While at RealMoney, I wrote a short series on data-mining. ?Copies of the articles are here: (one, two). I enjoyed writing them, and the most pleasant surprise was the favorable email from readers and fellow columnists. As a follow up, on April 13^th, 2005, I wrote an article on analyst coverage — and neglect. Today, I am writing the same article but as of today, with even more detail, and comparisons to prior analyses.

As it was, in my Finacorp years, I wrote a similar piece to this but it has been lost; I can?t find a copy of it, and Finacorp is in the ash-heap of financial firms. (Big heap, that.)

For a variety of reasons, sell-side analysts do not cover companies and sectors evenly. For one, they have biases that are related to how the sell-side analyst’s employer makes money. It is my contention that companies with less analyst coverage than would be expected offer an opportunity to profit for investors who are willing to sit down and analyze these lesser-analyzed companies and sectors.

I am a quantitative analyst, but I try to be intellectually honest about my models and not demand more from them than they can deliver. That’s why I have relatively few useful models, maybe a dozen or so, when there are hundreds of models used by quantitative analysts in the aggregate.

 

Why do I use so few? Many quantitative analysts re-analyze (torture) their data too many times, until they find a relationship that fits well. These same analysts then get surprised when the model doesn’t work when applied to the real markets, because of the calculated relationship being a statistical accident, or because of other forms of implementation shortfall — bid-ask spreads, market impact, commissions, etc.

This is one of the main reasons I tend not to trust most of the “advanced” quantitative research coming out of the sell side. Aside from torturing the data until it will confess to anything (re-analyzing), many sell-side quantitative analysts don’t appreciate the statistical limitations of the models they use. For instance, ordinary least squares regression is used properly less than 20% of the time in sell-side research, in my opinion.

 

Sell-side firms make money two ways.They can make via executing trades, so volume is a proxy for profitability.They can make money by helping companies raise capital, and they won?t hire firms that don?t cover them.Thus another proxy for profitability is market capitalization.

 

Thus trading volume and market capitalization are major factors influencing analyst coverage. Aside from that, I found that the sector a company belongs to has an effect on the number of analysts covering it.

 

I limited my inquiry to include companies that had a market capitalization of over $10 million, US companies only, and no ETFs.

 

I used ordinary least squares regression covering a data set of 4,604 companies. The regression explained 82% of the variation in analyst coverage. Each of the Volume and market cap variables used were significantly different from zero at probabilities of less than one in one million. As for the sector variables, they were statistically significant as a group, but not individually.Here’s a list of the variables:

 

Variable	?Coefficients	?Standard Error	?t-Statistic
?Logarithm of 3-month average volume	?0.57	0.04	?15.12
?Logarithm of Market Capitalization	?(2.22)	0.15	(14.69)
?Logarithm of Market Capitalization, squared	?0.36	0.01	?31.42
?Basic Materials	?(0.53)	0.53	?(1.01)
?Capital Goods	?0.39	0.54	?0.74
?Conglomerates	?(0.70)	1.95	?(0.36)
?Consumer Cyclical	?0.08	0.55	?0.14
?Consumer Non-Cyclical	?(1.40)	0.55	?(2.52)
?Energy	?2.56	0.53	?4.87
?Financial	?0.37	0.48	?0.78
?Health Care	?0.05	0.50	?0.11
?Services	?(0.30)	0.49	?(0.61)
?Technology	?0.82	0.49	?1.67
?Transportation	?2.92	0.66	?4.40
?Utilities	?(1.10)	0.60	?(1.82)

 

In short, the variables that I used contained data on market capitalization, volume and market sector.

An increasing market capitalization tends to attract more analysts. At a market cap of $522 million, market capitalization as a factor adds no net analysts. At the highest market cap in my study, Apple [AAPL] at $469 billion, the model indicates that 11 fewer analysts should cover the company. The smallest companies in my study would have 3.3 fewer analysts as compared with a company with a market cap of $522 million.

