Estimating EPS
March 5, 2009
UPDATED JANUARY 18, 2011
This post supercedes Estimating EPS (March 5, 2009) as BetterInvesting changed its provider of SSG data from S&P to Morningstar on December 22, 2010. And, the Morningstar data also comes with a long-term EPS estimate from Zacks in place of the previous estimate from S&P.
In addition to making those revisions here, I also took the opportunity to add new material as well as to check and recheck the accuracy of everything else. Because many of my other posts link to this one, I decided not to change its URL/web address and put all the new stuff in blue.
By far, the most important judgment on the SSG is our estimate of the company’s EPS growth rate for the next five years. This estimated rate directly determines the High and Low EPS in dollars which are then used (in part) to determine the Forecast High and Low Prices. No other SSG judgment has such a major impact on the final result.
Our estimate of future Sales growth is only used as part of the Better Investing/NAIC Preferred Procedure which is one method for estimating EPS. I use another method and here’s how I estimate EPS for all my SSGs:
(1) I ALWAYS check seven long-term EPS estimates for every SSG I do, from:
- FactSet CallStreet via CNNMoney.com
- FactSet via Reuters.com
- Thomson FN via YahooFinance.com
- Thomson Reuters via Reuters.com
- Zacks.com
- Zacks via BI.org [the only one I pay for which also comes with SSG data from Morningstar] [added 1-18-11]
- Value Line (from my library)
(2) The first five are free and easy to get from the Research Links page at StockCentral.com. Once there, just type in a ticker symbol and click the estimate link which will take you directly to the correct page at that website.
- The Stock Central Reasearch Links page is free and you will find a link to it on My Favorite Links page; click here.
(3) These are all long-term EPS estimates. Reuters doesn’t say how long, VL is for 3-5 years, and the remaining five are for the next 5 years.
If anyone knows of other free estimates from different data sources (not simply different web sites), please let me know.
- There’s some dispute about the meaning of VL’s 3-5 year period. Manifest Investing, a subscription site that uses the SSG, thinks it means 4 years and then extrapolates another year as our SSGs want a 5 year estimate. I disagree and think VL’s estimate is for a 3 and 4 and 5 year period. [added 3-17-11]
(4) Note that these are seven different estimates, not just from seven different websites. It’s not always easy to tell, but I believe the following is accurate [revised 1-18-11]:
- Thomson Reuters (old Thomson/First Call and Thomson FN still to some) provides EPS estimates to YahooFinance, Reuters.com, CNBC, and AOL’s Daily Finance;
- Zacks Investment Research furnishes EPS estimates to MSNMoneyCentral, NASDAQ.com, SmartMoney.com, StockSelector.com, ClearStation.com, and Zacks.com.
Thus, it’s not very helpful to check several websites when the EPS estimate is provided by the same source, say from MSN Money and SmartMoney since Zacks provides both estimates. On the other hand, it is very helpful to check different data sources (not just different websites) for their EPS estimates in order to learn which is high, low and wacky.
(5) I evaluated the different estimates before BI switched from S&P to Morningstar and compared the 7 estimates for 6 companies over 7 quarters together with EPS estimates from Take Stock [added 1-18-11]:
- All the estimates are tabulated and my findings set forth at: Evaluating EPS Estimates, click here.
** I confirmed that all seven of these estimates are still different;
** I found that Value Line’s estimates were often at the far end (either low or high) of the seven; and
** I did not include Take Stock in the averages I calculated because, compared to the seven mainstream estimates, it was repeatedly irrational and outrageously untrustworthy.
(6) Seperately, I learned that Zacks.com EPS estimates are also different from the Zacks EPS estimates that now come with SSG data from BI. To make sure, I checked 25 different stocks in 7 different industries and the two estimates were always different. [added 1-18-11]
(7) Only one of the seven estimates, from Reuters.com, provides more info than the other estimates which only provide the consensus estimate. In addition to the consensus estimate, Reuters also provides the number of contributing analysts and the high and low estimate.
