One of the first suggestions I received when I initially published my research here was that it would be useful to utilize weighting in my analysis. After all, not all ratios are equal and vary in importance from industry to industry. This is why I started working on weighting-related alterations. Readers who have seen my previous articles know that I divide my ratios into four categories: debt, profitability, efficiency, and growth. Prior to including any weighting, each category equally contributed to what eventually became the final score of a stock. This is illustrated by the following equation:
(Debt Score x .25) + (Profitability Score x .25) + (Efficiency Score x .25) + (Growth Score x .25) = Final Score
Since each category contributed exactly 25% to the final score, the model assumed that each category contributed an equal value to the determination of the final ranking/score. Yet, after doing some research, this was not the case- each type of ratio had different correlations to the stock's final return. Therefore, it is logical that there be some sort of weighting that more properly captures each category's relation to the final return of the stock.
Weight Determination Process
The basis of my weight determination process revolves around the comparison of a stock's rank in a particular category to its return over the past year. By comparing the two, I can see how closely results of each category correlates to the performance of the stock. If, for example, the debt category was more frequently aligned with how much a stock appreciated over a given year, it should be allotted more weight in the final score than a different category where less correlation is present. Below is a table that helps explain this idea.
Company: Valero (NYSE:VLO)
|'17 Debt Rank: 1st out of 5 (2)||'17 Return Rank: 1st|
|'17 Profitability Rank: 2nd out of 5 (2.67)||'17 Return Rank: 1st|
|'17 Efficiency Rank: 1st out of 5 (1.5)||'17 Return Rank: 1st|
|'17 Growth Rank: 1st out of 5 (1.6)||'17 Return Rank: 1st|
I construct tables like these for all of the companies in the industry, but for simplicity sake, I've just created one here. There are three things I look for when comparing the contents of each of the columns that help me determine how I'm going to make my weights.
1. How many times a category's rank was the exact same rank as the company's previous year's return.
In the case of Valero, this happened three times: Debt, Efficiency, and Growth. The stock saw the greatest return in 2017 and also had the best scores in these three categories. If a certain category places a company in the exact same place as how much it returned that year relative to its peers, then that category gains value when it's time to determine weighting.
2. How many times a category's rank was within 1 of the rank of the company's previous year's return.
This can be seen in the Profitability row for Valero, as it ranked 2nd out of the 5 companies in the analysis yet still placed 1st in its overall return in 2017. This is less impressive than the phenomena in point number one, and therefore, it should be valued a little lower in the weighting process if it were to occur. That being said, having a category's rank come within 1 of the final return rank still holds value because it holds a decent amount of accuracy. Hence, it is still seen as a positive in determining final weights despite it possibly occurring more often.
3. How many times a category's rank fails to come within 1 place of the rank of the company's previous year's return.
This can't be seen in the table for Valero, but it does happen. An example of this would be if a company's debt score ranked them 2nd overall, yet their last year's return was 4th among its 5 peers. Since that category seemed to show little to no correlation to the prior year's return, it should be given less of a weight when it comes time to calculate final weighting numbers.
I'll now move on to using this year's and last year's oil analysis to figure out general ideas of what weights could be. Figuring the weights isn't an exact science, as I assign them based on how each category performs according to the three points mentioned in the previous section. Below is the concept I follow for the oil companies.
# of times Debt rank aligned exactly with the stock's prior year return rank: 6
# of times Debt rank is within 1 place of stock's prior year return rank: 1
# of times Debt rank failed to be within 1 of stock's prior year return rank: 2
# of times Profit rank was equal to the stock's prior year's return rank: 0
# of times Profit rank was within 1 of stock's prior year's return rank: 4
# of times Profit rank failed to be within 1 of stock's prior year's return rank: 5
# of times Efficiency rank was equal to stock's prior year return rank: 5
# of times Efficiency rank was within 1 of stock's prior year return rank: 4
# of times Efficiency rank failed to be within 1 of stock's prior return rank: 0
# of times Growth rank was equal to stock's prior year return rank: 1
# of times Growth rank was within 1 of stock's prior year's return rank: 7
# of times Growth rank failed to be within 1 of stock's last year return rank: 1
With the above information, it becomes more clear how well each category was correlated to the oil companies' stock performance. The clear laggard was Profitability, as it not once gave a company a rank that correctly aligned with its ending return placement of the 5 companies included in the analysis. Therefore, there is no reason that profitability should have the same contribution to the final score as others, since it provides little predictive value on year end returns. The other three categories performed fairly well. Growth only had 1 complete miss, but it also only had 1 rank that was exactly right. For this reason, I placed it 3rd among the 4 categories. At the top of the list, it's slightly more difficult to determine the best category. Debt aligned exactly with prior year return 6/9 times, but still missed entirely twice. Efficiency did not have a single complete miss, as it aligned exactly with past year's returns 5 times and came within 1 rank the remaining 4 times. The fact that Efficiency did not have a miss led me to determine it was the best category with Debt right behind it at 2nd place. When all is said and done, the order from the most weighted to least weighted in my new model looks like this:
Profitability, Growth, Debt, Efficiency
Keep in mind that this order is only for my major oil company analysis. In other industries, such as telecommunications, different categories might have a greater correlation to stock price performance, thus changing the order.
Now comes the determination of what the numerical weights will be. Remember that before adding weighting to my system each category had a 25% weight to it (everything equal). After seeing that some categories exhibit more strength and value than others, I need to come up with new weights that properly reflect this. Since Efficiency displayed the most strength in my opinion, I gave it a 32% weight. Debt followed with a 28% weight and then Growth with 24%. Lastly, since it performed so poorly, I gave Profitability just a 16% weight. There was no mathematical process in determining these numbers, but I do believe that they sufficiently reflect how each category was correlated to the stock's prior year's return. So, the new equation for finding the final score of a big oil stock is as follows:
(Debt Score x .28) + (Profitability Score x .16) + (Efficiency Score x .32) + (Growth Score x .24) = Final Score
I'll spare the math, but I'll also share how this new weighting system affected the scores in my most recent oil analysis.
|Old '17 Score||New '17 Score||Old '18 Score||New '18 Score|
|Royal Dutch Shell (RDS.A) (NYSE:RDS.B)||4.11||4.17||2.82||2.96|
|Exxon Mobil (XOM)||3.49||3.61||3.17||3.36|
In the end, every company stayed in the same place as before with just some minor changes in their final scores. In future analyses, however, these minor changes could potentially change company's ranks among their competition and more accurately encompass their financial position relative to other stocks included in the analysis.
I will continue to revise this weighting system as more information becomes available, but eventually, I hope to develop some concrete ideas on how each category should be weighted, supported with several years of data. I'll also probably try to speed up this process by back-testing my system to see if the current trends in the four categories hold up. Do note that these weights will most likely be different for each industry I analyze and that the above numbers are just what I came up with for my oil analysis- although the process of determining these weights will largely stay the same. In future analyses, the determined weighting will only be briefly touched upon as more energy will be placed on the research itself and analyzing trends. As always, I appreciate any feedback or suggestions that readers think will add value to my system.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.