 

Market Cap	?Analyst additions
?10.00	?2.30
?30.00	?3.40
100.00	?4.61
300.00	?5.70
522.20	?6.26
?1,000.00	?6.91
?3,000.00	?8.01
10,000.00	?9.21
30,000.00	?10.31
100,000.00	?11.51
300,000.00	?12.61
469,400.30	?13.06

 

The intuitive reasoning behind this is that larger companies do more capital markets transactions. Capital markets transactions are highly profitable for investment banks, so they have analysts cover large companies in the hope that when a company floats more stock or debt, or engages in a merger or acquisition, the company will use that investment bank for the transaction.

 

Investment banks also make some money from trading. Access to sell-side research is sometimes limited to those who do enough commission volume with the investment bank. It’s not surprising that companies with high amounts of turnover in their shares have more analysts covering them. The following table gives a feel for how many additional analysts cover a company relative to its daily trading volume. A simple rule of thumb is that (on average) as trading volume quintuples, a firm gains an additional analyst, and when trading volume falls by 80%, it loses an analyst.

 

Daily Trading Volume (3 mo avg)	Analyst Additions
3	0.6
10	1.3
30	1.9
100	2.6
300	3.2
1,000	3.9
3,000	4.5
10,000	5.2
30,000	5.8
100,000	6.5
300,000	7.1
1,000,000	7.8
3,000,000	8.4
4,660,440	8.7

 

An additional bit of the intuition for why increased trading volume attracts more analysts is that volume is in one sense a measure of disagreement. Investors disagree about the value of a stock, so one buys what another sells. Sell-side analysts note this as well; stocks with high trading volumes relative to their market capitalizations are controversial stocks, and analysts often want to make their reputation by getting the analysis of a controversial stock right. Or they just might feel forced to cover the stock because it would look funny to omit a controversial company.

Analyst Neglect

The first two variables that I considered, market capitalization and volume, have intuitive stories behind them as to why the level of analysts ordinarily varies. But analyst coverage also varies by industry sector, and the reasons are less intuitive to me there.

 

Please note that my regression had no constant term, so the constant got embedded in the industry factors. Using the Transportation sector as a benchmark makes the analysis easier to explain. Here’s an example: On average, a Utilities company that has the same market cap and trading volume as a Transportation company would attract four fewer analysts.

 

Sector	?Addl Analysts	?Fewer than Transports
?Transportation	?2.92
?Energy	?2.56	?(0.37)
?Technology	?0.82	?(2.10)
?Capital Goods	?0.39	?(2.53)
?Financial	?0.37	?(2.55)
?Consumer Cyclical	?0.08	?(2.84)
?Health Care	?0.05	?(2.87)
?Services	?(0.30)	?(3.22)
?Basic Materials	?(0.53)	?(3.46)
?Conglomerates	?(0.70)	?(3.63)
?Utilities	?(1.10)	?(4.02)
?Consumer Non-Cyclical	?(1.40)	?(4.32)

 

Why is that? I can think of two reasons. First, the companies in the sectors at the top of my table are perceived to have better growth prospects than those at the bottom. Second, the sectors at the top of the table are more volatile than those toward the bottom (though basic materials would argue against that). As an aside, companies in the conglomerates sector get less coverage because they are hard for a specialist analyst to understand.

 

My summary reason is that “cooler” sectors attract more analysts than duller sectors. To the extent that this is the common factor behind the variation of analyst coverage across sectors, I would argue that sectors toward the bottom of the list are unfairly neglected by analysts and may offer better opportunities for individual investors to profit through analysis of undercovered companies in those sectors.

Malign Neglect

Now, my model did not explain 100% of the variation in analyst coverage. It explained 82%, which leaves 18% unexplained. Some of the unexplained variation is due to the fact that no model can be perfect. But the unexplained variation can be used to reveal the companies that my model predicted most poorly. Why is that useful? If my model approximates “the way the world should be,” then the degree of under- and over-coverage by analysts will reveal where too many or few analysts are looking. The following tables lists the largest company variations between reality and my model, split by market cap group.

 

Behemoth Stocks

?

