1/10/2015 More Yearly Return Metrics

Sector Model	XLB	-0.92%
Large Portfolio	Date	Return	Days
ESI	8/4/2014	-36.34%	159
EDU	10/27/2014	-6.13%	75
PLT	11/6/2014	-0.98%	65
UVV	12/2/2014	0.54%	39
JOY	12/8/2014	-13.99%	33
MRVL	12/10/2014	8.68%	31
RS	12/11/2014	-6.73%	30
BBRY	12/24/2014	-3.89%	17
MWW	12/29/2014	1.93%	12
JNPR	1/5/2015	2.67%	5
(Since 5/31/2011)
S&P	Annualized	12.29%
Sector Model	Annualized	24.66%
Large Portfolio	Annualized	20.85%
S&P	Total	52.01%
Sector Model	Total	121.77%
Large Portfolio	Total	98.28%
Sector Model	Advantage	12.37%
Large Portfolio	Advantage	8.57%
	Previous	2015
S&P	53.06%	-0.68%
Sector Model	122.60%	-0.37%
Large Portfolio	101.13%	-1.42%

 

Rotation: selling BBRY (corrected on 1/11/2015) PLT; buying COG.

A careful reader noted that I had not included the rate of return for the Full Model last year.

I had a computer crash over the summer that eliminated some of my background data, but this is what I could reconstruct from the previous blog posts.

I’ve taken the annualized rates of return for the Full Model, the Sector Model, and the S&P for the posts on 1/1/2013, 1/5/2014, and 1/1/2015 to reproduce the total return from 5/31/2011 (when the models went live) on those dates.  Next, I reconstructed the return for the year from each date to the next.  Since one of the dates is 1/5/2014 instead of 1/1/2014, the returns will be off by a small amount.  But this is the best I could do from the blog itself:

From 5/31/2011
		Full	Sector	S&P
1/1/2015	Annualized	21.50%	24.97%	12.59%
1/5/2014	Annualized	28.75%	22.65%	12.61%
1/1/2013	Annualized	27.20%	16.22%	3.75%
1/1/2015	Total Return	101.17%	122.57%	53.06%
1/5/2014	Total Return	92.95%	70.07%	36.19%
1/1/2013	Total Return	46.62%	27.01%	6.03%
2014	Year Return	4.26%	30.87%	12.38%
2013	Year Return	31.60%	33.90%	28.45%

 

I’ll dig through my files in the next week or so to see if I can reconstruct 2012 as well.

The Full Model had a bad 2014, tied to an error in my Fundamental Correlation Matrix that I’ve since corrected.  It should catch back up to the Sector Model in 2015.

Also, Steve Cohen’s STAR fund at Folio Institutional returned 29% in 2014, which was an accurate track of the Sector Model.  A good start to our little fund.

To this I’ve added on the Full Model report the metrics that I keep on file.  The additional lines are not usually reported, because they don’t add value on a week by week basis, but I’ll include them in the future at least once a quarter.  On 12/31/2014 I took a snapshot of the total returns, which are used to calculate year to date returns.  Currently those are all down for the year.

The market pundits are screaming about bear markets and such, but I don’t time.  Even if I were 100% convinced of a bear market I would not time.  There is a reason behind that strategy, best elaborated in Taleb’s book Antifragile. The book is a sheer delight and covers an important topic about constructing systems that gain from disorder.

My own model isn’t quite “antifragile” in respect to the dollar, but its defensive nature makes it robust enough to appear antifragile in relation to the S&P.

To show how this works, here are the return rates against SPY for the Sector Model back-tests:

Date	SPY	SPY%	Sector	Sector%	Advantage
12/31/2014	205.54	13.46%	24833.09	36.12%	22.66%
12/31/2013	181.15	32.30%	18243.38	42.36%	10.05%
12/31/2012	136.92	15.99%	12815.31	28.95%	12.95%
12/30/2011	118.04	1.89%	9938.44	6.23%	4.34%
12/31/2010	115.85	15.06%	9355.89	17.54%	2.48%
12/31/2009	100.69	26.35%	7959.91	58.07%	31.72%
12/31/2008	79.69	-36.79%	5035.68	-16.37%	20.43%
12/31/2007	126.08	5.15%	6021.19	21.85%	16.70%
12/29/2006	119.91	15.84%	4941.50	17.82%	1.97%
12/30/2005	103.51	4.83%	4194.23	-0.49%	-5.32%
12/31/2004	98.74	10.70%	4214.90	30.96%	20.26%
12/31/2003	89.20	28.18%	3218.55	36.48%	8.30%
12/31/2002	69.59	-21.58%	2358.30	-7.58%	14.00%
12/31/2001	88.74	-11.76%	2551.61	25.68%	37.44%
12/29/2000	100.57	-9.74%	2030.29	18.47%	28.21%
12/31/1999	111.42	20.39%	1713.72	35.30%	14.91%
12/31/1998	92.55		1266.59

 

These can then be used to create the following forecast metric on expected returns in different market conditions:

SPY%	Sector%	Advantage%
50%	58%	8%
40%	50%	10%
30%	41%	11%
20%	33%	13%
10%	25%	15%
0%	16%	16%
-10%	8%	18%
-20%	-1%	19%
-30%	-9%	21%
-40%	-17%	23%
-50%	-26%	24%

 

The graph of these relationships shows that the worse the market gets, the better the model outperforms:

 

So, if there is a bear, I’ll lose money – but will be so far ahead of the market that I will recover at a much better advantage than I was before the chaos.

I have to keep reminding myself of this during these painful whipsaws we are having lately.  Losing money hurts, and it’s easy to get spooked and cash out at the bottom!

