This is a copy of an e-mail that I recently sent Irving Roth. I originally posted this with I-Club. I hope this will encourage further discussion.  Italicized words indicate questions asked by Irving Roth.

Hi Irving:

“So, my question is, if I'm intrerested in what EPS will be ONE year from the now, using the last 4 quareters of EPS, why would I be concerned as to where I start the FIVE year trend line ?”

Short Answer: Because I am more interested in the future projection of EPS based on my trend line and assumptions then I am in extrapolating from the past 4 quarters. I don’t want to wait another 4 quarters to see the recession results on the SSG – on the SSG graph and on the FTM EPS in Section 3.

Long Answer: This may not make such a difference in a bull market. But, in a bear market where analysts are publishing EPS estimates way below current EPS levels – and where no one (analysts or Wall Street or investors, etc) really knows what to expect from the current recessionary economy one, two, or even three years from now – establishing EPS based on my own conservative trend lines seems to be the prudent thing to do to protect against any loss of principal.

I have done several SSG’s where the current 3 options offered in Toolkit 6 for the start of my projected trend line is nowhere near the analysts EPS estimates. Why would I have any faith in my projections then? I would then have to enter a negative number for my estimated growth rates. That negative growth estimate would have an adverse effect on the FTM EPS in Section 3.

“What difference does it make from where you start the trend line?”

Case in Point: Please look at my attached SSG of VSEC. I have eliminated the years that I think are outliers. I did this to create a historical trend line that seems to be at the mid-points of the re-drawn historical EPS line. (I am using an old technique I learned when I first joined NAIC. Where the investor would draw historical trend lines using either the: 1) Mid-point method; 2) the High point method; or, 3) the Low point method…if my memory serves me correctly.) I did this to start my projected trend line at a lower EPS level on the SSG graph then using either: 1) last annual EPS, or 2) latest quarterly earnings. Unfortunately, the Future EPS calculated in Section 3 of Toolkit 6 SSG is based only on the last 4 quarters which does not take into account my beginning EPS for my projected 5 year trend line.

So my rhetorical question is: What good is a one year future EPS based on past results that 1) don’t take into consideration my trend line assumptions; and, 2) don’t take into consideration the current market environment?

What good is a projected trend line that can not start with a user imputed numerical value?

If I am going to make investment mistakes, I want the option to control all assumptions I make in toolkit 6.

I have several suggestions to help implement my ideas for Toolkit 6.

1) On Toolkit 6 is it possible to assign numerical values on the most recent year vertical line on the SSG graph for all 3 trend line options? (Last Annual Data, Last Quarterly Data, End of Trend Line)

2) Or, would it be easier to develop a SSG Graph option where the user can enter a numerical value for the start of the user’s projected trend line?

I would like to suggest you use both # 1 and # 2 as options for defining a starting EPS numeric value for the start of my projected trend line. # 2 would be the easiest to implement.

# 1 would be implemented by using a PV calculation to discount the 5 year future EPS back to the start of the projected trend line. This could be used for all three current options available on Toolkit 6. (Last Annual Data, Last Quarterly Data, End of Trend Line)

The formula is as follows:

Present Value(start of Projected Trend Line)= Future Value (5 year projected EPS)/[(1+i)n], where i is the periodic discount rate (estimated growth rate) and n is the number of periods.

You might be able to use and edited PV function of Microsoft Excel’s built in function, PV.

=PV(rate,nper,fv)  where rate is the growth rate per period, nper is the number of periods, and fv is the future EPS value.

I hope this helps. Please let me know if you have any further questions.

Nick A. DiVirgilio

The graph is a semi-logarithmic graph.  If the EPS has a constant growth rate, the graph will be a straight line, and the slope of the line is related to the growth rate.

If the graph turns upward, the growth rate is increasing with time.

If the graph turns downward, the growth rate is decreasing with time.

Let me quibble a bit with your words.

The graphs produced by Toolkit are not inaccurate.  They are absolutely accurate.  The plotted points represent actual data.

Toolkit's trend lines are the "best-fit" straight lines to the data.  The R-squared number associated with the line is a number that measures "goodness of fit."  When R-squared is 1, that means that the line goes through each of the data points.  When R-squared is less than 1, that means that the data points deviate from a straight line.

You are absolutely correct in saying that the EPS plot is never a constant growth rate.  But ... there are many examples of companies whose histories exhibit a nearly constant growth rate.

When the EPS and Sales growth rates are nearly constant and nearly the same, we call that graph "railroad tracks."  We look for that! 

When the EPS and Sales growth rates are separately nearly constant but differ one from another,  we might be concerned that these do not track.  Since we believe that earnings are produced by sales, this can be troubling.

You are right that Value Line doesn't assume that historical growth rates will be the same as future growth rates.  They give a range.  I don't have a problem with this.  If their predicted range does not include the historical growth rate, that would be a red flag for me.