How to Use the Mohanram G-Score to Identify Strong Growth Stocks

The G-Score is a financial analysis model developed by Professor Partha S. Mohanram to identify potential winners and losers among growth stocks (i.e., low book-to-market firms). The G-Score uses eight financial signals to evaluate the strength of growth companies' fundamentals and identify those most likely to outperform or underperform in the future.

This article will explain how the G-Score works, how to calculate it, and how to interpret the results. We'll cover each of the eight financial variables used in the model and explain the logic behind them. We'll also provide a real-world example and discuss the research supporting the model's effectiveness in identifying high-performing growth stocks. Lastly, we'll examine the model's limitations and practical considerations for its application in investment strategies.

By learning how to calculate and interpret the G-Score, investors can better identify promising growth stocks, avoid potential underperformers, and make more evidence-based investment decisions.

G-Score Explained

The G-Score is a mathematical model that uses eight financial variables to identify which growth stocks are likely to outperform or underperform in the future. The score is calculated by assigning binary values (0 or 1) to each of eight financial signals and then summing them up, resulting in a score between 0 and 8. Higher scores suggest stronger growth fundamentals and better future performance potential.

Developed by Partha S. Mohanram, an Associate Professor then at Columbia Business School (now at the University of Toronto's Rotman School of Management), the model was first presented in his 2004 paper "Separating Winners from Losers among Low Book-to-Market Stocks using Financial Statement Analysis." The study analyzed U.S. stock market data from 1979 to 1999 (21 years) to develop and test the model.

To understand the G-Score, we must first understand the book-to-market (B/M) ratio, which compares a company's book value to its market value:

Book-to-Market (B/M) Ratio = Book Value of Equity (aka Shareholders' Equity) / Market Value of Equity (aka Market Capitalization)

Notably, the B/M ratio is often used to categorize stocks into two broad investment styles:

• Growth Stocks (Low B/M): Companies valued much higher than their book value due to expected future growth potential.
• Value Stocks (High B/M): Companies valued closer to or below their book value, potentially indicating undervaluation.

The G-Score was developed against the backdrop of extensive research on the "book-to-market effect" in stock returns. This effect, documented by Fama and French (1992) and Lakonishok, Shleifer, and Vishny (1994), shows that value stocks tend to outperform growth stocks over time.

While most financial analysis frameworks like Piotroski's F-Score work better for value stocks, the G-Score takes a fundamentals-based approach to analyzing growth stocks, which traditionally have been evaluated more on future potential rather than current performance.

Mohanram recognized that conventional financial statement analysis required modification to be effective for growth stocks, focusing on three key aspects:

1. Traditional fundamentals: Mohanram found that current profitability and cash flow generation matter even for growth companies. Firms with stronger return on assets (ROA) and cash flows can better maintain competitive advantages and fund future growth without shareholder dilution.
2. Naive extrapolation: Markets often project recent performance too far into the future. Mohanram's model identifies companies with stable earnings and sales growth patterns, which are less likely to disappoint investors who have extrapolated temporary high performance.
3. Accounting conservatism: Growth companies often invest heavily in R&D, advertising, and capital expenditures (CapEx) that accounting rules require to be immediately expensed rather than capitalized. Mohanram's model rewards these investments as they create "hidden reserves" that signal future growth potential not reflected in current book values.

By combining these three perspectives, Mohanram created a scoring system that captures both current performance and future growth potential.

G-Score Evidence and Performance

Mohanram's research demonstrated that the G-Score effectively identifies winners and losers among growth stocks.

His study found that growth stocks (i.e., low book-to-market firms) underperformed compared to similar-sized companies, with negative returns of -6.0% in the first year and -4.2% in the second year following portfolio formation (which occurs four months after fiscal year end).

However, when these companies were sorted using the G-Score, a clear and consistent pattern emerged that separates potential winners from losers.

Overall Performance

Companies with high G-Scores (6-8) performed much better, earning positive returns of 3.3% and 2.4% above their size benchmarks in years one and two.

In contrast, companies with low G-Scores (0-1) performed much worse, with returns of -17.9% and -13.3% below their benchmarks. Mohanram called these poorly performing stocks "torpedo stocks" because they could severely damage portfolio performance.