Tim

 

1/4/2015 Two Updates

After updating the constituent stocks in the sector ETFs, the call has changed from XLI to XLB.  I'll trade a favorable gap in the morning.  If there is no favorable gap I'll recalculate in the afternoon.

Also, for the full model, I have a possible rotation tomorrow: selling SWHC; buying JNPR (on a 1% favorable gap or better). The rotation scheduled is a manual trade if there is a 1% favorable gap (if SWHC is up .5% and JNPR down .5%, for instance, there would be a trade).  It doesn’t matter if they are both up or both down, so long as the gap between them is favorable.

1/1/2015 End of the Year review

Sector Model	XLI	-1.43%
Large Portfolio	Date	Return	Days
ESI	8/4/2014	-32.84%	150
EDU	10/27/2014	-8.68%	66
PLT	11/6/2014	1.84%	56
UVV	12/2/2014	10.73%	30
JOY	12/8/2014	-9.46%	24
SWHC	12/9/2014	-1.66%	23
MRVL	12/10/2014	-1.69%	22
RS	12/11/2014	0.10%	21
BBRY	12/24/2014	1.67%	8
MWW	12/29/2014	-0.86%	3
(Since 5/31/2011)
S&P	Annualized	12.59%
Sector Model	Annualized	24.97%
Large Portfolio	Annualized	21.50%

 

A look back and a look ahead.

First, a look ahead.

I have no crystal ball.  I don’t time the market.  I just look for stocks that seem cheaper than other stocks in respect to adaptive metrics of my model.  If I were to start January 1, 2015 with cash I would look to buy:

JDS Uniphase	JDSU	ELECTRNX
Career Education	CECO	EDUC
Cliffs Natural Res.	CLF	STEEL
Arch Coal	ACI	COAL
Monster Worldwide	MWW	ADVERT
LeapFrog Enterpr. 'A'	LF	RECREATE
Southwestern Energy	SWN	GASDIVRS
Kelly Services 'A'	KELYA	HUMAN
Albemarle Corp.	ALB	CHEMDIV
Amer. Vanguard Corp.	AVD	CHEMSPEC

 

A lot of these stocks have had a no good, horrible, very bad year.  I’ve also lost money with CLF in a most spectacular collapse. ACI is being bankrupted by a President who came into office promising to do just that.

What goes down does NOT have to come back up.  Some stocks go to zero.  If they didn’t, then investing would be relatively easy.  Well managed ETFs are less likely to go to zero, of course, and I noted the other day that XLE would be a reasonably sound lifetime hold.

Now for a look back.

My worst trade: ESI lost 45% in the first few hours I held it, and if I had simply bought it one day later I would be up over ten percent instead of down over thirty.  Stocks are dangerous, and if you push the fundamentals too far you can find yourself in a rather painful situation.  ESI still doesn’t have corrected earnings for a good portion of the year, and they are over burdened with some real estate that they need to unload (I sympathize with that position from my own painful real estate adventures).

The Sector Model, on the other hand, ticked along as expected for the year:

	2013	2014
S&P	29.60%	11.39%
Sector	42.36%	36.12%

 

It beat the S&P, and it beat its back-tested benchmark.  That’s about all anyone can ask for.

It would have done better in the early part of December if I had better market data.  The Yahoo feed was corrupted and I rode XLB down for well over a week instead of XLF during that nasty drop.  To make matters worse, December ended with a series of whipsaws that caused me to miss a trade or two – (again) a better real time data feed would have solved that problem.

My New Year’s resolution is to ditch Yahoo and find a better feed.

Although my own account had a good final return, at times it felt like it was more trouble than it was worth:

 

A linear regression for the year would show my account ending very close to the median forecast line, but the standard deviations are uncomfortably large.  Fama would not approve.

To make matters worse, I would have been down for the year if I eliminated the best two week period in late October.  I was extremely pessimistic in early October, and would have cashed out if I have been trading on gut instinct or trying to time.

I’m showing this chart as an object lesson: timing the market is a fool’s errand.  This is NOT a clean year.  It had some really bad trades mixed in (one of which I’m still holding).  The volatility was hair-raising at times.

If you are prone to panic, sometimes you are better off NOT looking.

Another object lesson to take here is the trouble a Hedge Fund would have experienced this year.  Hedge Funds are designed to manage risk.  The volatility of individual stocks created an environment in which a hedged approach would have wiped out any potential gains one could have.  The only type of managed funds that could have systematically made money by the end of the year would be unhedged approaches with a small management fee.

Read that last paragraph again.

Yes, I meant what I wrote.

Hedging COSTS you money in times of volatility because such a fund typically holds individual stocks against an index.  That makes the fund’s volatility to be greater than the index itself by virtue of the fact that it holds less stocks than the total index.  Rather than managing volatility, it locks it into a loss.

The reason is that funds typically mistake “volatility” for “risk.”  The precise opposite is true.  When stocks are cheap, they are at the maximum point of volatility.  Beta is through the roof.  A value investor buys fear and sells complacency.  The control of risk therefore should not be in terms of Beta, but in terms of earnings and price mean reversion.  A stock with an unusually bad earnings year will have a high current P/E because the earnings are down worse than the price, but the regression for both earnings and price indicate that a reversion to the mean would reduce current P/E even as price is increasing, because earnings will be increasing as well.  (Note that I’m talking about current P/E instead of cyclically adjusted P/E).

Conversely, a stock with an unusually good year is likely to revert down to the mean instead of up.

All of this brings us to the conclusion that “value” is found when current fundamentals are worse than normal. And risk is reduced by buying stocks that have already terrified investors away.  In other words, buying at the greatest point of perceived risk is the best way to profit as perceived risk begins to fade.

I had an annoying year. Hedge Funds, on the other hand, should have had an extremely bad year.

Tim