Highly valued growth stocks fall the deepest when the substance is missing. The G-Score checks whether the growth is fundamentally backed.
Save the strategy and StockScorer checks its rules for you every day. You'll get a heads-up whenever a stock crosses the upper or lower score threshold.
Gap to MSCI World: +24.3 percentage points total, +0.8 per year
Index, start = 100
Past performance is not a reliable indicator of future results.
Based on all 16,182 stocks with a score. Not rated are stocks that the profile does not classify because of a size limit or an exclusion rule.
| Score | Number of stocks | High match | Medium match | Low match | Not rated |
|---|---|---|---|---|---|
| 0 | 1,167 | 0 | 0 | 170 | 997 |
| +1 | 4,713 | 0 | 849 | 0 | 3,864 |
| +2 | 5,046 | 0 | 1,269 | 0 | 3,777 |
| +3 | 3,150 | 0 | 1,153 | 0 | 1,997 |
| +4 | 1,951 | 941 | 0 | 0 | 1,010 |
| +5 | 155 | 107 | 0 | 0 | 48 |
Every stock runs through the same disclosed rules. The points add up to a score, traceable down to the individual rule.
Each rule checks a metric against a threshold, here for example “Return on assets above industry average (ROA)”.
In this profile, each rule that is met counts +1 point. Exclusion rules take a stock out of the rating regardless of its points: it is then not rated.
The sum is the score. That gives three classes: high, medium, low match.
These terms describe only the match with the criteria, not a recommendation to buy or sell.
Return on assets (ROA) and cash-flow return are measured against the industry median, not absolute thresholds.
Operating cash flow must exceed net income: book profits without cash coverage do not count.
Above-average R&D and capital expenditure depress today's profits but secure tomorrow's market position.
All comparisons run sector-relative: a software stock is measured against software, not utilities.
These exact rules run over every stock daily.
After signing up for free, you can copy the full profile in the rule editor, adjust its thresholds, and save it as your own starting profile.
See and copy all rules for freeIndex, start = 100
The backtest shows a total return of +264.5%. The MSCI World reaches +240.1% over the same period.
The same strategy with the same settings, started at every possible date. Each dot is a separate run over 5 years.
The periods overlap heavily, neighbouring runs share almost all periods. The spread therefore looks smaller than it would for independent periods. Past results say nothing about future ones.
Strategy calculated without fees and taxes, dividends reinvested.
Point-in-time composition since June 30, 2009: each period uses the members at that time. Stocks that left the market early may be missing, which favours the result.
Historical period, quarterly rebalancing. Past performance is not a reliable indicator of future results.
Purely mechanical rule application, no curated-list effect: stocks with a high match can fall just like any other stock.
Dots per card, in this order: Profitability, Growth, Valuation. Filled means at least one rule in the category is met.
For growth investors who do not avoid expensive stocks but want a safety net: the G-Score answers whether fundamental strength stands behind the high valuation or just a story.
While the classic 9-point balance-sheet check was developed for cheaply valued stocks, the G-Score targets the opposite: highly valued growth stocks. Partha Mohanram showed in 2005 that winners can be separated from "fallen stars" in the expensive market segment too: with binary criteria relative to the industry, applied exclusively to the lowest book-to-market quintile, meaning classic growth names with a high P/B. StockScorer reconstructs that quintile through an exclusion rule on a P/B from 3. If a company beats its industry median on returns on assets and cash flow, if operating cash flow covers profit, and if its research and capital investment sit above the industry midpoint, the high growth is fundamentally underpinned. Whoever scores below the industry midpoint everywhere is a crash candidate whose valuation rests on expectations alone.
An exclusion rule admits only growth stocks: a P/B from 3 (an approximation of Mohanram's lowest book-to-market quintile; a stock without a P/B is not scoreable). Only then do five binary criteria count, one point each: return on assets (ROA) above the industry median, cash-flow return on assets (CFROA) above the industry median, operating cash flow above net income (earnings quality), R&D intensity above the industry median and investment intensity (CapEx relative to total assets) above the industry median.
The investment criteria are the clou of the model: high R&D and CapEx spending depresses current profit, conservative accounting therefore punishes exactly the companies investing most in their future. The G-Score turns that around and treats above-average investment as a quality signal. The upper threshold is a uniform 4 points across all size classes (Mohanram's paper finds size irrelevant); the lower threshold is 1 point, closer to the original (0 to 1 of 8 criteria as a short candidate) than the previous threshold of 2.
Growth gate (exclusion rule): P/B from 3, an approximation of the original's lowest book-to-market quintile.
Profitability: (1) ROA above the industry median, (2) CFROA above the industry median.
Earnings quality: (3) operating cash flow above net income.
Investment: (4) R&D spending relative to total assets above the industry median, (5) CapEx relative to total assets above the industry median.
Documented deviations: five of the eight original criteria are implementable, advertising intensity and the variances of returns and sales are dropped for lack of data; the thresholds are scaled from 8 to 5 criteria accordingly. Comparisons run against industry medians, as in the original: the median is robust against single outliers that would distort a mean in skewed distributions (such as R&D intensity).
Fidelity update as of 23 August 2026: the growth gate (P/B from 3) bounds the universe on the lowest book-to-market quintile as in the paper, the lower threshold is lowered from 2 to 1 point (closer to Mohanram's own 0-to-1-of-8 short candidacy), and the cap thresholds are unified because size plays no role in the paper.
In the backtest, the industry-relative criteria are evaluated on each rebalancing date against the industry median of the stocks that belonged to the chosen universe at that time, using the figures published by then. Two caveats: sector assignment has only been historised since September 2026 (earlier periods use the sector recorded when tracking began), and industries with fewer than 20 stocks in the universe count as missing. The rule-coverage panel in the backtest scales the thresholds proportionally for missing values and discloses this transparently.
For growth investors as a quality filter over the expensive market segment, ideal combined with a momentum or Rule-of-40 view that measures the growth itself.