Ticker	Company	Sector	Excess analysts
BRK.A	Berkshire Hathaway Inc.	07 – Financial	(25.75)
GE	General Electric Company	02 – Capital Goods	(20.47)
XOM	Exxon Mobil Corporation	06 – Energy	(19.32)
CVX	Chevron Corporation	06 – Energy	(14.64)
PFE	Pfizer Inc.	08 – Health Care	(14.57)
MRK	Merck & Co., Inc.	08 – Health Care	(12.76)
GOOG	Google Inc	10 – Technology	(11.44)
JNJ	Johnson & Johnson	08 – Health Care	(11.39)
MSFT	Microsoft Corporation	10 – Technology	(10.39)
PM	Philip Morris International In	05 – Consumer Non-Cyclical	(10.21)

?

Too many

?

Ticker	Company	Sector	Excess analysts
V	Visa Inc	09 – Services	?2.58
DIS	Walt Disney Company, The	09 – Services	?2.95
SLB	Schlumberger Limited.	06 – Energy	?4.15
CSCO	Cisco Systems, Inc.	10 – Technology	?5.22
QCOM	QUALCOMM, Inc.	10 – Technology	?5.34
ORCL	Oracle Corporation	10 – Technology	?5.98
FB	Facebook Inc	10 – Technology	?8.28
AMZN	Amazon.com, Inc.	09 – Services	?9.34
AAPL	Apple Inc.	10 – Technology	?10.57
INTC	Intel Corporation	10 – Technology	?11.85

?

Large Cap Stocks

?

Ticker	Company	Sector	Excess analysts
SPG	Simon Property Group Inc	09 – Services	(16.15)
BF.B	Brown-Forman Corporation	05 – Consumer Non-Cyclical	(16.03)
LUK	Leucadia National Corp.	07 – Financial	(15.93)
L	Loews Corporation	07 – Financial	(15.90)
EQR	Equity Residential	09 – Services	(15.87)
ARCP	American Realty Capital Proper	09 – Services	(15.75)
IEP	Icahn Enterprises LP	09 – Services	(15.50)
LVNTA	Liberty Interactive (Ventures	09 – Services	(15.36)
ABBV	AbbVie Inc	08 – Health Care	(15.01)
GOM CL	Ally Financial Inc	07 – Financial	(14.87)

?

Too Many

?

Ticker	Company	Sector	Excess analysts
UA	Under Armour Inc	04 – Consumer Cyclical	?16.68
BRCM	Broadcom Corporation	10 – Technology	?17.29
RRC	Range Resources Corp.	06 – Energy	?17.33
SWN	Southwestern Energy Company	06 – Energy	?17.70
RHT	Red Hat Inc	10 – Technology	?18.08
NTAP	NetApp Inc.	10 – Technology	?19.82
CTXS	Citrix Systems, Inc.	10 – Technology	?19.84
COH	Coach, Inc.	09 – Services	?20.87
VMW	VMware, Inc.	10 – Technology	?21.60
CRM	salesforce.com, inc.	10 – Technology	?22.64

?

Mid cap stocks

?

Ticker	Company	Sector	Excess analysts
FNMA	Federal National Mortgage Assc	07 – Financial	(13.84)
UHAL	AMERCO	11 – Transportation	(12.23)
O	Realty Income Corp	09 – Services	(12.06)
CIM	Chimera Investment Corporation	07 – Financial	(11.49)
SLG	SL Green Realty Corp	09 – Services	(11.46)
NRF	Northstar Realty Finance Corp.	09 – Services	(11.34)
FMCC	Federal Home Loan Mortgage Cor	07 – Financial	(11.14)
EXR	Extra Space Storage, Inc.	11 – Transportation	(10.97)
KMR	Kinder Morgan Management, LLC	06 – Energy	(10.94)
CWH	CommonWealth REIT	09 – Services	(10.51)

?

Too Many

?

