5/21/2015 Mean Reversion -- good and bad

Sector Model	XLE	-3.53%
Full Model	Date	Return	Days
JOY	12/8/2014	-18.78%	164
SSYS	3/3/2015	-44.40%	79
PWR	3/9/2015	4.59%	73
BHE	3/31/2015	-1.39%	51
CBI	4/2/2015	20.16%	49
MTZ	4/9/2015	-7.14%	42
NE	5/7/2015	-3.98%	14
DRQ	5/15/2015	-1.12%	6
RES	5/19/2015	-2.49%	2
CRR	5/19/2015	-6.02%	2
(Since 5/31/2011)
S&P	Annualized	12.21%
Sector Model	Annualized	20.87%
Full Model	Annualized	19.03%
S&P	Total	58.03%
Sector Model	Total	112.37%
Full Model	Total	99.81%
Sector Model	Advantage	8.67%
Full Model	Advantage	6.82%
	Previous	2015
S&P	53.06%	3.25%
Sector Model	122.60%	-4.59%
Full Model	101.13%	-0.66%

 

I won’t be able to do an update this weekend, so a quick note on performance statistics, benchmark, and mean reversion.

A full look at the entire 1998-2010 back-test, compared to 2011-2015 live performance shows two major deviations from the long term benchmark:

 

The low deviation from benchmark (shown here as the exponential trend-line) was back in first quarter of 2009.

The high deviation from benchmark was in the first quarter of this year.

The outperformance of last year should create a drag for this year, and in fact the rolling 12 month performance of the model, the market, and benchmark show that the model is far closer to the market than benchmark:

Rolling S&P	13.51%
Rolling Sector	14.57%
Benchmark	24.39%

 

Still outperforming the market, but barely.

What gives?

Mean reversion.

Kahneman writes in his book Thinking, Fast and Slow about a training session he had with the Israeli air force. He pointed out that people responded better to praise than punishment, when an officer contradicted him and explained that when he criticized bad performance from pilots he got better performance, and when he praised good performance he got worse performance.

Kahneman also notes the so called Sports Illustrated curse where an athlete who makes the cover one year will have a bad year the following one.

It’s a perverse experience to find that we get bad results from those we praise and good results from those we punish.

The reality is that we all underestimate the randomness of life: half of the time we are above our average and half of the time we are below our average. The key isn’t to focus on how we compare to our average, but instead how our average compares to other averages.

Take a look at our performance chart above. For the full time period we are above our average trend-line, which means that we run the risk of under-performing in the short term.

The question isn’t what to do in the short term.

The question is whether our average is better than the market’s average.

 

Another perverse effect of human response to mean reversion is that we tend to get excited about the market just when it is ready to tank, and we bail out of the market just when it is ready to rally. This again has to do with our mental blindness to mean reversion.

Hedge Funds are especially susceptible to this effect, and respond by closing their funds to new clients when the market gets over heated. Hedge funds take a percentage of the profit, but also take a percentage of the loss. It is possible for a hedge fund to out-perform the market every single year and still go bankrupt if they get over loaded with clients at tops and under loaded at bottoms. Their profit will be from a smaller client base and their loss from a larger client base: making money for their clients who ride it out, but bankrupting the hedge fund owners.

Currently the Sector Model is above benchmark by 0.22 standard deviations. If it were 2 standard deviations above benchmark, I’d get concerned.  If it were 2 standard deviations below, I’d get excited. Now it’s… eh… slightly high, but nothing to panic about.

 

After the last two trades the Full Model owns 4 oilfield stocks and 1 coal stock. The Sector Model is in energy.

I’m in oil up to my eyeballs.

Mean reversion in oil would be a very good thing right now J.

 

Tim