The difference between high and low G-Score companies created a significant investment opportunity, with a return gap of 21.2% in the first year and 15.8% in the second year.

Performance Across Different Market Segments

What makes the G-Score particularly valuable is how consistently it worked across many different types of companies:

• Company size: Mohanram divided firms into three equal groups based on market capitalization. Among small companies, high G-Score firms (6-8) outperformed low G-Score firms (0-1) by 22.8%. For medium-sized companies, the performance gap was 23.1%. Even for large companies, high G-Score firms beat low G-Score firms by 19.8%.
• Analyst coverage: For companies that no analysts followed, high G-Score firms outperformed low G-Score firms by 21.3%. The performance gap was 16.3% for companies with limited analyst following (below industry median) and 18.1% for widely-followed companies (at or above industry median).
• Stock exchanges: On the NYSE and AMEX exchanges, high G-Score companies beat low G-Score companies by 12.7%. On the NASDAQ, the performance difference was even larger at 26.4%. Mohanram noted that NASDAQ firms contained more potential "torpedo stocks," although this may no longer be true today.
• Company age: The strategy worked for both newly public companies (IPOs within one year of portfolio formation) and established firms.
• Growth rates and industries: For fast-growing companies (those with above-median sales growth in their industry), high G-Score firms outperformed low G-Score firms by 20.3%. For slower-growing companies, this gap was 18.7%. Among technology companies, the performance difference was 17.8%.

Perhaps most impressively, the G-Score maintained its effectiveness throughout the entire study period. The strategy delivered positive returns in all 21 years of Mohanram's research (1979-1999). In 16 of those years, the performance gap between high and low G-Score portfolios was statistically significant, showing this wasn't just due to chance.

Beyond Risk Factors

Some finance experts might argue that these return differences simply reflect compensation for taking higher risk (as traditional finance theory suggests that higher returns come from riskier investments), but Mohanram's analysis disproved this theory.

The G-Score's effectiveness held up even after controlling for well-known risk factors like size, book-to-market, momentum, and accruals. After accounting for these factors, each one-point increase in G-Score was associated with a 3.7% increase in returns.

Notably, the G-Score wasn't simply identifying riskier stocks. High G-Score firms (6-8) had similar levels of market risk (beta) compared to low G-Score firms (0-1). In fact, low G-Score firms had higher return volatility but still performed worse—the opposite of what traditional risk-return theory would predict.

Connection to Future Financial Performance

What explains these return differences? The answer lies in how well the G-Score predicted actual business performance:

• High G-Score firms had much better future profitability (11.9% ROA vs. -8.5% for low G-Score firms).
• High G-Score firms were far less likely to delist due to poor performance (0.3% vs. 7.4%).
• High G-Score firms were more likely to beat analyst earnings forecasts. Across the four quarters following portfolio formation, high G-Score firms surprised analysts more positively than low G-Score firms by 1.23%.

These findings suggest that the market didn't fully understand how current financial information predicted future performance, creating an opportunity for investors using the G-Score to gain an advantage.

The Eight G-Score Signals Explained

The G-Score is calculated by assigning binary values (0 or 1) to eight financial signals and then summing them:

G-Score = G1 + G2 + G3 + G4 + G5 + G6 + G7 + G8

where:

• G1 =1 if ROA > Ind. Median ROA
• G2 = 1 if CFROA > Ind. Median CFROA
• G3 = 1 if CFO > NI
• G4 =1 if VARROA < Ind. Median VARROA
• G5 =1 if VARSGR < Ind. Median VARSGR
• G6 = 1 if RDINT > Ind. Median RDINT
• G7 = 1 if CAPINT > Ind. Median CAPINT
• G8 = 1 if ADVINT > Ind. Median ADVINT

These signals were specifically chosen to address three key aspects of growth stock analysis: current profitability, stability of performance, and investment in future growth.

The G-Score ranges from 0 (weakest fundamentals) to 8 (strongest fundamentals). The G-Score gives equal weight to each signal, making it straightforward to calculate and interpret.

Category #1: Signals Based on Earnings and Cash Flow Profitability (G1-G3)

The first three signals focus on a firm's current profitability measured by earnings and cash flows.

Signal #1: Return on Assets (G1)

Return on assets (ROA) measures how efficiently a company uses its assets to generate profits. Even for growth companies, current profitability matters. Firms that can generate higher returns on their assets demonstrate superior operational efficiency and management capability.

