Skip to main content

Building and Using an Ass-Kicking Dividend Model

I wish you could see me on a good day.


I'm working with some new features in INVRS that makes it easier to screen using custom investment models.  I'm looking through the whole Canadian universe for kick-ass dividend stocks and I'm using the following metrics:
  1. Years of consecutive dividend growth (I set a minimum of eight for this analysis),
  2. Dividend coverage,
  3. Value relative to historical dividend yield. and
  4. Total return (defined as projected dividend yield plus projected dividend growth).
My initial screen is a dividend yield greater than 3%.  There were several hundred and I winnowed these down by running an analysis that extracted the companies that had increased their dividends per share every year for at least the past eight years. And this is where we begin...

Constructing the Template for Year Over Year Dividend Growth

To construct this model I selected eight years of dividend per share data, then wrote if(A>B,0,1) eight times, just changing the letters as I went back in time.  This says that if dividends per share in year t is greater than the dividends per share in year t-1 then assign a value of zero, if not 1.
Next, I sum the results of all the 'if' statements.  If the sum is zero, the company has increased its dividends per share every year for the past eight years.

Running the Model for Year Over Year Dividend Growth

I ran the analysis on my initial portfolio of hundreds.  INVRS lets you export the results to a new portfolio (or excel or google sheets document) and to specify new screening criteria.  I screened by selecting the column that summed the if statements and set the screening criteria equal to zero.
New portfolio size equals 33 companies.
Next factor is dividend coverage.  I wanted companies where dividends were no more than 75% of free cash flow.

Constructing the Template for Dividend Coverage

Super simple, two pieces of data: dividends per share and cash flow per share, both for the current year.  With INVRS there's no need to specify particular dates, Y is always the most current, Y-1 is last years etc.  Divide A by B.

Running the Model for Dividend Coverage

I run this one against the new portfolio with 33 stocks.  I screen out the stocks where dividend coverage is less than or equal to .75 and greater than zero.
I get a lovely little portfolio of 10.
Value is next.

Constructing the Template for Dividend Value

Since this is a dividend model it makes sense to look at value relative to dividends, right?  Right.
The value measure is the forward dividend yield over the average dividend yield for the past five years.
I calculate the forward dividend yield by dividing the current dividend rate (the latest dividend multiplied by the number of times per year the company pays dividends) by the current price.  I select dividend yield data for each of the past five years and take the average.  I divide them.

Running the Model for Dividend Value

I run this model against my portfolio of 10.  If the quotient is greater than 1 then the company is relatively cheaper than it has been over the past five years.  If needed I would have created a screen for this, but wow, they are all relatively cheap, and by a significant amount.  Thanks pandemic.
Take a look:

My portfolio of 10 remains unchanged.
Finally I look at total return.

Constructing the Total Return Template

This model calculates both elements of total return and sums them.
Projected dividend yield is the same one used in the dividend value model above.
Projected dividend growth uses a modified 5 year average growth rate as a proxy.  The first step is to calculate the five year average growth rate for the dividends (using six years of dividend per share data, then calculating the year over year growth for Y-6 to Y-5, Y-5 to Y-4, Y-4 to Y-3, Y-3 to Y-2, Y-2 to Y-1, Y-1 to Y and then taking the average for the five values).
Next I calculate the standard deviation of those five year over year growth rates (it's easy, there's a short cut std(L...P), where L to P are the columns with the growth rates).
My modification is to multiply the 5-year average by 1 minus the standard deviation.  This "punishes" companies with a lot of variability in their growth rates and "rewards" those with a consistent growth rate.
Here's the results:

Deadly huh?

Here's the 10:

  • CTC.A - Canadian Tire - Retail, Specialty Store
  • CNQ - Canadian Natural Resource - Oil & Gas
  • XTC - Exco Technologies - Manufacturing, Auto Parts
  • TD - Toronto Dominion Bank - Major Bank
  • IMO - Imperial Oil - Integrated Oil & Gas
  • TCL.A - Transcontinental - Publishing, Newspapers
  • TCL.B - Transcontinental - Publishing, Newspapers
  • LB - Laurentian Bank - Regional Bank
  • RY - Royal Bank - Major Bank
  • CWB - Canadian Western Bank - Region Bank


Popular posts from this blog

11 Reasons Why INVRS is Better Than Excel Alone

If you work with your own investment models you likely use excel to build them, but excel isn’t ideal for many reasons and it costs you in other ways.   First, you need good data and it isn’t just lying around in an easy to import format.   You’re either keying it in yourself or paying money for excel downloads.   When you need data from numerous sources – price information, data from different statements and across multiple years - you must merge it from multiple sheets.   It's a n inefficient process which can lead to data corruption. After all this data is collected and merged, models built and tested, the net result is one statistic for one company.   A stand-alone number without context has limited use.   To be meaningful, you need to compare it to similar companies.       This is just a sample of the challenges.  You need software that overcomes these problems and is designed for investment model creation. Surprise!  This software exists,

NextEra - Good Dividend in the Renewable Energy Sector

NextEra had good results relative to a group of peers in a factor-based analysis. NextEra has an appealing profitability and income profile. Its price momentum looks decent, with a caveat. Its relatively small size (a small mid-cap) coupled with its industry (renewable energy) further weight the odds that this company could be a strong performer in the future. If you want in on renewable energy, we recommend NextEra. The Analysis Overview I created a portfolio of stocks in the alternative energy sector, looking specifically for companies with a market cap over $1B but less than $4B.  This is a sweet spot that offers strong potential for growth but is also substantial enough not to be too speculative. It's my believe that alternative energy is on the ascendance, where as fossil fuels will inevitably decline (NextEra isn't a pure play in this regard however, natural gas assets are part of its portfolio).  If you share this belief and you want exposure to this ma

Analyzing Results - 2013/Y-6

 There were eight companies in Group 8 (no 9) in 2013/Y-6: EW  STZ LRCX ALXN MU MCHP AME DE This is how they each fared from 2014 to 2020 - the year runs from April 1 to March 31st:   EW         92%, -58%, 5%, 48%, 41%, -1% STZ        36%, 31%, 7%, 42%, -22%, -18% LRCX     27%, 10%, 57%, 60%, -6%, 32% ALXN     12%, -18%, -15%, -7%, 24%, -34% MU         11%, -61%, 162%, 82%, -17%, 0 MCHP     5%, 3%, 56%, 27%, -4%, -21% AME        1%, -4%, 9%, 42%, 12%, -14% DE           -2%, -9%, 45%, 45%, 24%, -34% Six Year Return: EW  69% STZ  81% LRCX  355% ALXN  -42% MU  72% MCHP  60% AME  42% DE  68% What score did each of the above have in 2014/Y-5? EW - 6 STZ - 8 (no 9) LRCX - 4 ALXN - 8 (no 9) MU - 6 MCHP - 5 AME - 6 DE -5 What score did each have in 2015/Y-4? EW - 5 STZ -6 LRCX - 6 ALXN -5 MU - 6 MCHP - 2 AME - 3 DE - 1 What score did each have in 2016/Y-3? EW - 8 STZ - 7 LRCX - 7 ALXN - 6 MU - 7 MCHP - 5 AME - 1 DE - 2 What score did each have in 2017/Y-2? EW - 8 STZ - 4 LRCX - 8 ALXN - 6 MU -