- The extra info provides helpful context and I put less weight on estimates that are made by only a few analysts. The high and low estimates are also useful to indicate whether those analysts are in close agreement or far apart. [revised 1-18-11]
(8) StockCentral.com, like BetterInvesting.org, also provides the same SSG data from Morningstar that also includes an EPS estimate from Zacks. Previously, StockCentral’s SSG data did not include any EPS estimate. [revised 1-18-11]
- Both StockCentral and BetterInvesting are subscription sites, and you can find a link to them, their rates, and free trials on my “Favorite Links” page; click here.
(9) We don’t know the names of the analysts that make these EPS estimates and therefore can’t determine the extent of overlap among the seven that I routinely check. However, we DO know that the these estimates are far from identical. Consider these current examples [added 1-18-11]:
- Value Line’s long-term EPS estimate for Baidu (BIDU), a high-flying Chinese Internet search engine, was 44.00% on 1-13-11, FactSet CallStreet via CNN Money was 50.00%, Thomson FN via YahooFinance was 54.63%, Thomson Reuters via Reuters.com was 56.17%, FactSet via Morningstar.com was 56.80%, and Zacks.com was 59.19% while Zacks via BI was 70.50%.
** Here, VL was the lowest, Zacks via BI the highest, and the difference was a whopping 26.50 percentage points [hereafter: pp]. Moreover, the difference between just the two Zacks estimates was a surprising 11.30 pp. And, if VL was eliminated as an atypical outlier, the spread between the highest and lowest was still a substantial 9.19 pp;
- Cardinal Health (CAH) is another example where VL was again low at 8.50%, FactSetCallStreet via CNNMoney was high at 14.00%, and the spread between highest and lowest was 5.50 pp. Moreover, the difference between the two Zacks estimates was just over 3.00 pp;
- Intel (INTC) is a third example where VL was high this time at 17.50%, Zacks.com was low at 9.14%, and the spread was a hefty 8.36 pp.
** Without VL, the spread between the highest and lowest was 2.70 pp and the difference between the two Zacks estimates was 1.86 pp.
(10) After collecting all seven estimates, there are several ways to utilize the data:
- Excel can easily calculate the average and the average less one Standard Deviation. That, in my judgment, results in a reasonably conservative estimate and adequately “wrings-out” over-optimism. If you want to be tougher, you could reduce the average by two Standard Deviations although I think that is usually too tough as this is not rocket science;
- Also, any odd-ball estimate could be treated as an outlier and eliminated from the average. And, the lowest estimate of the five (or six) could turn out to be more reasonable than the average less one (or two) SDs;
- You could also compare the average and the lowest estimate to your own EPS estimate, especially one from BI/NAIC’s Preferred Procedure which involves making four other estimates for the next 5 years: Sales growth, Pre-Tax Profit Margin, Tax Rate, and Shares Outstanding. I stopped using the PP because, for me, it involves too many uniformed estimates and too much guesswork. See:
Determining What’s Reasonable and What’s Not: An Update (probing the PP of CTSH)
Pondering the Preferred Procedure (peering at the PP of ORCL)
(11) There are several reasons why I check all seven EPS estimates:
- I learn which are out-of-whack and which are in the ball park;
- I learn what is conservative and what is optimistic; and
- I have some rational basis for what seems reasonable and what’s not.
Moreover, when I do my SSGs, I do not have to rely on my gut feelings or some ambiguous sense of sleeping well. I sleep better knowing that I have tried to be reasonable by doing my homework.
Armin
### Please let me know what you think about this update by leaving a comment in the box below and/or rating this post by mousing-over and clicking the star scale also below. Your feedback is important to me.
Hello!
Very Interesting post! Thank you for such interesting resource!
PS: Sorry for my bad english, I’v just started to learn this language
See you!
Your, Raiul Baztepo
Hi Raiul:
Your English is plenty good and you are welcome to visit my blog anytime.
Armin
Hello !!! ^_^
My name is Piter Kokoniz. oOnly want to tell, that your blog is really cool
And want to ask you: will you continue to post in this blog in future?
Sorry for my bad english:)
Thank you:)
Your Piter
Hi Piter:
Your English is plenty good and you are welcome to visit my blog anytime.
If you’re serious about using the SSG to invest in growth stocks, you need to start using the SSG. You might try to redo one of my recent SSG posts and see if you can duplicate each number.