ROA is calculated and scored as follows:

ROA = Net Income Before Extraordinary Items / Beginning Total Assets

Scoring: G1 =1 if ROA > Ind. Median ROA

For example, a software company with an ROA of 15% compared to its industry median of 8% would score a point for this signal, suggesting it converts its assets to profit more efficiently than peers.

When comparing ROA against industry peers, this signal helps control for industry-specific factors that might affect profitability levels, identifying companies with sustainable competitive advantages that often lead to stronger future performance.

Signal #2: Cash Flow ROA (G2)

Cash flow ROA examines profitability from a cash flow perspective, which can be more meaningful than earnings-based measures for growth companies, particularly early-stage ones.

These firms often have large non-cash expenses like depreciation and amortization (D&A) due to significant investments in fixed and/or intangible assets.

Cash flow ROA is calculated and scored as follows:

Cash Flow ROA = Cash Flow from Operations / Beginning Total Assets

Scoring: G2 = 1 if CFROA > Ind. Median CFROA

This signal complements G1 by examining if reported profitability is backed by actual cash generation rather than accounting effects.

Companies with above-median cash flow ROA demonstrate an ability to convert their business model into real cash, providing greater financial stability and reducing dependence on external financing.

Signal #3: Cash Flow Greater Than Net Income (G3)

This signal examines the relationship between a company's reported earnings and its cash generation, based on the "accrual anomaly" documented by Sloan (1996), which shows that firms with a higher proportion of accruals in their earnings generally underperform in the future.

Here's how G3 is scored:

Scoring: G3 = 1 if CFO > NI

When cash flow exceeds net income (negative accruals), it suggests the company has higher-quality earnings. For example, if a company reports $10 million in net income but generates $12 million in operating cash flow, this indicates conservative accounting practices.

Conversely, when earnings exceed cash flow, it could signal aggressive revenue recognition (i.e., recognizing sales too early), delayed expense recognition, or other accounting choices that inflate current earnings. These practices often lead to earnings disappointments in future periods when the accruals reverse.

Growth companies often have large negative accruals due to their rapid expansion, so this signal helps identify those with the healthiest relationship between reported earnings and actual cash generation.

Category #2: Signals Related to Naive Extrapolation (G4-G5)

These signals address the market's tendency to naively extrapolate current performance into the future.

Signal #4: Earnings Variability (G4)

Earnings variability measures the consistency of a company's profitability over time. This signal identifies firms with stable earnings patterns, as stable earnings suggest that current strong performance is more likely to persist rather than being a lucky one-time occurrence.

Earnings variability is measured and scored as follows:

Earnings Variability = Variance of ROA over past 3-5 years

Scoring: G4 =1 if VARROA < Ind. Median VARROA

For example, a company with ROA of 12%, 11%, and 13% over three years would have much lower variability than one with ROA of 20%, 5%, and 15%. The first company's performance is more predictable and likely sustainable.

The market often overvalues firms with temporary earnings spikes and undervalues those with consistent performance. Lower earnings variability suggests that a company operates in a more predictable business environment or has better risk management practices, reducing the likelihood of negative earnings surprises.

Note: Minimum 3 years of data required. Companies with less history receive a score of 0.

Signal #5: Sales Growth Variability (G5)

Sales growth variability examines the consistency of a company's revenue expansion. Similar to earnings stability, stable sales growth indicates that a company's expansion is more predictable and sustainable.

Sales growth variability is measured and scored as follows:

Sales Growth Variability = Variance of Annual Sales Growth over past 3-5 years

Scoring: G5 =1 if VARSGR < Ind. Median VARSGR

This signal focuses on sales rather than earnings because sales growth is typically more persistent and less affected by accounting choices, as observed in Damodaran's 2001 book "The Dark Side of Valuation."

Growth companies with consistent sales increases demonstrate more reliable business expansion and market penetration. In contrast, firms with erratic growth patterns often disappoint investors who extrapolated from temporary high growth periods.

Note: Minimum 3 years of data required. Companies with less history receive a score of 0.

Category #3: Signals Related to Accounting Conservatism (G6-G8)

These signals identify firms making growth-oriented investments that may depress current earnings but fuel future growth.

