Wednesday, July 05, 2017

Price Momentum & MFI

Price Momentum & MFI

When I posted my MFI tracking results last weekend, I showed (as usual) that the subset of dividend stocks have done better than the overall 50 stocks over $100m screen.

A reader who follows me on Twitter commented that he had heard that MFI stocks that had gone up more in the past 6 months, were more likely to sustain that momentum.  This is kind of the reverse of "dogs of MFI" approach I had been tracking.

It was an interesting enough idea that I decided to at least roll up my sleeves and see if it was doable.  It was, to some extent - with the problem that I went further and further back in time, it became more manual and more frustrating.

So I looked at a subset.  I looked at MFI portfolios i started in 2013-2016.  For each stock, I then tried to find the price 6 month prior.  I will be the first to admit I was not super duper thorough (with stock splits or special dividends or tickers going away or changing causing problems).

But I think I did enough to get a decent sample.  Here is the table and then I will explain it:


Decile Pct Change Cumulative Average
1 -6.4% 5.8%
2 2.5% 7.1%
3 3.3% 7.7%
4 8.4% 8.3%
5 5.5% 8.3%
6 9.0% 8.9%
7 13.0% 8.8%
8 11.3% 7.4%
9 3.8% 5.5%
10 7.2% 7.2%


So I did it on a decile basis instead of absolute.  There could be problems with this that I can easily think of.  So after scrubbing the data, I had about 1700 stock years.  I split these into deciles with a "1" being the 10% of stocks that had seen their price drop the most in past 6 months (like an ESI).  A 10 are the 10% of stocks that had seen their price rise the most in 6 months prior to the portfolio inception. 

Pct Change is just the change for each decile.  So overall, this group was up a mediocre 5.8%.  But the 10th decile was up 7.2% and the 1st decile was the only loser, down 6.4%.  Even 2nd and 3rd deciles were subpar.

Cumulative Pct Change first assumes you just bought the 10% that went up most (7.2%) and then assumes you buy the 20% best, 30% best etc.  So it maxes out at "6" (8.9% and corresponds to 50% of best moving stocks in past 6 months) which means a good strategy might be to 

  1. take the 50 stocks great than 100m from the screen,
  2. Find their price 6 months ago.
  3. Calculate pct change from 6 months ago to today.
  4. Then only buy stocks in the top 1/2.
That would be 8.9% versus 5.8% per the table above.

I know I have not done this as rigorously as I should.

  1. I should test bear market and bull market years.
  2. I should cross tabulate dividends and market caps.
  3. I should clean up data better.
  4. I should run the backtest exactly as I described the strategy.

The clearest point to me is the "don't catch a falling knife" adage.  The cut off here for the first decile is stocks that have dropped 34% or more... so my advice would be in MFI, if a stock has dropped 34% or more in the past 6 months - give it a good long thought before buying it.


Not that it is a great sample, but let us now look at the MFI stocks I have bought since 2012 and see if this filter would have helped:


