Small Pond, Big Fish: Calls Claiming Fast Growth in a Tiny Market
This study examines 35,111 earnings calls — 21.3% of a 165,182-call corpus spanning 1990 to 2026 — where a language model answered YES to both "Early Products Growing Fast" and "A Tiny Fraction of the Market." These calls skew promotional: promotion scores 5.92 versus a 5.05 baseline (+0.87), and confidence runs +0.30 higher. They overrepresent scale-dependent advantage claims (2.15x) and deferred revenue growth (1.82x). Among 4,404 calls with return data, the median next-period return is -6.11% versus -7.16% for the 22,449-call base, and 42.6% beat expectations versus 39.5% at baseline — a modest difference, not an edge. The share of such calls peaked at 27.33% in 2021.
- Calls matching the pattern make up 21.3% of the 165,182-call corpus, with a 95% confidence interval of 21.1% to 21.5%.
- Promotion language scores 5.92 versus a 5.05 baseline (+0.87), the largest behavioral gap among the six measured traits.
- The pattern overrepresents Scale-Dependent Advantage Claims (2.15x) and Deferred Revenue Growing (1.82x), and underrepresents The Finished-Story Tell (0.70x).
- Among 4,404 calls with returns, the median next-period return is -6.11% versus -7.16% at baseline, and 42.6% beat expectations versus 39.5%.
1Introduction
Every earnings-call lexicon has its dream pairing: management describing an early product line growing fast while the addressable market remains a tiny fraction of the whole. It is the classic early-stage bull case — and also, on paper, one of the most abused framings in corporate storytelling, since "tiny market" conveniently makes any growth rate look impressive. For analysts who parse calls for tone and framing, understanding how often this combination appears, and how its language differs from the typical call, is a useful calibration exercise. This study measures how frequently the pairing occurs across 165,182 calls from 1990 to 2026, how its language profile, guidance behavior, and topic mix differ from the corpus, and how subsequent results compare.
2Data & methodology
The corpus comprises 165,182 earnings-call transcripts published between 1990 and 2026, each scored independently by a large language model on an identical 37-field battery: seven categorical business verdicts, eight 0–9 behavioral meters, and twenty yes/no judgments. The study group is defined as calls where the model answered YES to "Early Products Growing Fast" AND the model answered YES to "A Tiny Fraction of the Market" (n = 35,111; 21.3% of the reference set, 95% Wilson interval 21.1%–21.5%). Baseline figures use all scored calls. Market outcomes join a fixed sample of 22,449 calls with twelve-month total returns in excess of SPY, measured from the first close after each call; this sample skews toward liquid U.S. names and is reported as descriptive history only.
3Results
The behavioral profile is distinctly promotional: promotion scores 5.92 against a 5.05 baseline (+0.87) and confidence 7.51 versus 7.21 (+0.30), while candor is slightly lower (6.59 versus 6.86). Guidance is raised on 26.5% of these calls versus 21.1% at baseline, and guidance is less often lowered (9.7% versus 11.6%). Topic lifts point to scale narratives: Scale-Dependent Advantage Claims appear 2.15x more often and Deferred Revenue Growing 1.82x more often. The annual share climbed from 17.51% in 2016 to a peak of 27.33% in 2021, easing to 19.26% in 2025. Returns are not clearly favorable: the median next-period return of -6.11% is only modestly above the baseline -7.16%, with a wide interquartile range from -31.58% to +18.23%.
| Meter | Study group | Baseline | Δ |
|---|---|---|---|
| Candor | 6.59 | 6.86 | -0.27 |
| Evasion | 2.80 | 2.70 | +0.10 |
| Specificity | 7.45 | 7.56 | -0.11 |
| Stress | 2.41 | 2.43 | -0.02 |
| Promotion | 5.92 | 5.05 | +0.87 |
| Confidence | 7.51 | 7.21 | +0.30 |
| Action | Study group | Baseline |
|---|---|---|
| Raised | 26.5% | 21.1% |
| Maintained | 44.5% | 48.8% |
| Lowered | 9.7% | 11.6% |
| Withdrawn | 2.0% | 2.7% |
| Signal | Lift | In group | Baseline |
|---|---|---|---|
| Scale-Dependent Advantage Claims | 2.15× | 23.8% | 11.1% |
| Deferred Revenue Growing | 1.82× | 16.1% | 8.9% |
| Founder-Led Companies | 1.72× | 34.5% | 20.0% |
| Volume About to Step Up | 1.64× | 46.6% | 28.5% |
| The Hidden Segment | 1.51× | 31.9% | 21.1% |
| When the CFO Dominates | 0.65× | 9.2% | 14.1% |
| The Finished-Story Tell | 0.70× | 3.1% | 4.4% |
| Pricing Recovering | 0.70× | 15.2% | 21.5% |
| Statistic | Study group | Returns sample |
|---|---|---|
| Median excess return | -6.1% | -7.2% |
| Interquartile range | -31.6% to +18.2% | — |
| Share beating SPY | 42.6% (95% CI 41%–44%) | 39.5% |
| Observations | 4,404 | 22,449 |
| Ticker | Quarter | Call date | Call grade |
|---|---|---|---|
| FRST | Q2 2025 | 2025-07-25 | A |
| AN | Q2 2025 | 2025-07-25 | B |
| CHTR | Q2 2025 | 2025-07-25 | C+ |
| DBOEY | Q2 2025 | 2025-07-25 | B+ |
| LBTSF | Q2 2025 | 2025-07-25 | C+ |
| ASPS | Q2 2025 | 2025-07-25 | D |
| AAFRF | Q1 2026 | 2025-07-25 | B |
| PHPPY | Q2 2025 | 2025-07-25 | C |
4Discussion
A careful reader should conclude that this framing is common — roughly one call in five — and that it travels with more promotional, confident language and scale-advantage rhetoric. The slightly better median return and beat rate (42.6% versus 39.5%) are descriptive comparisons in an observational sample; they do not show that the pattern predicts outcomes, and the wide return dispersion means individual outcomes vary enormously. Nothing here establishes that the language caused anything, or that the gap would survive costs, timing choices, or different sampling. Treat the profile as context, not a signal.
5Limitations
The YES/NO fields are produced by a language model and are noisy; topic and trait labels inherit that noise. The returns sample covers 4,404 of these calls against a 22,449-call baseline and is skewed toward liquid names, so return comparisons may not generalize. Our own forward tests of similar patterns failed to show directional prediction. Additionally, language models partially remember famous stocks' histories, which can contaminate any apparent backtest performance in corpora that include well-known companies. See the full methodology, including the C1 pattern’s forward-test failure and the LLM-memorization finding.