Piotroski's F score

A little bit over 2 decades ago, Joseph Piotroski published a research paper titled " Value Investing: The use of Historical Financial Statement Information to Separate Winners from Losers ". This is when the F-score was public for the first time.

The reasoning and the initial selection of companies

Joseph Piotroski was working on creating an investment strategy that is based on the fundamentals of the companies that have high book/market values (or in modern terms, low price/book ratios).

What is very common about companies with low price/book ratios is that they're likely to be financially distressed, could go bankrupt, and hence are trading at a price close to their book values. The reasoning behind this was that these particular companies are less followed by the analyst community and have a low level of investor interest. In many cases, there are no forecasts available or stock recommendations.

Therefore, the companies that are going to recover from this more difficult financial position are likely to provide returns above the market. Now the question was, how can someone identify the companies that are going to do better within this category? That leads us to the F-score.

The Piotroski's F-score

Joseph Piotroski came up with 9 binary signals that are based on the financial statements of a company and based on how many of those are present, a company has a score between 0 and 9. These signals were grouped into 3 buckets:

Bucket #1: Profitability

1. Positive return on assets - If the company is profitable, then it would have a positive return on assets and therefore get a tick for this signal being present.
2. Positive cash from operations - If the cash makes money from its operations, then the signal is present
3. Positive change in RoA year-over-year - If the company's return on assets improved compared to the previous year, it means it's moving in the right direction
4. Cash from operations > Net profit - This takes into account the accruals.

All of these questions together answer whether the company is profitable enough and brings enough cash to survive.

Bucket #2: Leverage, liquidity and source of funds

1. Decrease in long-term debt/total assets year-over-year - One of the worst possible signals for a struggling company is if the debt keeps increasing. However, if it decreases, it means it is moving in a good direction.
2. Increase in current ratio - Although the idea behind this is great, it has its flows. If a company's current ratio decreases by 0.0001 or 1, it is irrelevant. The magnitude of the change is not taken into account. If a certain company have excess cash and puts it into use or gives it back to the shareholders, even though it is a great decision, it would be seen as a negative event for the F-score
3. Equal or decreasing shares outstanding - Not diluting the existing shareholders is important to have as one of the signals. Otherwise, a company could issue more shares in exchange for cash and use part of that to pay its long-term debt. In this case, the 2 signals above would be met (as the debt would decrease and the current assets would increase)

Bucket #3 Operating efficiency

1. Increase in gross margin - As the idea is to identify companies that are improving, this signal is a great one. Although, it can be argued that the operating margin would do an even better job as it takes into account the operating expenses.
2. Increase in asset turnover - This particular signal is focused on the usage of the assets.

Calculation of returns

Now, if we put ourselves in the shoes of Joseph, we have identified the companies that have low price/book ratio, we have calculated the F-score for each one and now what? Well, we need to calculate the return of these companies and compare them with the market return.

Joseph measured the return from the beginning of the fifth month after the company's fiscal year-end for a year (or two, as he runs multiple backtests). What this means is, that if a company's fiscal year is January 1st until December 31st, he would measure the 1-year return starting from May 1st up until April 30th the following year.

His reasoning for choosing the fifth month is to ensure that the necessary annual information is available to investors at the time of the portfolio creation.

The outcome

The backtesting was done for the period between 1976 and 1996 and the outcome is what made this paper so famous.

The mean performance of the companies with an F-score of 0 (Meaning, not profitable, not having positive cash from operations, etc, etc.) was 6.1% below the market.

The mean performance of the companies with an F-score of 9 was 15.9% above the market!

Now, personally, every time I hear something too good to be true, I tend to be sceptical and think that it is probably too good to be true. So, what's the catch?

14,043 companies were part of the backtesting, and 333 of them had an F-score of 9. Let's focus on these as they had the best score!

Well, the worst performing 10% of them underperformed the market by 46%. The worst 25% underperformed the market by 27%. The median company with an F-score of 9 underperformed the market by 1.2%. The best 10% overperformed the market by 89%! So, it is quite clear that the winners do exceptionally well, but how many are there?

171 out of the 333 companies underperformed in the market. Although the average return of the companies with a high F-score did better than the average company with a low F-score, the score doesn't provide information to pick a stock. That means, if someone wants to put this into practice, the only way would be to buy a piece of all the 333 companies.

Joseph also did another test. He divided all the companies into two groups: High F-score (if the score is greater than 5) and low F-score (If the score is below 5) and compared their returns for every year between 1976 and 1996. In almost all of the years (except 3) the high F-score companies did better. How better? 9.7%.

Recent back-testing

There have been many backtests done recently and the outcome is quite similar. There are years when Piotroski's F-score beats the market, but there are plenty of them when the score doesn't do a good job. It is significantly more volatile compared to the S&P and it requires investing in a significant number of companies.

In the original paper Joseph made a point that if an investor shorts the companies with low F-scores and buys the companies with high F-scores, the return would be 23% annualized!

Unfortunately, if the same strategy was applied in the last 2 decades, the investor would've lost money.

Common mistake

Knowing that the F-score was created based on companies with low price/book ratios, it would not be reasonable to use it to assess a stable company's financial strength. That doesn't mean the 9 signals are not relevant, but that a score greater than 7 is more common for companies with a high price/book ratio.

If someone wants to go through the published paper, here's the link: https://www.anderson.ucla.edu/documents/areas/prg/asam/2019/F-Score.pdf