06/07/2014 You Can Never Be Fast Enough

Style Model	Small Value
Sector Model	XLU	1.36%
Large Portfolio	Date	Return	Days
ABX	4/11/2013	-32.96%	422
BX	4/14/2014	13.28%	54
TIVO	4/23/2014	0.91%	45
SHOO	4/28/2014	-5.23%	40
UNF	5/2/2014	6.85%	36
PWR	5/12/2014	3.13%	26
JRN	5/19/2014	3.73%	19
BT	5/22/2014	5.01%	16
PM	5/27/2014	2.03%	11
SR	6/2/2014	2.58%	5
(Since 5/31/2011)
S&P	Annualized	13.07%
Sector Model	Annualized	26.55%
Large Portfolio	Annualized	26.64%

 

Rotation: selling UNF; buying CFI.

The Sector Model continues to recover from its recent weakness:

 

And the Style Model has shifted to Small Value:

	Small Value	Mid Value	Mid Blend	Small Growth	Large Value	Small Blend	Large Growth	Large Blend	Mid Growth
Utilities	1	2	5	9	10	15	25	30	48
Staples	3	6	11	19	20	27	36	40	62
Finance	4	7	13	21	22	29	38	42	63
Healthcare	8	16	26	33	35	45	56	58	74
Materials	12	23	32	43	44	52	60	65	75
Industrial	14	24	34	46	47	55	64	67	76
Technology	17	28	39	49	51	57	66	69	78
Cyclicals	18	31	41	53	54	59	68	72	80
Energy	37	50	61	70	71	73	77	79	81

 

The Sectors are defensive.  The Styles are neutral.

My guess is that money is moving between sectors instead of running for the hills – but that’s just a guess.  My models only show WHERE to invest, not WHETHER to invest.  These are not timing models.

As for timing models…

The Kirkpatrick and Dahlquist book, Technical Analysis, probes each traditional method used to measure price and volume and finds one resounding conclusion over and over again: each one stopped working after the turn of this century.

The work is encyclopedic, fascinating, and hopeless.

In other words, by the time everyone realizes something works, it will stop working.

In some cases a time limit is implied in the observation itself.  For instance:

On a purely price basis, the S&P continues to stay on track, with both long term and short term regression forecasts intersecting at 2000 before the end of the year.

As would any “technical” observation, it works until it doesn’t.  But in this case there is a time limit – the forward regressions terminate at their intersection point in November 2014.  After that the two targets diverge and volatility should increase as imagined inevitability gives way to realistic uncertainty.

But the problem here isn’t just the time limit.  Rising wedges (as in this regression graph) often fail before their intersection point just because that point is too obvious.  The trade itself is crowded.

It is precisely this aspect of “crowded” trades that can affect popular quant based formulas.

The Greenblatt formula is a good case in point.  In my own observations, Greenblatt’s one-year holding period substantially underperforms three-month and three-year holding periods.  The reason?  Everyone who uses it is trying to hold for one year and that trade is too crowded.

Does that mean Greenblatt doesn’t work?  No.  It just means that it doesn’t work as advertised.

NOTHING works as advertised, because your competition has read the same advertisements you have.

I don’t advertise the guts of my own model.  But even so I’m comforted by the fact that in the few cases I’ve tried to describe the model, my audience shook their heads and said it didn’t make any sense.

One hedge fund manager told me, “That’s weird.”  I replied, “It’s weird, but it works.” (I actually have solid theoretical analysis for why my “weird” model works, but I didn’t give that to the gentleman).

I was both disappointed and encouraged.  I was disappointed that I wouldn’t be doing business with him, but I was encouraged that he wouldn’t try to duplicate my own work.

So my weird model works – for now.  How LONG it works into the future depends on how long it takes for other folks to duplicate it, but I’ve built into the model some self-adaptive features that should keep it away from my potential competition for as long as possible.

Another reason that formulas fail is that the market itself changes in character. There is a core of human nature within the market that falls prey to successful strategies, but those strategies compete against each other as well.  Mean reversion investors sell to momentum traders, and both hope to scalp the hapless retail short trades (on the mean reversion end) and the gullible long trades (on the momentum end).  Humans scalp dollars from other humans, and black box algorithms slice off an infinite amount of pennies in between.

Thus, the average retail trader underperforms the market; and the average mutual fund underperforms the market too.

Index investing, therefore, will become more popular over time.

So where does that leave us?

Pretty much the same as always: invest in companies instead of stocks.  A stock is worth the value of its company’s future cash flows.  That’s all stocks have ever been worth.  And that’s all that stocks will ever be worth.  Any other movement is random noise. Sometimes a stock will go up because a company is ruthlessly selling assets and laying off workers, and even hopelessly outdated business models can bounce when such trimming occurs. These things affect future cash flows in the short term, but cannot sustain a company for the long term.  I remember one year that Value Line gave Earthlink a good projection, not because dial up had any hope of survival, but because they were slashing their work force faster than their business was shrinking and the stock might bounce.

It has indeed bounced, many times, on the long road down.

In the end you have to invest into companies that you can hold for years if you need to.  Even with my own trading models I have metrics in place to track stocks AFTER I sell them, in order to know how long to hold them if something pulls me away from the market in the near term.

You have to be able to walk away.

It’s summer.  Go to the beach.  Take a walk.

That’s what HUMANS need to do.  Black box algorithms are reading the news now, trying to anticipate what people are going to do mid-day.

People have no business looking at the market mid-day.  We have little business looking at it mid-week.  If you want to stop getting chewed up by high powered algorithms, less is more.

My grandfather picked stocks that would outperform for a decade.  When he passed away my grandmother left his choices in place – against the urging of her advisors.

My grandmother was right, because my grandfather was right.  For the rest of her life her stocks outperformed the broad market.

When the world thinks short term, think long term.

Even if you trade short term, think long term.

Long term isn’t based on timing or momentum.

Long term is based on value.

Tim