1%	332,007
1/30/15	NUS	-20.2%	2,430
6/26/15	OMC	18.5%	17,307
12/31/15	PBI	-24.6%	4,089
11/27/15	PBI	-26.0%	4,170
10/30/15	PBI	-8.8%	4,089
9/25/15	PBI	-6.0%	4,064
8/28/15	PBI	-2.0%	4,030
7/30/15	PBI	-3.1%	4,131
6/26/15	PBI	-13.9%	4,319
5/29/15	PBI	-11.7%	4,392
4/24/15	PBI	-2.6%	4,605
3/27/15	PBI	-7.0%	4,689
2/27/15	PBI	-19.2%	4,657
12/31/15	PPC	-9.9%	5,728
11/27/15	PPC	-5.2%	5,447
10/30/15	PPC	27.3%	4,924
9/25/15	PPC	12.2%	5,461
8/28/15	PPC	22.6%	5,404
7/30/15	PPC	18.4%	5,692
6/26/15	PPC	16.4%	6,196
5/29/15	PPC	8.4%	6,626
4/24/15	PPC	1.4%	6,118
3/27/15	PPC	1.6%	6,141
2/27/15	PPC	-13.6%	7,104
12/31/15	QCOM	40.1%	76,177
11/27/15	QCOM	44.3%	73,975
10/30/15	QCOM	19.5%	93,765
9/25/15	QCOM	33.6%	83,981
8/28/15	QCOM	13.8%	90,656
7/30/15	QCOM	-0.3%	101,639
6/26/15	QCOM	-14.9%	105,606
5/29/15	QCOM	-17.1%	113,787
4/24/15	QCOM	-20.2%	111,436
1/30/15	QCOM	-25.1%	103,309
6/26/15	RTN	42.0%	29,914
4/24/15	TSRA	-21.0%	2,029
12/31/15	VIAB	-7.2%	16,382
11/27/15	VIAB	-24.6%	20,362
10/30/15	VIAB	-21.1%	19,620
9/25/15	VIAB	-9.4%	17,317
8/28/15	VIAB	4.4%	16,278
7/30/15	VIAB	-18.3%	22,381

Eyeballing this, you see GME, BKE, VIAB and CALM really drug things down.

Medium Stocks From 2015

Date	Stock	Percent Change	Mkt Cap
12/31/15	BKE	-9%	1,493
9/25/15	CPLA	16%	625
7/30/15	CPLA	21%	619
6/26/15	CPLA	-1%	680
5/29/15	CPLA	2%	651
12/31/15	IILG	21%	897
11/27/15	IILG	22%	897
10/30/15	IILG	-5%	1,014
11/27/15	IQNT	22%	653
10/30/15	IQNT	-18%	699
9/25/15	IQNT	-26%	757
8/28/15	IQNT	-10%	625
6/26/15	IQNT	10%	634
4/24/15	IQNT	-13%	648
7/30/15	NSU	7%	685
6/26/15	NSU	-20%	793
5/29/15	NSU	-16%	833
4/24/15	NSU	2%	763
12/31/15	OUTR	47%	630
11/27/15	PDLI	-30%	649
10/30/15	PDLI	-23%	749
9/25/15	PDLI	-31%	854
8/28/15	PDLI	-44%	924
7/30/15	PDLI	-32%	942
6/26/15	PDLI	-45%	1,065
5/29/15	PDLI	-44%	1,092
4/24/15	PDLI	-39%	1,131
3/27/15	PDLI	-51%	1,135
2/27/15	PDLI	-53%	1,125
1/30/15	PDLI	-52%	1,217
2/27/15	RGR	37%	974
1/30/15	RGR	49%	784

First note only about 30 items here. So data does get thin. And one can quickly see PDLI is a major culprit.  While the average of all stocks is -9.6%, it becomes a respectable +6.5% without PDLI.  Any anyone who was paying attention knew PDLI was unsustainable as their royalty payments were ending.

So I have convinced myself... the approach has hit a speed bump but I do not see enough evidence to conclude it no longer works. My formula approach does allow me to discard one stock and PDLI has been my pick regularly.

Conclusion

I know this is more data than I generally throw out there. But I think it is important to give some serious consideration on Magic Formula Investing as in total the results have not been anywhere near the book.  I truly believe that is because many stocks hit the screen that do not have sustainable income streams moving forward... if you can weed those out the approach still works. And a mechanical way to weed those out is to pick larger cap stocks that pay dividends from the screen as typically a stock would not pay a dividend if its income was not sustainable and then larger cap are just less volatile to loss of a customer.

Whike my Formula approach takes all of this explicitly into account. you can also see I have a bias in my Select picks.

12 of my 20 positions pay regular dividends.  Just for fun, here are the results from my Select portfolios split between dividends and no dividend.

dividend	22%
non dividend	16%
qtr Div	21%

So you can see I am also seeing a differential (note this excludes my Nov 2016 picks as they are so recent). Now this is just dividend versus no dividend.  Not the 2.6% floor. So a CBI makes the list even though yield under 1%.

Open Positions

2/1/2016	Start	Current	Dividend	Pct Gain	R3K Gain
VEC	$19.64	$22.96	$0.00	16.9%	17.4%
QCOM	$45.18	$68.13	$1.54	54.2%	17.4%
PERI	$2.45	$1.07	$0.00	-56.3%	17.4%
DEPO	$15.34	$20.32	$0.00	32.5%	17.4%
TDC	$24.02	$27.35	$0.00	13.9%	17.4%
Totals				12.2%	17.4%
5/6/2016	Start	Current	Dividend	Pct Gain	R3K Gain
BRCD	$7.97	$12.49	$0.11	58.1%	10.1%
CBI	$37.82	$33.75	$0.00	-10.8%	10.1%
HSII	$18.51	$21.00	$0.26	14.9%	10.1%
RHI	$38.51	$44.80	$0.66	18.0%	10.1%
UTHR	$107.99	$131.39	$0.00	21.7%	10.1%
Totals				20.4%	10.1%
8/15/2016	Start	Current	Dividend	Pct Gain	R3K Gain
CSCO	$31.20	$29.71	$0.00	-4.8%	2.2%
GNC	$19.79	$15.40	$0.19	-21.2%	2.2%
VLO	$54.73	$64.72	$0.60	19.3%	2.2%
MSGN	$16.68	$19.60	$0.00	17.5%	2.2%
GILD	$79.33	$75.30	$0.47	-4.5%	2.2%
Totals				1.3%	2.2%
11/15/2016	Start	Current	Dividend	Pct Gain	R3K Gain
MPAA	$26.55	$25.90	$0.00	-2.4%	2.0%
ATHM	$22.61	$28.06	$0.00	24.1%	2.0%
AMGN	$145.14	$145.03	$0.00	-0.1%	2.0%
RHI	$43.23	$44.80	$0.22	4.1%	2.0%
CA	$31.11	$31.60	$0.00	1.6%	2.0%
Totals				5.5%	2.0%