Armin
Armin,
I found Yahoo.Finance rates the analyst and rates how the analyst did on a particular stock. For those stocks that don’t receive a lot of coverage this might be a better way to look at the eps guess. Those large stocks that have extensive coverage then using excel sounds to be fruitful. Have you tracked how well the std dev method works? Also SmartMoney.com has free coverage on eps projections.
EPS is very important. How do you determine the PE to go with it?
Bob
Hey Bob:
Thanks for the comment.
- I saw no way to look at the EPS estimates of the analysts tracked by YahooFinance. I saw their names and ratings, but not their estimates. I could not even learn if these estimates were for the long term.
- As I pointed out, SmartMoney’s EPS estimates come from Zacks which I already review at Zacks.com. Checking the same data source at different websites doesn’t help, what does is checking different data sources.
- As you know, High PE x High EPS = High Price. For every SSG I do, I check to see that I don’t substantially exceed Value Line’s High Price estimate. If I do, I lower my High PE. For those stocks where VL doesn’t make a High Price estimate, I begin my SSG by using Toolkit’s Alt-M command which is usually the most conservative option.
Armin
Armin,
I’m having a lot of problems with the High PE x High EPS = High Price. My trouble is with High PE and the elements used to compute it. The High PE is computed by taking the year’s high price and dividing by that year’s eps. The problem is the full year’s eps was not known at the time the high price was used. The eps used was some future projection at the time of high price. If the projected earnings turned out to be higher than the actual eps then the computing the High PE using the lower number would result in PE that is too high and vice versus. A similar problem would occur for low PE.
Bob
Bob:
I already explained how I handle forecasting the High PE and can’t help you any further.
Here’s some large-scale guidance: my overall goal is to achieve a reasonably conservative SSG and I use comparisons all the time to guide me. Does that help at all?
You could ask the BI and/or the StockCentral Discussion Lists and you’ll find links to them on my Favorite Links page. If you ask, I suggest that you provide a concrete example rather than hypothetical generalizations.
Armin
Thank you,
very interesting article
Armin:
You don’t mention Manifest Investing, which also offers a 5-year EPS. It likely does come from VL, which would not make it particularly useful, unless you don’t take time to get VL yourself.
Ron
Thanks Ron:
I don’t mention Manifest because, as I tried to explain in my blog post, it is redundant as it is based on VL which is one of the seven estimates I always check. Also, I’m always on the lookout for more free estimates that are from a source other than the seven (Manifest is neither free nor different) so let me know if you learn anything.
In my recent post on Contemplating Cognizant, I eliminated VL because it distorted the average. As I discussed in Estimating EPS, sometimes VL is
way high and other times atypically low. The larger point here is that we need to check all seven of the long-term EPS estimates to learn which are in the ballpark and which are out-of-whack.
Armin
Armin from Item #8 you refer to:
Better Investing.org provides SSG data from S&P which does include an EPS estimate for the next 5 years.
I have a BI Subscription W/O the online SSG upgrade as I use Toolkit 6 exclusively. Were do I find a 5 year EPS estimate from the BI site you refer.
-Bruce
Hi Bruce:
I pay for a subscription to S&P data from BI (which includes access to the Online SSG and which I almost never use because it is inferior to SSG software) and that data includes an EPS estimate for the next 5 years.
- if you don’t subscribe, what data do you use for your SSGs?
Armin
Armin, I think you refered to Moneycentral,and Yahoo as both using Thompson Reuters data. And they say they do. However the EPS numbers issued are very unique. Then add in Reuters.com and you get three sets of different numbers. All from Reuters based sites. Quite confusing.
-Bruce
Hey Bruce:
I agree, long-term EPS estimates on the web can be very confusing, especially when the same data source reports a different estimate to different websites.
- by sticking to my seven different analysts, I avoid confusion AND get all the different ones that are free.
Armin
And Finally,
Are you aware of this site?
http://www.tickerpedia.com/stockEstimates.htm?quote=COH
It shows multiple sites all on one page. You can change the symbol. Coach was used in the example.
-Bruce
No, I was not aware of the site you mentioned. However, I saw no long term EPS estimates AND the five sites it links to are not as comprehensive as the five sites at StockCentral’s Research Links page.
