Research › Hypotheses Tested
Artul.ai Research LibraryStudy No. 36Hypotheses TestedUpdated 2026-08-28

Light at the End of the Tunnel Is Often a Train: Recovery Talk on Calls

By Artul.ai Research Group · n = 215 earnings calls · First published 2026-08-28
Abstract

This study examines 215 earnings calls from a 500-call corpus (2015-2024) that answered YES to the research hypothesis "Coming out of the tunnel" - management claims that the company is emerging from a difficult period. These calls show elevated stress (2.97 vs 2.71), lower confidence (6.88 vs 7.33), and lower specificity (7.54 vs 7.65) than the baseline. Management lowered guidance on 22.8% of these calls versus 13.2% of the corpus. Among 83 calls with post-call return data, the median return was -16.4% versus -10.5% for the 199-call base. The strongest behavior lift was "Results Worse Than Direction" at 1.64.

Key findings
  • Only 215 of 500 calls (43%) answered YES to the "Coming out of the tunnel" hypothesis.
  • These recovery-narrative calls show stress of 2.97 versus 2.71 at baseline, a +0.58 delta.
  • Guidance was lowered on 22.8% of tunnel calls versus 13.2% of the broader corpus, a 1.72x ratio.
  • The 83 calls with return data had a median post-call return of -16.4% versus -10.5% for the base.

1Introduction

Few phrases on an earnings call carry more narrative weight than management declaring that the worst is over - that the company is "coming out of the tunnel." Investors must decide whether such language reflects genuine inflection or wishful framing. These moments matter because they often coincide with guidance decisions, analyst skepticism, and volatile post-earnings trading. This study examines 215 calls from a 500-call corpus spanning 2015-2024 that answered YES to the "Coming out of the tunnel" hypothesis, profiling their candor, stress, specificity, guidance behavior, and post-call returns against the broader corpus.

2Data & methodology

The corpus comprises 500 earnings-call transcripts published between 2015 and 2024, 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 that answered YES to the research hypothesis "Coming out of the tunnel" (n = 215; 43.0% of the reference set, 95% Wilson interval 38.7%–47.4%). Baseline figures use the set of calls on which this question was tested. 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 of tunnel calls is striking: stress runs 2.97 versus 2.71 and confidence 6.88 versus 7.33, suggesting the recovery talk coexists with visible strain. The lift table reinforces this - "Results Worse Than Direction" appears 1.64x more often than baseline (46.8% vs 23.6%), and "Underused Fixed Costs" 1.44x. Guidance cuts are the clearest divergence: 22.8% of tunnel calls lowered guidance versus 13.2% corpus-wide. The trend peaked at 0.32 in 2023 before falling to 0.11 in 2024. Returns add a sobering note: median -16.4% across 83 calls versus -10.5% for the base, with 38.6% beating versus 38.2% baseline.

Table 1. Mean behavioral scores (0–9 scale), study group versus baseline
MeterStudy groupBaselineΔ
Candor6.986.91+0.07
Evasion2.782.71+0.07
Specificity7.547.65-0.11
Stress2.972.39+0.58
Promotion4.955.11-0.16
Confidence6.887.33-0.45
Table 2. Guidance actions, study group versus baseline
ActionStudy groupBaseline
Raised11.6%21.8%
Maintained51.6%53.2%
Lowered22.8%13.2%
Withdrawn2.3%1.0%
Table 3. Co-occurring battery signals ranked by lift (group prevalence ÷ baseline prevalence)
SignalLiftIn groupBaseline
Results Worse Than Direction1.64×76.7%46.8%
Underused Fixed Costs1.44×56.3%39.0%
The Hidden Segment1.44×34.0%23.6%
Pricing Recovering1.37×24.2%17.6%
The Question Left Hanging1.30×55.8%42.8%
Skeptic Reassured0.75×53.5%71.6%
20150.11%
20160.12%
20170.15%
20180.12%
20190.01%
20200.00%
20210.14%
20220.22%
20230.32%
20240.11%
20250.00%
Figure 1. Share of all analyzed calls matching the study definition, by year.
Table 4. Twelve-month excess total returns versus SPY (descriptive history, not a signal)
StatisticStudy groupReturns sample
Median excess return-16.4%-10.5%
Interquartile range-36.0% to +7.8%
Share beating SPY38.6% (95% CI 29%–49%)38.2%
Observations83199
Table 5. Most recent calls matching the study definition
TickerQuarterCall dateCall grade
MCDQ2 20242024-07-29D
ASOQ1 20242024-06-11C+
SFIXQ3 20242024-06-04C+
NOAHQ1 20242024-05-30D
IFSQ1 20242024-05-14C+
RCELQ1 20242024-05-14F
PRAQ1 20242024-05-11D
WRBYQ1 20242024-05-09A

4Discussion

A careful reader should note that tunnel calls are, on these measures, more stressed and more likely to accompany guidance cuts and weaker median returns than the average call. But none of this establishes that the recovery language caused anything, or that it predicts future performance. The beat rate, 38.6% versus 38.2% baseline, is essentially indistinguishable. The 2023 peak and 2024 decline in prevalence likely reflect macro conditions rather than any change in management behavior. Treat these statistics as description of a pattern, not as a signal to act on.

5Limitations

The candor, evasion, specificity, stress, promotion, and confidence scores are AI-read fields and inherently noisy. The returns sample covers 22,449 calls and is skewed toward liquid names, so the 83-call tunnel subset may not generalize. Our own forward tests falsified directional prediction, and LLMs partially remember famous stocks' history, contaminating any backtest. The lift figures describe co-occurrence within this corpus, not causal relationships, and the corpus window includes unusual macro regimes. See the full methodology, including the C1 pattern’s forward-test failure and the LLM-memorization finding.

Companion page: every company matching this hypothesis is listed at the question’s own page.

Cite this study Artul.ai Research Group (2026). “Light at the End of the Tunnel Is Often a Train: Recovery Talk on Calls.” Artul.ai Earnings-Call Research Library, Study No. 36. https://artul.ai/research/hypothesis-coming-out-of-the-tunnel

Related studies

The Numbers Are Fine; Everything Else Is Pending: FlRoom to Run and Nowhere to Hide: Calls With UncontesDon't Stick to the Script: Earnings Calls Where AnswHave Your Cake and Expand the Base Too: Repeat GrowtPressed Harder, Answered Straight: A Profile of 183 Second Front, Funded and Fueled: Who Says Yes on the
Not investment advice. Artul.ai publishes AI-generated earnings-call quality grades and expected-volatility estimates — never buy or sell recommendations. We tested over 1,600 predictive hypotheses against 165,000 transcripts; the honest result, including what failed, is documented in our methodology.