Signal #6: R&D Intensity (G6)

R&D intensity measures a company's investment in research and development relative to its size.

R&D intensity is calculated and scored as follows:

R&D Intensity = R&D Expenses / Beginning Total Assets

Scoring: G6 = 1 if RDINT > Ind. Median RDINT

Under conservative accounting principles, R&D expenditures are immediately expensed rather than capitalized. This reduces current earnings and book values even though these investments often create long-term value not reflected in financial statements.

This signal is particularly important for technology, pharmaceutical, and biotechnology companies where R&D drives future product development

High R&D intensity relative to industry peers indicates greater investment in innovation. Research by Lev and Sougiannis (1996) has shown that firms spending heavily on R&D tend to earn excess returns in subsequent periods. These investments can drive sales and earnings growth, helping companies meet or exceed market expectations.

Signal #7: CapEx Intensity (G7)

CapEx intensity measures a company's investment in CapEx (capital expenditures) relative to its size. CapEx refers to money spent on physical assets like buildings, equipment, and machinery.

Unlike R&D (G6), which focuses on innovation and new product development, this signal concentrates on investment in tangible, long-term assets.

CapEx intensity is calculated and scored as follows:

CapEx Intensity = CapEx / Beginning Total Assets

Scoring: G7 = 1 if CAPINT > Ind. Median CAPINT

Industries like manufacturing, telecommunications, energy, and utilities typically have high CapEx needs. While CapEx is capitalized rather than expensed, it still represent a commitment to future growth that may not be fully valued by the market.

It's worth noting that high CapEx spending isn't always positive - in some cases, it could indicate overinvestment or poor capital allocation.

However, for growth companies investing strategically, above-industry CapEx spending may indicate positioning for future growth by expanding production capacity or improving infrastructure in ways that will generate future returns.

Signal #8: Advertising Intensity (G8)

Advertising intensity measures a company's investment in marketing and brand-building relative to its size. Like R&D, advertising expenditures are immediately expensed despite their potential to create long-term value.

Advertising intensity is calculated and scored as follows:

Advertising Intensity = Advertising Expenses / Beginning Total Assets

Scoring: G8 = 1 if ADVINT > Ind. Median ADVINT

This signal is particularly relevant for consumer products companies, retail businesses, and service providers where brand recognition drives customer acquisition and loyalty.

High advertising intensity can indicate investment in market share and brand equity, creating "hidden reserves" not reflected on the balance sheet.

Research by Chan, Lakonishok, and Sougiannis (2001) found these investments often lead to stronger future performance through improved customer loyalty, premium pricing ability, and market share gains.

Note: Advertising data can be less consistently reported than R&D or CapEx, as some companies classify marketing expenses differently.

Interpreting G-Score Ranges

The G-Score adds up eight financial signals to create a total score between 0 and 8, with higher scores suggesting better growth fundamentals. Based on Mohanram's research, we can interpret these scores in three main groups:

• Low G-Score (0-1): Companies in this range show the weakest growth fundamentals. These firms experienced very poor future performance, with mean size-adjusted returns of -17.9% in the first year and -13.3% in the second year after portfolio formation. About 7.4% of these companies delisted due to poor performance.
• Medium G-Score (2-5): For the scores between these groups (2-5), Mohanram observed an almost perfect monotonic relationship - as G-Score increased, returns steadily improved. While he doesn't explicitly categorize these middle scores as a separate group, the data shows firms with scores of 2-3 still earned negative returns, while those with scores of 4-5 performed closer to market averages.
• High G-Score (6-8): These companies demonstrated the strongest growth fundamentals. They earned positive size-adjusted returns of 3.3% in the first year and 2.4% in the second year. High G-Score firms had very low delisting rates (only 0.3%) and showed stronger future earnings performance with much better ROA than low G-SCORE firms.

Overall, the most effective investment strategy based on the G-Score was to go long on firms with high G-Scores (6-8) and short firms with low G-Scores (0-2). This strategy produced a size-adjusted return difference of 21.2% in the first year after portfolio formation and 15.8% in the second year.

Notably, the G-Score's effectiveness was consistent across different market segments. It worked for both large and small firms, companies with and without analyst coverage, and across different time periods. The strategy earned positive returns in all 21 years of Mohanram's study (1979-1999).