- Tickerpedia misses Morningstar and Zacks.
Armin
I do subscribe to Betterinvesting.org and use it for my data source to Toolkit. May be a grandfather feature. I pay $50 per year rather than the $79. But with the $50 fee I do not get the Online SSG. I do get the monthly BI magazines also. I was just curious if I was missing something. Thanks for all your responses. Keep up the good work.
Hey Bruce:
Just to be sure: import a MSFT SSG data file into your Toolkit, and then click the “data” button at the top center of the page. You should see an 11.00% EPS estimate from S&P at this time.
If you don’t see any estimate, check your TK 6 data feed page (click: options ->->preferences ->-> data feed). Make sure your BI login and password are correct, and that the Better Investing data feed is selected.
I pay $79 for the S&P data and I’d like to get it (with the 5 yr EPS estimate) for $59.
Armin
Armin
The comments above pertty much say what I was thinking. I just wanted to say thank you for taking the time and effort.
Thanks Chuck, I appreciate your feedback.
I hope you check out some other posts of mine: Savoring Sysco (SYY) and Appraising Aeropostle (ARO) are the most recent and you’ll find links to them on my home page.
Armin
Nice work! great website
Armin,
great job on the due diligence!
Please make sure when comparing percentages to be accurate in differentiating between percent and percentage points. The two Zacks estimates differ by 11.3 percentage points; the larger exceeds the smaller by 19 percent.
Thanks Bob, I appreciate the good advice.
I always try to express the difference between two percentages as percentage points (11.3%), and don’t ever use the other percentage (19%) in the context of our SSGs as it confuses me.
My next Blog post will be on GE as it was this month’s Online Stock Study. I was impressed with TEVA and bought some after I wrote it up for the Blog.
Have you done any recent write-ups for BI’s First Cut?
Armin
11.3 (in our example) should not be written as 11.3% as it is the difference between two percentages and is not itself a percent. Percentage points when comparing percent values; Percent (%) only when dividing one number by another.
No new First Cuts since Dec/Jan. Time to do some more!
Bob, thanks again
I misunderstood your first comment and thought your concern was my math rather than my expression.
I made the corrections as you suggested.
A
I appreciate all you do. I try to follow your method of checking the sources for estimates for my SSG’s. Thank you for putting me on to it.
I wonder why you don’t check Manifest Investing? Is it because they use Value Line? Is it because one has to pay to access MI?
I thought MI used sales to project eps as one would do a preferred procedure? I think Value Line just projects the earnings which is slightly different.
I checked CAH. VL shows estimated eps rate at 3.82 in 2016 and MI $3.36 in five years. This becomes $3.68 also in six years (9.5% eps projected per Value Line). This is only 4% difference if I do the math correct. Pretty close using two different procedures or not? Thank you in advance.
Hey Pete:
Thanks for the kind words.
I don’t check EPS estimates by Manifest Investing for several reasons, perhaps most importantly because I fundamentally disagree with its methods. See: http://arminfields.wordpress.com/2009/01/31/studying-stryker-syk-and-mulling-over-methods/
Other reasons: MI extends VL Sales estimates for one or more years which misinterprets VL in my opinion; and MI uses the Preferred Procedure which I think is inaccurate and full of guess work, and also mistakenly extends these estimates 1 or more years. See: http://arminfields.wordpress.com/2009/03/28/pondering-the-preferred-procedure/
And, MI is a pay site.
I also don’t include EPS estimates from Take Stock which I think are wacky and unreasonable and irrational. See: http://arminfields.wordpress.com/2010/10/19/evaluating-eps-estimates-7/
Check out these other posts and let me know what you think.
Armin
Armin, how do you know the difference between the FactSet data and the Thompson Reuters data at Reuters.com? When I enter a symbol and click on Analysts I get one set of data, am I doing something wrong?
Hey Woody
Only Reuters.com usesThompson-Reuters data while Morningstar.com uses FactSet data and CNNMoney.com uses FactSet CallStreet data.
All three provide different long-term EPS estimates and I urge that everyone check all seven estimates I identified for every SSG to learn which is high, low, and/or nutzo.
Armin