Ticker	Company	Sector	Excess analysts
AEO	American Eagle Outfitters	09 – Services	?17.00
DRI	Darden Restaurants, Inc.	09 – Services	?17.40
RVBD	Riverbed Technology, Inc.	10 – Technology	?17.50
CMA	Comerica Incorporated	07 – Financial	?17.74
GPN	Global Payments Inc	07 – Financial	?18.30
WLL	Whiting Petroleum Corp	06 – Energy	?19.67
DO	Diamond Offshore Drilling Inc	06 – Energy	?21.57
URBN	Urban Outfitters, Inc.	09 – Services	?24.06
RDC	Rowan Companies PLC	06 – Energy	?24.48
ANF	Abercrombie & Fitch Co.	09 – Services	?26.02

?

 

Small cap stocks

 

Ticker	Company	Sector	Excess analysts
BALT	Baltic Trading Ltd	11 – Transportation	?(7.96)
ERA	Era Group Inc	11 – Transportation	?(7.45)
PBT	Permian Basin Royalty Trust	06 – Energy	?(7.42)
SDR	SandRidge Mississippian Trust	06 – Energy	?(7.18)
PHOT	Growlife Inc	02 – Capital Goods	?(6.79)
SBR	Sabine Royalty Trust	06 – Energy	?(6.74)
CAK	CAMAC Energy Inc	06 – Energy	?(6.64)
FITX	Creative Edge Nutrition Inc	09 – Services	?(6.57)
BLTA	Baltia Air Lines Inc	11 – Transportation	?(6.53)
VHC	VirnetX Holding Corporation	10 – Technology	?(6.49)

 

Too many

 

Ticker	Company	Sector	Excess analysts
WLT	Walter Energy, Inc.	06 – Energy	?12.19
ANGI	Angie’s List Inc	10 – Technology	?12.31
FRAN	Francesca’s Holdings Corp	09 – Services	?12.58
ZUMZ	Zumiez Inc.	09 – Services	?13.49
GDP	Goodrich Petroleum Corp	06 – Energy	?15.02
DNDN	Dendreon Corporation	08 – Health Care	?15.89
ACI	Arch Coal Inc	06 – Energy	?16.04
HERO	Hercules Offshore, Inc.	06 – Energy	?16.19
AREX	Approach Resources Inc.	06 – Energy	?17.64
ARO	Aeropostale Inc	09 – Services	?20.80

 

Microcap Stocks

 

Ticker	Company	Sector	Excess analysts
SGLB	Sigma Labs Inc	06 – Energy	?(6.18)
AEGY	Alternative Energy Partners In	10 – Technology	?(5.97)
WPWR	Well Power Inc	06 – Energy	?(5.83)
TTDZ	Triton Distribution Systems In	10 – Technology	?(5.53)
SFRX	Seafarer Exploration Corp	11 – Transportation	?(5.15)
PTRC	Petro River Oil Corp	06 – Energy	?(4.99)
UTRM	United Treatment CentersInc	08 – Health Care	?(4.82)
BIEL	Bioelectronics Corp	08 – Health Care	?(4.80)
DEWM	Dewmar International BMC Inc	01 – Basic Materials	?(4.74)
FEEC	Far East Energy Corp	06 – Energy	?(4.61)

 

Too many

 

Ticker	Company	Sector	Excess analysts
PRSS	CafePress Inc	09 – Services	?3.99
SANW	S&W Seed Company	05 – Consumer Non-Cyclical	?4.03
KIOR	KiOR Inc	01 – Basic Materials	?4.06
PRXG	Pernix Group Inc	02 – Capital Goods	?4.08
EYNON	Entergy New Orleans, Inc.	12 – Utilities	?4.17
PARF	Paradise, Inc.	05 – Consumer Non-Cyclical	?4.40
SUMR	Summer Infant, Inc.	05 – Consumer Non-Cyclical	?4.52
LAND	Gladstone Land Corp	05 – Consumer Non-Cyclical	?4.57
JRCC	James River Coal Company	06 – Energy	?6.38
GNK	Genco Shipping & Trading Limit	11 – Transportation	?7.11

My advice to readers is to consider buying companies that have fewer analysts studying them than the model would indicate.? This method is certainly not perfect but it does point out spots where Wall Street is not focusing its efforts, and might provide some opportunities.

 

 

Full disclosure: long BRK/B & CVX