G-Score Calculation Example

To calculate the Mohanram G-Score for a real-world company, here are the steps investors can follow:

• Step #1: Group companies by industry: First, identify companies in the same industry using 2-digit SIC codes (Standard Industrial Classification).
• Step #2: Identify low B/M ratio stocks: Within each industry group, select companies with the lowest B/M ratios (typically the bottom ~20%).
• Step #3: Access financial data:
  • If using a financial database: Collect the financial metrics for your target company and all other low B/M companies in the same industry.
  • If collecting data yourself: Collect 3-5 years of financial statements (income statement, balance sheet, and cash flow statement) for your target company and all other low B/M companies in the same industry.
• Step #4: Calculate financial metrics (if necessary): If you don't have access to a financial database, calculate/find these metrics for your target company and all the industry peers:
  • Return on assets (ROA)
  • Cash flow return on assets (CFROA)
  • Cash flow from operations (CFO) and net income (NI)
  • Earnings variability (VARROA; over 3-5 years)
  • Sales growth variability (VARSGR; over 3-5 years)
  • R&D intensity (RDINT)
  • CapEx intensity (CAPINT)
  • Advertising intensity (ADVINT)
• Step #5: Determine industry benchmarks: Calculate/find the median value for each metric among all the low B/M companies in the same industry.
• Step #6: Assign values for each signal: Compare your target company to the industry medians. For each signal, assign 1 point if the criterion is met and 0 if not.
• Step #7: Calculate and interpret the G-Score: Add up all eight values to get the final score (0-8). Scores of 0-1 suggest high risk of underperformance, while scores of 6-8 indicate better potential returns.

Reference the Excel model linked to this article to follow our example below, where we'll calculate and interpret the G-Score over the last three fiscal years.

Step #1: Group Companies by Industry

The first step in calculating the G-Score is to group companies by industry. This is important because the G-Score compares a company's financial metrics to its industry peers. Since seven of the eight G-Score signals involve industry-relative measures, we need to ensure we're making appropriate comparisons.

You can find industry codes (called SIC codes) on the NAICS Association website. Each company has a four-digit SIC code, but for G-score calculations, we only need the first two digits to group similar businesses.

For this example, we'll focus on "SIC 20 - Food and Kindred Products." This includes companies that manufacture or process foods and beverages.

After choosing your industry, identify all publicly traded companies in that group. We used Fintel to find all stocks with the SIC code 20, then exported this list to Excel for filtering.

To match Mohanram's original study, we kept only U.S. stocks (from any exchange). We also removed several types of securities that would skew our results:

• Preferred shares: Have different characteristics than common stock.
• Depositary shares: Represent foreign companies trading on U.S. exchanges.
• Equity warrants: Derivative securities, not operating companies.
• ETFs: Investment funds, not individual companies.
• Different share classes of the same company: To avoid counting the same company twice.
• Stocks trading below $1: Penny stocks often have less reliable financial information.
• Stocks with market caps below $1 million: Micro-cap stocks typically have limited liquidity.

After applying these filters, we had a list of 69 SIC 20 companies for our industry comparison group.

Step #2: Identify Low B/M Ratio Stocks

After creating our industry group of 69 Food and Kindred Products companies, we needed to identify which ones are growth stocks.

Following Mohanram's approach, we calculated the book-to-market (B/M) ratio for each company (shareholders' equity / market capitalization). We used trailing twelve months (TTM) data to get the most current financial picture.

Then, we ranked all 69 companies by their B/M ratios and selected the lowest 20%. This gave us 13.8 companies, which we rounded up to 15 companies. These 15 companies represent our "growth stock" subset within the Food and Kindred Products industry.

We included companies with negative B/M ratios, just as Mohanram did in his original study. Negative ratios occur when a company has negative book value, usually because their accumulated losses exceed their equity. These companies often have market values that far exceed what their accounting numbers would suggest.

Here's our final list of 15 companies:

Values in USD Millions
Data Collected: 05/15/2025
Data Source: SEC 10-K Annual Reports

For our G-Score calculation example, we selected Celsius (CELH), a global beverage company that makes lifestyle energy drinks. CELH falls within the lowest ~20% of B/M ratios in the SIC 20 group, making it a suitable candidate for G-Score analysis.

