Z-score composite stock ranking, explained

A z-score, or standard score, is a basic statistical tool: it expresses how far a value lies from the average in units of standard deviation. Combining z-scores across several measures is a common technique in quantitative factor research, because it puts different measures on the same scale while keeping information about how large each gap is.

That is the difference from ranking. If one company earns a return on capital twice as far above the average as another, a rank treats them as one place apart. A z-score sees the gap.

1. Standardize each metric. For each metric, the stock's z-score is its distance from the peer-group average, divided by the peer-group standard deviation.
2. Flip lower-is-better metrics. For measures such as net debt / EBITDA, the sign is reversed so that a positive z-score is always good.
3. Cap at ±3. Each z-score is limited to between −3 and +3, so one extreme value cannot dominate.
4. Combine with your weights. The composite is the weighted average of the metric z-scores.
5. Show it on 0–100. The composite is standardized across all the scored stocks and passed through the normal distribution. 50 is average, about 84 is one standard deviation better than average, and about 98 is two.

Price multiples such as P/E are measured by their inverse, a yield, so a loss ranks as the worst reading rather than the cheapest. Target-range metrics such as payout ratio and RSI use their score against the target range.

Assume Company A is compared with a fictional peer group whose average and spread on each metric are shown below. The figures are made up for illustration and do not describe any listed company.

ROIC is 2.0 standard deviations above the peer average, net debt / EBITDA is 1.0 better (lower debt), and revenue growth is 1.0 below. With weights 3, 2 and 1, the composite is (2.0 × 3 + 1.0 × 2 − 1.0 × 1) ÷ 6 = +1.17.

Suppose the composites of all the scored stocks average +0.10 with a standard deviation of 0.70. Company A's standardized value is (1.17 − 0.10) ÷ 0.70 ≈ 1.52, which the normal distribution turns into a score of about 93.6 out of 100. The score describes A's position among these companies; it is not a forecast of its return.

1. Choose Z-score composite in the scoring-model picker.
2. Select the metrics and weights exactly as for weighted ranking, and set any filters.
3. Open a stock's breakdown to see each metric's score and its share of the total.

As in weighted ranking, a stock needs complete data for every metric you score, at least ten stocks are needed, and you can compare each stock within its own sector (a sector with fewer than five eligible stocks falls back to the whole list). Equal overall scores are ordered by ticker. See the full methodology.

• Scores bunch near 50 when stocks are similar. The scale is relative to the group, so tiny differences can be stretched or compressed.
• One strong metric can lift a stock more than under weighted ranking, because a large gap earns large credit.
• Averages and spreads depend on the group. Outliers and the filters you choose change the benchmark every stock is measured against.
• Standard deviations assume a typical spread. Skewed metrics such as growth rates do not follow a bell curve, which is why the z-scores are capped.

Stock Meadow's scores are research tools based on reported financial data. They do not predict returns or recommend buying a particular stock. For the plainer position-based version, read the weighted ranking guide.

Each measure has its own page with a formula, how to read it and common pitfalls.

• Return on Invested Capital (ROIC): profit made on all money invested in the business.
• Net Debt to EBITDA: years of earnings needed to pay off debt.
• Revenue Growth: how fast sales grew last year.
• P/E Ratio (Price-to-Earnings): what you pay for each dollar of profit.

What is a z-score in stock screening?

A z-score says how many standard deviations a value is above or below the average of its peers. A z-score of 0 is average, +1 is one standard deviation better than average, and +2 is two.

How is a z-score composite different from weighted ranking?

Weighted ranking uses only each stock's position in the list. A z-score composite also uses the size of the gaps, so a stock far ahead of its peers earns more credit than one that is just ahead.

What does a score of 50 mean in the Z-score composite?

50 is the average composite across the stocks being scored. Scores of about 84 and 98 are roughly one and two standard deviations better than that average.

Why are z-scores capped at plus or minus 3?

A single extreme value could otherwise dominate the result. Stock Meadow limits each metric's z-score to between −3 and +3, so no one figure can swamp the others.