Develop a series of quantitative screens in order to narrow the investable universe down to a manageable number of securities that can then be thoroughly analyzed on a fundamental basis. Ten ratios, five categories, applied identically to all 1,322 companies in the benchmark.
There are two situations, and almost every firm is in one of them.
The first is that the resources do not exist to analyse every name properly. This is the position of the overwhelming majority of managers. Even with one analyst per sector, each would be responsible for well over a hundred companies before doing any thinking.
The second is subtler and applies to the firms large enough to escape the first. Where the resources do exist to cover every name in depth, getting every analyst onto the same methodology and the same assumptions is a herding cats operation. Fail, and the portfolio is built by comparing work that is not comparable.
The framework in the previous two chapters gives country and sector boundaries. Screens give a consistent way of filtering what sits inside them, so that deep analysis is spent on names selected by the same rules everywhere.
Two measures in each of five categories.
Price to free cash flow, on a trailing twelve month basis. Price to tangible book value, on the most recent quarter.
Tangible book, deliberately: the gap between reported book and tangible book is where goodwill and intangibles sit, and a balance sheet that looks solid on one and thin on the other is telling you something worth knowing.
Return on average equity and return on average assets, both trailing twelve months.
Average rather than period-end, so that a balance sheet reshaped during the year does not flatter the ratio.
Current ratio and total debt to equity, both on the most recent quarter.
The pairing matters: strong returns produced on heavy leverage are a different proposition from the same returns produced without it, and reading the two together is what separates them.
Book value per share growth and earnings per share growth, measured across an eight year window.
Eight years rather than the ten the original framework used. It still spans a full cycle, and it is the longest history available consistently across every market in the benchmark. Consistency wins the tie, as it did with the macro factors.
Asset turnover and receivables turnover, both trailing twelve months.
The least glamorous pair on the list and among the most revealing. Asset turnover shows whether a growing capital base is being worked or parked. Receivables turnover shows whether reported sales are being collected.
Note what is not there: no margin measure. Normal margins vary enormously by industry, so ranking the whole universe on margin would quietly become a bet on sector composition. Every metric on the list is one that means the same thing in a bank and a miner, because the screens must not smuggle in a view the allocation framework has not sanctioned.
Banks and insurers are the exception that proves the rule. Several of these ratios are meaningless for a balance sheet that is the business, so financial companies are ranked on a sector-appropriate set of the same size and structure, rather than being scored on measures that cannot apply to them. An honest blank beats an invented number.
And what it costs.
Every company in the universe is scored on each screen and ranked against every other company. A percentile is taken, so that the best name on a measure sits at 100 and the worst at 0, and the ten results are combined into one figure.
The cost of ranking is real and worth stating. A company can be twice as good as the next name on a metric and finish one place ahead. Ranking discards the distance between names.
What it buys is a consistent, positive value for every company on every measure, which is what makes combining ten unlike ratios possible at all. It also means metrics can be added, removed or substituted, provided it is done uniformly across the universe. There is no natural limit to how many screens can be applied. The only binding condition is that each must be comparable across the whole universe.
And the distance is not thrown away, only set aside. It comes back in the fundamental work in the next chapter, where relative magnitude is exactly what is wanted.
The unglamorous rules that decide whether the output is trustworthy.
It is not set to zero. A company is averaged over the metrics it genuinely has, so a missing data point does not masquerade as a bad score.
Below that threshold there is not enough left to rank on, and a ranking built on six of ten measures is not comparable to one built on all ten.
The interest is at both ends of the distribution, where mispricing tends to be most pronounced, not in the middle where most companies sit and little is being said.
Nothing in this chapter has asked what a company does, who runs it, or what it is worth. It has decided where the accounting work is worth doing.
MSFT · United States · Information Technology
| The ten screens, as measured | Value |
|---|---|
| Price to free cash flow | 54.55x |
| Return on average equity | 36.3% |
| Return on average assets | 21.3% |
| Current ratio | 1.23x |
| Book value per share growth | 25.5% |
| Earnings per share growth | 17.4% |
| Asset turnover | 0.53x |
| Receivables turnover | 5.94x |
Trailing twelve months, or most recent quarter for the balance sheet measures.
| Ranked against the universe | Percentile |
|---|---|
| Valuation | 12 |
| Profitability | 93 |
| Financial strength and liquidity | 60 |
| Growth | 80 |
| Turnover | 35 |
What the screen is for. The left column is what the accounts say. The right column is the only thing the screen actually uses: where that figure sits against every other company in the benchmark. A filter, not a verdict. Nothing here has yet asked what the company does or what it is worth.