Note: The G-Score only works for companies already in the lowest 20% B/M ratio of their industry. It was designed to find winners and losers among existing growth stocks, not to identify which companies are growth stocks in the first place.

Step #3: Access Financial Data

After identifying our industry group and target company, we need to gather financial information for both Celsius and the other 14 companies in our low B/M group.

For each company, we need the income statement, balance sheet, and cash flow statement, all covering at least 3-5 years of the most recent historical financial data.

Our data came from SEC filings (i.e., 10-K annual reports). Investors can also use QuickFS, Yahoo Finance, or financial databases like Blooomberg, FactSet, or S&P Capital IQ.

However, when using stock market data websites, you'll likely need to read the notes to the financial statements to find accurate numbers on R&D and advertising spend.

From these financial statements, we need to find:

• Income Statement:
  • Revenues (for calculating sales growth and variance (VARSGR))
  • Research & development expenditures (RDINT)
  • Advertising expenditures (ADVINT)
  • Net income (NI)
• Balance Sheet:
  • Total assets (TA)
• Cash Flow Statement:
  • Cash flow from operations (CFO)
  • Capital expenditures (CAPINT)

We collected data from FY2019-2024 (6 years) to properly calculate the variance-based signals (G4 and G5) and to assess how the G-Score changed over the most recent 3 fiscal years.

Step #4: Calculate Financial Metrics

With financial data gathered for Celsius and our industry peers, we now need to calculate the eight financial metrics used in the G-Score. These calculations will let us compare Celsius to other growth companies in its industry.

The table below provides an overview of the G-Score signals and how to calculate them, with our Celsius calculations for FY2022-2024 (the last 3 fiscal years):

Reference the list below to ensure you're performing these calculations accurately:

• Net income before extraordinary items: Mohanram uses this for ROA because it excludes unusual one-time events (e.g., restructuring costs, asset write-downs, legal settlements, gains from selling divisions). In most cases, you can simply use "net income" as extraordinary items are relatively rare in modern financial reporting.
• Beginning total assets: Use the total assets figure from the start of the year (end of previous year), not the current year's total assets or an average of beginning and ending assets.
• Variance calculation: For earnings and sales growth variability, use the VAR.S function in Excel (sample variance) on the past 3-5 years of data. VAR.S is appropriate because we're working with a sample of years, not the entire population (VAR.P). If a company doesn't have at least 3 years of data, assign a value of 0 for these signals.
• Missing data: If R&D or advertising expenses aren't separately reported (check financial statement footnotes), use zero for that metric. For some companies, you can also make a reasonable estimate (e.g., by using a percentage of the total SG&A expense).

After completing these calculations for Celsius and all other industry peers, the next step is to calculate the industry median values for each metric (except G3, which is a direct comparison not based on industry medians).

Step #5: Determine Industry Benchmarks

After calculating financial metrics for all 15 companies in our low B/M food industry group, we need to establish proper benchmarks for comparison. The G-Score method requires comparing a company's metrics to similar growth companies in the same industry, which is why we use median values for our low B/M group.

As previously discussed, seven of the eight G-Score signals require industry comparisons. These benchmarks let us evaluate whether a company is stronger or weaker than its peers on each metric. Using industry-specific benchmarks accounts for the fact that "normal" profitability, investment amounts, and earnings stability vary greatly across different industries.

For each financial metric (except G3), we calculate the median value using the MEDIAN function in Excel. This function arranges all values from the 15 companies in order from lowest to highest and selects the middle value.

We use median rather than average (mean) values because medians aren't as affected by extreme outliers, which are common in samples of growth companies.

Here are the median values we calculated for our SIC 20 industry group across the last 3 fiscal years:

These medians represent the middle point of our Food and Kindred Products industry growth stocks. They become the benchmarks against which we'll compare Celsius in the next step to determine its G-Score.

Step #6: Assign Values for Each Signal

With industry benchmarks established, we can now determine which of the eight G-Score signals Celsius meets. For each signal, we compare Celsius' metrics to the appropriate benchmark and assign either 1 point (if the criterion is met) or 0 points (if not).