Symbol Purchase Date Purchase End Gain 6 Month Prior Price Change
VNCE 11/15/15 $3.99  $4.10  3% $18.15  -78.0%
DEPO 02/01/16 $15.34  $18.09  18% $32.79  -53.2%
SAVE 11/15/15 $34.48  $53.39  55% $66.59  -48.2%
NSR 05/06/14 $25.82   30.69  19% $48.25  -46.5%
SALE 08/17/15 $9.34  $11.75  26% $15.47  -39.6%
LCI 11/15/15 $34.01  $24.25  -29% $53.04  -35.9%
HSII 05/06/16 $18.51  $21.65  20% $28.75  -35.6%
AAPL 02/01/13 $450.60  $500.66  13% $665.24  -32.3%
UTHR 05/06/16 $107.99  $122.84  14% $157.16  -31.3%
QCOM 02/01/16 $45.18  $53.43  23% $63.02  -28.3%
RHI 05/06/16 $38.51  $46.91  24% $52.60  -26.8%
CBI 08/15/14 $58.25  $51.70  -11% $79.42  -26.7%
BRCD 05/06/16 $7.97  $12.57  60% $10.83  -26.4%
WU 11/15/12 $12.77  $16.40  32% $16.84  -24.2%
VEC 02/01/16 $19.64  $22.54  15% $25.41  -22.7%
GNC 08/16/16 $19.79  $8.58  -55% $25.56  -22.6%
PPC 11/15/15 $18.98  $19.46  17% $24.33  -22.0%
AWRE 11/15/14 $4.40  $3.07  -30% $5.61  -21.6%
VIAB 02/02/15 $63.91  $44.90  -27% $80.82  -20.9%
TDC 02/01/16 $24.02  $29.36  22% $30.00  -19.9%
ATHM 11/15/16 $22.61  $45.67  102% $27.45  -17.6%
RCII 05/06/14 $28.78  $31.77  14% $34.35  -16.2%
NUS 02/01/13 $41.94  $116.59  181% $49.48  -15.2%
KFY 08/15/12 $13.98  $18.95  36% $16.46  -15.1%
CBI 05/06/16 $37.82  $29.30  -22% $44.39  -14.8%
FLR 11/15/14 $64.80  $47.46  -25% $75.90  -14.6%
VDSI 08/17/15 $18.66  $17.45  -7% $21.76  -14.2%
MPAA 11/15/16 $26.55  $27.26  3% $30.75  -13.7%
KORS 05/06/15 $62.24  $50.87  -18% $72.06  -13.6%
CF 08/15/13 $187.26  $247.43  35% $216.65  -13.6%
CF 05/06/13 $183.97  $243.15  34% $212.50  -13.4%
NSU 08/15/12 $3.62  $3.33  -5% $4.16  -13.1%
VIAB 11/15/14 $72.29  $50.69  -28% $83.13  -13.0%
HSII 08/17/15 $20.14  $19.19  -3% $23.03  -12.5%
PERI 02/01/16 $2.45  $1.86  -24% $2.77  -11.6%
QCOM 02/02/15 $62.73  $45.34  -25% $70.85  -11.5%
CA 05/06/14 $29.54  $31.41  10% $32.85  -10.1%
AGX 08/15/13 $15.60  $36.71  140% $17.29  -9.8%
GILD 08/16/16 $79.33  $70.86  -8% $87.25  -9.1%
TGNA 08/17/15 $25.30  $21.37  -13% $27.61  -8.4%
VEC 11/15/15 $23.60  $22.25  -6% $25.45  -7.3%
CBI 05/06/15 $50.38  $38.23  -24% $54.19  -7.0%
IPCM 11/15/14 $38.45  $80.03  108% $41.33  -7.0%
AMGN 11/15/16 $145.14  $174.26  21% $155.41  -6.6%
RDA 05/06/13 $9.74  $16.79  73% $10.20  -4.6%
ITRN 08/15/14 $21.32  $25.63  24% $22.07  -3.4%
AVG 11/15/13 $17.22  $19.21  12% $17.77  -3.1%
BRCD 08/15/14 $9.34  $10.09  10% $9.62  -3.0%
KLIC 08/15/13 $11.29  $14.26  26% $11.48  -1.7%
MSFT 08/15/12 $30.19  $31.92  9% $30.56  -1.2%
UIS 02/02/15 $22.03  $10.87  -51% $22.29  -1.2%
AAPL 11/15/12 $537.62  $527.28  0% $542.44  -0.9%
ATVI 02/01/14 $17.16  $21.53  27% $17.23  -0.4%
KLIC 05/06/13 $11.40  $14.45  27% $11.43  -0.3%
DLB 08/15/12 $34.43  $33.04  8% $34.49  -0.2%
VLO 08/16/16 $54.73  $67.07  26% $54.82  -0.2%
RPXC 02/01/14 $16.13  $12.93  -20% $15.84  1.9%
ABC 11/15/12 $40.21  $69.23  74% $39.35  2.2%
FLR 08/15/13 $66.06  $73.03  11% $64.51  2.4%
CA 02/01/13 $25.07  $32.08  32% $24.33  3.0%
MSGN 08/16/16 $16.68  $22.25  33% $15.99  4.3%
CA 11/15/16 $31.11  $34.48  12% $29.81  4.4%
LPS 11/15/12 $23.89  $34.26  45% $22.84  4.6%
NHTC 02/02/15 $12.44  $19.95  61% $11.83  5.2%
UIS 08/15/12 $19.96  $24.52  23% $18.97  5.2%
CSCO 05/06/14 $22.79  $28.89  30% $21.54  5.8%
CSCO 08/17/15 $28.95  $31.20  11% $27.12  6.7%
VIAB 11/15/12 $49.12  $80.94  67% $45.87  7.1%
KLAC 02/01/14 $61.35  $61.24  30% $57.07  7.5%
CSCO 08/15/14 $24.40  $30.95  30% $22.56  8.2%
VEC 02/02/15 $28.00  $19.76  -29% $25.50  9.8%
SYNA 11/15/13 $48.01  $61.90  29% $43.71  9.8%
GA 11/15/13 $9.04  $11.90  34% $8.19  10.3%
RHI 11/15/16 $43.23  $48.76  14% $38.99  10.9%
PSDV 05/06/15 $4.12  $2.87  -30% $3.67  12.4%
AVG 05/06/14 $19.57  $23.27  19% $17.40  12.5%
RPXC 11/15/13 $17.04  $13.91  -18% $15.08  13.0%
NVDA 05/06/13 $13.85  $18.43  35% $12.10  14.5%
ATVI 11/15/13 $17.71  $20.10  15% $14.90  18.9%
TSRA 05/06/15 $38.94  $30.46  -20% $31.92  22.0%
RPXC 08/15/13 $15.92  $15.09  -5% $12.94  23.0%
CYOU 05/06/13 $29.24  $28.02  -4% $23.75  23.1%
SNDK 02/01/14 $70.20  $75.78  9% $55.94  25.5%
WNR 02/01/14 $39.80  $37.71  2% $31.04  28.2%
AGX 08/15/14 $36.71  $40.69  13% $27.94  31.4%
GA 02/01/13 $6.10  $11.01  91% $4.48  36.1%
CSCO 08/16/16 $31.20  $31.11  1% $22.65  37.7%
STRZA 02/01/13 $16.06  $27.98  74% $10.83  48.3%
DEPO 05/06/15 $23.50  $15.51  -34% $15.00  56.7%
NHTC 11/15/14 $12.19  $51.46  323% $6.60  84.7%

So these cks I bought between aug 2012 and Nov 2016.  I have listed the stock price 6 months prior to my purchase.  And then I sorted in ascending order.

So overall, these 90 stock years average 20% gain (close to my 19.4% IRR).  A second statistic is Median of 14.1%.

If I apply the simple rule of get rid of anything that dropped 34% or more in prior six months I get average gain of 20.7% and median of 13.7%.  Pretty marginal. 

If I apply broader rule (which I cannot actually do, but can approximate) the average gain jumps to 25.0% and the median goes to 16.8%.

That is a little more interesting.

Summary

A lot of meandering here.  But the data does seem to suggest that you are better off taking the top 1/2 subset of the screen where the stocks have gone up the most in the prior 6 months.

I will start to track this a bit more rigorously and may even start some real $ this way.

No comments: