An overview of the applied stock rating strategy concept is given. Its idea and basic assumptions are outlined on a broader, conceptual level.
In the previous section “The project – Approach” a brief explanation of how stock rating is applied to enhance the decision making process and accomplish the overall task is given. In the current section the stock rating concept is explained further. Examples are provided to illustrate the details and to improve comprehension of the matter.
The stock rating process can be divided into three successive steps:
- prefiltering
- stock rating
- stock choosing
Though not part of the stock rating concept itself, it should be mentioned that the monitoring of the performance of the chosen stocks is an essential part of applying a stock rating strategy, too.
1. Prefiltering
The step number one of the stock rating process, the prefiltering, is about excluding stocks from the stock rating. If, for example, we only accept stocks of companies which were able to generate positive earnings in the last three business years, we check for earnings and exclude the stocks from the rating if we have found negative ones. If we don’t want to buy stocks with falling prices, we can exclude these as well. The idea of prefiltering can be well summarized by the following words by Sherlock Holmes:
When you have eliminated the impossible, whatever remains, however improbable, must be truth.
Sherlock Holmes
And, well, we do want stock prices that go up in the long run, don’t we?
Another way of exclusion would be to filter stocks manually after the stock rating. I personally don’t like this way, because it lets you sit in front of a much larger number of potential stock options. The filtering process then might become very time consuming and the indecisiveness becomes greater. And that is exactly what we want to overcome by our stock rating concept. A further advantage of the definitive way of exclusion by prefiltering is that we really need to think about the exclusion parameters. What do we think makes up “bad” stocks? What does “bad” stocks separate from “good” stocks? And how can these features be expressed by measurables? By which measurables that are accessible through available data? So, it becomes quite clear that prefiltering is a very important part in the design of our stock rating strategy.
2. Stock rating
Let’s assume we have successfully designed and applied the prefilter and now are dealing with step number two, the stock rating of the remaining subset of stocks. As mentioned in the previous section, the stock rating is about calculating a number, the total score. This calculation is based on evaluating and classifying certain features, i.e. financial key metrics and stock prices. To classify a feature value, three classes are defined. For calculation reasons, each of these classes is associated with a number:
- outperformer: +1
- neutral: 0
- underperformer: -1
The numbers which define the range and borders of the classes are designed by asking the question “Which feature value makes a stock a stock we want to buy?”
To give an example, let’s have a look at the widely used stock feature “P/E ratio”. The P/E ratio is calculated by dividing the stock price by the earnings. The resulting number is interpreted. Usually, lower numbers are preferred, because lower numbers imply that company’s future earnings will cover the amount of our investment at a faster rate. A P/E ratio of 10, for example, implies that the total sum of earnings of ten years are equal to the price of the stock today – assuming the earnings remain the same over the whole period. And since a stock is a share in a company’s equity, we got kind of our invested money back.
If we are looking for stocks with low P/E ratios, then it might make sense to set the range of the performer classes as follows:
| P/E | class | score |
|---|---|---|
| < 10 | performer | +1 |
| 10 – 15 | neutral | 0 |
| > 15 | underperformer | -1 |
The same we can do with other features, e.g. the PEG ratio. The PEG ratio is defined as P/E divided by percentage growth of earnings. The following performance classification and scoring might make sense if we are looking for reasonably priced stocks:
| PEG | class | score |
|---|---|---|
| < 1 | performer | +1 |
| 1 – 2 | neutral | 0 |
| > 2 | underperformer | -1 |
More features can be taken into account. Also, the features might be grouped into feature classes. E.g. the mentioned features P/E ratio and PEG ratio can be grouped into the feature class “price”. Other feature types might form other categories, e.g. “quality” and “sentiment”. Depending on which feature category our strategy wants to stress, different category weights might be introduced. As an example, the categories might be weighted as follows:
| category | weight |
|---|---|
| price | 25 % |
| quality | 50 % |
| sentiment | 25 % |
The single feature scores are weighted according to the category weights. If, taking the above example, the price category has two features, then each of the two features has a weight of 12.5 %. If the quality category has five features, then each feature in the quality category has a weight of 10 %. If the sentiment category consists of only one feature, this single feature has a weight of 25 %. To check if we have assigned correct weights to all features, we check if all the single features sum up to 100 % again.
Finally, to calculate the total score of a single stock, all weighted scores of all categories are summed up. The highest possible total score, again, is 100 %.
3. Stock choosing
The third and last step is about choosing stocks from the list of rated stocks. So, all stocks that passed the prefiltering have been assigned a total score during the rating process. One approach to choose stocks is simply to choose the first stocks with the highest total scores, the first 20 to 30 stocks for example.
I personally don’t like this approach a lot. I think it is soulless. I think, now, having the list of reasonable choices, the fun part of stock buying can begin. The fun of exploring the companies, e.g. their history, their business, their technology, their industry sector and the market conditions. I think it’s reasonable to have a closer look at the stocks and the companies before buying.
And still, there are features that are hard to score during an automated routine and therefore are not reflected by the calculated total score. How could we rate a business model during an automated routine? The business model is without a doubt worth to check, especially when it comes to buy stocks we want to hold for a long time. And, what about the rumors, the integrity of the executive board, the scandals the company was or is involved in? The more complex facts and the gossip?
Also, as mentioned in the previous section, further checks on single stocks can be done e.g. by examining single feature values closer or by comparing feature values with the ones of industry sector competitors. Furthermore, it might be reasonable to factor in future growth estimates of the companies or their industry sector.
This also leads to a limitation of this stock rating concept. In its simplest form, it’s applying the same rating rules to all stocks from all industry sectors. Hence, the rating is a bit unfair, because feature values of stocks are usually in ranges typical for their industry sector. A feature value might be considered low for a stock of industry sector one, but might be normal for industry sector two. Software industry stocks for example have relatively large PEG ratios in comparison to other industry sectors.
So, I recommend to check twice before finally buying a stock. The stock rating gives a reasonable choice of stocks worth considering to buy. Then, the cheerful part of stock buying can begin.