We performed this analysis for Celsius across three fiscal years (FY2022-2024), but we'll focus on FY2024 scorings:

For profitability and investment signals (G1, G2, G6, G7, G8), higher values are considered better, so the company scores 1 when its value exceeds the industry median.

For stability signals (G4, G5), lower variability is considered better, so the company scores 1 when its value is below the industry median.

For G3, the company simply needs to have cash flow from operations greater than net income to score 1 point.

Step #7: Calculate and Interpret the G-Score

For Celsius' FY2024, if we add up the points from all eight signals:

G-Score (CELH: FY2024) = 1 (G1) + 1 (G2) + 1 (G3) + 0 (G4) + 0 (G5) + 1 (G6) + 0 (G7) + 1 (G8) --> 5 (out of 8)

Celsius achieves a G-Score of 5 for FY2024, up from a score of 4 in both FY2022 and FY2023.

As previously explained, based on Mohanram's research, G-Scores can be interpreted as follows:

With a G-Score of 5, Celsius sits at the top end of the middle range, showing moderate-to-good growth fundamentals compared to its industry peers. According to Mohanram's findings, companies with mid-range G-Scores have mixed performance prospects, though performance tends to improve as scores increase.

Celsius shows strong cash flow characteristics (G2, G3), with cash flow ROA exceeding the industry median and cash flow exceeding net income. It also has above-average profitability (G1) with ROA exceeding industry median. The company demonstrates commitment to future growth through above-average marketing investment (G8) and technically above-average R&D intensity (G6).

The median R&D expenditure in SIC 20 is nearly zero (0.01%), as food companies typically focus on product development rather than formal R&D. Celsius' R&D spending was only marginally higher at 0.03%, but this was enough to score a point.

However, Celsius falls short in some key areas. Its CapEx intensity (G7) is slightly below industry standards (1.5% vs. industry median of 1.7%). Most notably, Celsius shows much higher variability in both earnings and sales growth compared to peers. Its ROA variability was 18.3% versus an industry median of just 0.5%, and its sales growth variability was 34.8% compared to an industry median of 1.7%.

These high variance scores are particularly concerning based on Mohanram's findings. His research showed that companies with unstable earnings and sales growth are more likely to disappoint investors in the future. The high variability suggests that Celsius' strong current performance might not be as predictable or sustainable as investors hope, potentially making it more vulnerable to future underperformance if its growth pattern becomes less favorable.

It's worth noting that we used 3-year data for our variance calculations. A 5-year analysis might have yielded different results, possibly showing more or less stability over a longer timeframe.

Regardless, the G-Score of 5 in FY2024, up from 4 in prior years, indicates some improvement in Celsius' financial characteristics relative to its industry peers.

Limitations of the G-Score

While Mohanram's G-Score can help identify winning and losing growth stocks, investors should understand its specific shortcomings before applying it to investment decisions. These limitations affect the score's reliability and usefulness in different market conditions:

1. Basic design limitations: The G-Score gives equal weight to all eight signals despite some being more effective than others. For example, G7 (CapEx Intensity) and G8 (Advertising Intensity) were demonstrably weaker predictors in Mohanram's research. Moreover, each measure is simplified to 0 or 1, effectively ignoring the degree of outperformance.
2. Data requirements: Signals G4 and G5 need at least three years of financial history, making the G-Score less effective for newer companies and IPOs. In Mohanram's study, no IPO firms scored higher than 6 on the G-Score due to these data requirements.
3. Industry comparison challenges: Many signals rely on industry peer comparisons, which becomes problematic when there aren't enough similar companies or when industry classifications are inaccurate. This is especially true for companies in industries heavily influenced by external factors rather than internal financial metrics, including oil and gas, technology, real estate, financial services, utilities, consumer staples, companies in transition, and businesses dependent on external conditions.
4. Growth stock definition problems: The study defines growth stocks simply as companies with the lowest book-to-market ratios, creating a mixed group that includes non-growth companies. This lack of homogeneity creates challenges for investors focused solely on growth firms, as portfolios may not align precisely with growth-focused strategies, requiring additional work to separate true growth firms from others.
5. Lack of qualitative factors: The G-Score examines only quantitative financial data and ignores important factors like management quality, competitive advantages, and market trends. This means it misses critical aspects that often drive a growth company's success.
6. Time and market evolution: The original research used data from 1979-1999. Today's growth companies have an increasing proportion of intangible-intensive operations. While the G-Score includes R&D and advertising signals, it may not fully capture how modern growth companies create value.
7. Fiscal reporting differences: Companies report financial results at different times during the year, potentially creating look-ahead bias. High performing and low performing groups might have unequal average compounding periods, which could skew results.
8. Portfolio concentration risk: Using the G-Score might concentrate investments in smaller companies, less-established firms, those with limited analyst coverage, or thinly traded stocks, potentially increasing portfolio risk across changing market conditions.
9. Theoretical contradictions: The G-Score strategy shows that safer firms (high G-Score) outperform riskier ones (low G-Score), contradicting traditional theories that higher returns require higher risk. This suggests the strategy works due to market mispricing rather than risk factors, which might not persist if markets become more efficient. The success of fundamental analysis for growth firms is also surprising given that current financials often don't predict future performance for such companies.
10. Limited return adjustment methodology: The study primarily used size-adjusted returns rather than the more complete Fama-French multifactor models. This means other important risk factors (like value, momentum, or quality) might explain some of the performance differences attributed to the G-Score.

Overall, the G-Score works best when used alongside other analytical methods rather than as a standalone strategy. Combining it with qualitative analysis and industry-specific considerations gives a more complete picture of a growth stock's potential and helps avoid the risks of relying too heavily on any single quality score method.

Practical Applications for Investors

Despite its limitations, the G-Score remains a valuable scoring model for investors interested in growth stocks. Here are some practical guidelines for applying the G-Score:

• Use as a screening tool: Start with the G-Score to identify promising growth stocks and filter out potential underperformers. Focus on avoiding companies with very low scores (0-2), which Mohanram found were most likely to perform poorly.
• Apply industry context: Calculate G-Score signals by comparing companies to appropriate industry peers. Different industries have unique financial characteristics that affect how you should interpret the results.
• Consider time horizon: The G-Score works best for medium-term investing. Mohanram's research showed the biggest performance differences in the first and second years after forming portfolios.
• Build better portfolios: The G-Score can improve various investment approaches:
  • Long-only investors should focus on high G-Score growth stocks.
  • Long-short investors can buy high G-Score stocks while selling low G-Score stocks.
• Update regularly: Recalculate the G-Score annually since a company's fundamentals change over time. Today's high-scoring company might become tomorrow's low scorer.
• Works across company sizes: Mohanram found the strategy effective for companies of all sizes. High G-Score companies outperformed low G-Score companies by 19.8% for large firms, 23.1% for medium firms, and 22.8% for small firms.

By examining profitability, earnings stability, and investment patterns, investors can make more informed decisions about which growth companies have stronger fundamentals. While originally tested on data from 1979-1999, the G-Score provides a systematic method that helps investors look beyond market hype to evaluate the financial strength of growth companies.

The Bottom Line

The Mohanram G-Score provides a structured approach for analyzing growth stocks (i.e., low book-to-market firms) through fundamental financial data. It combines standard profit measurements with signals addressing naive extrapolation (the tendency to assume recent performance will continue indefinitely) and accounting conservatism (when companies expense growth investments immediately rather than spreading costs over time).

Mohanram's research showed that companies with high G-Scores (6-8) earned positive size-adjusted returns of 3.3%, while companies with low G-Scores (0-1) lost 17.9%, creating a performance gap of 21.2% in the year following analysis. Mohanram called these low-scoring companies "torpedo stocks" because they could suddenly and dramatically underperform, severely damaging portfolio performance.

Notably, the strategy generated positive returns in all 21 years of the 1979-1999 study period. Even when controlling for known factors like size, book-to-market ratio, momentum, accruals, and equity offerings, the G-Score remained a strong predictor of performance.

The G-Score does have limitations worth considering. It gives equal weight to all signals despite varying effectiveness, uses a simplified binary scoring system, requires several years of financial history, and misses qualitative factors like management quality and competitive advantages, among other limitations.

The model is most useful as a screening tool rather than a complete investment strategy, with investors typically finding more value in avoiding low-scoring stocks (0-2) than identifying those with perfect scores. For best results, apply proper industry comparisons, recalculate scores annually as fundamentals change, and combine the G-Score with qualitative analysis for a more complete picture of a company's prospects.