Age-Normalized Testosterone Peaks at Series B for Male Startup Founders

The study examines whether and how startup lifecycle stages relate to founders’ hormonal profiles, particularly testosterone, which influences mood, motivation, dominance, and energy. Prior research links testosterone fluctuations to dominance, competitive success, and social status, while stress, sleep, and activity modulate hormone levels. Startup founders face intense workloads, stress, and irregular sleep at early stages, potentially depressing hormonal health. As companies progress from pre-seed to later rounds, changes in lifestyle, stress, and self-perception may alter hormone levels. The authors tested 107 male and 32 female founders at a Y Combinator reunion using an at-home micro-sample blood testing kit to measure multiple biomarkers, aiming to identify patterns between age-normalized testosterone and company funding stages to understand the interplay between entrepreneurial activities and hormonal health.

The paper references work showing testosterone’s impact on mood and physiology, its association with perceptions of dominance and competitive outcomes, and the roles of sleep deprivation and aging on testosterone and DHEA-S. It highlights the dual-hormone hypothesis, wherein testosterone’s link to dominance manifests primarily when cortisol is low (Mehta & Josephs, 2010). These literatures motivate investigating hormonal patterns across entrepreneurial stages, where stress and status cues vary markedly.

Participants were founders attending a Y Combinator reunion who received a SiPhox Health at-home test kit. Finger-prick blood samples (200–400 µL) were self-collected onto ADX100 serum separator cards and processed in a CLIA/CAP laboratory. All assays were previously validated as LDTs with equivalence to venous draws on Beckman immunoassay/chemistry analyzers. Samples were collected over eight afternoon hours to note diurnal patterns; participants were randomly distributed in the testing period. Company stage was assigned by matching participant emails to LinkedIn, PitchBook, RocketReach, or Crunchbase to identify their companies and corresponding stages; unmatched cases were labeled “Unknown.” To normalize for age, participants were binned into 5-year age groups and compared against >5,000 anonymized SiPhox Health reference samples. For each biomarker and age bin, mean and SD were computed; participant values were converted to z-scores relative to their bin and then mapped back using the overall population mean and SD to generate age-normalized values. Statistical analysis used non-parametric methods: Kruskal–Wallis tests for overall between-stage differences; Mann–Whitney U tests for post-hoc pairwise comparisons among stages with n ≥ 5; Spearman correlation to assess ordered trends across stages. Very small stages (e.g., Series B n=2; Late Stage VC n=3) were excluded from inferential testing due to low power. Analyses were conducted in Python 3.11 using SciPy and statsmodels; data handling used pandas; visualization employed matplotlib and seaborn.

• Testosterone increased substantially from pre-seed to seed: median 2.21 to 3.44 ng/mL (+55.7%; p=0.0478), then modestly to Series A: 3.65 ng/mL (+6.1% from seed; p=0.3635). Testosterone peaked at Series B: 4.41 ng/mL (+20.8% from Series A; n=2, no statistical test). From Series B to Late Stage VC, testosterone fell to 2.55 ng/mL (−42.2%; n=3, no test). Acquired median 2.56 ng/mL was lower than Series A (p=0.2561); difference between Acquired and Shut Down (3.49 ng/mL) was not significant (p=0.4923).
• Relative to SiPhox-defined ranges (healthy 2.49–7.81 ng/mL; optimal 4.00–7.81 ng/mL), pre-seed and acquired medians were below healthy; seed and Series A were in healthy but below optimal; only Series B median reached the optimal range (not inferentially tested due to n=2).
• Spearman correlation for testosterone across ordered stages (excluding Shutdown and Unknown) indicated a positive, significant trend (p=0.035).
• Cortisol remained relatively stable through Series B, then rose at Late Stage VC: 6.03 to 8.44 nmol/L (+39.9%; small n, no significance testing). Seed and Series B medians were below the optimal cortisol range (6.7–19.4 nmol/L), whereas pre-seed, Series A, and Late Stage VC were within range.
• DHEA-S showed a slight net decrease from pre-seed (225.99 µg/dL) to Series B (220.50 µg/dL, −2.4%), followed by an increase at Late Stage VC (247.27 µg/dL, +12.1%). All medians were within the stated healthy range (70–690 µg/dL).
• Age normalization was necessary because earlier-stage founders were generally younger; trends likely reflect stress and lifestyle associated with company progression rather than age alone.

Findings suggest that early-stage fundraising success coincides with increased testosterone, potentially reflecting heightened dominance, confidence, and reduced stress. The subsequent testosterone decline alongside a cortisol rise at later stages aligns with the dual-hormone hypothesis: elevated cortisol can attenuate testosterone’s dominance-related effects. This may reflect mounting pressures of scaling, investor expectations, and potential exits. An alternative selection-based interpretation is that founders with inherently higher testosterone are more likely to reach later funding stages, which a cross-sectional design cannot disentangle. The cortisol spike at Late Stage VC temporally aligns with the testosterone downturn, consistent with cortisol’s inhibitory effects. DHEA-S patterns diverged somewhat from testosterone, implying additional regulatory factors beyond precursor availability. Together, the results emphasize the importance of stress management and hormonal health for entrepreneurial performance and well-being, while highlighting the need for longitudinal data to infer causality.

The study provides evidence that age-normalized testosterone in male startup founders rises from pre-seed to a peak around Series B and then declines as cortisol increases at later stages, consistent with the dual-hormone hypothesis. These results underscore a potential physiological dimension to entrepreneurial success and stress across the startup lifecycle. Future work should: (1) analyze biomarker associations with actual funding amounts rather than categorical stages; (2) track founders longitudinally to clarify causality between hormonal dynamics and company progression; and (3) test interventions (e.g., lifestyle changes or pharmacologic options like enclomiphene) to assess whether increasing testosterone meaningfully impacts advancement through funding rounds.

• Small sample sizes in key later stages (Series B n=2; Late Stage VC n=3) precluded reliable inferential testing and limit generalizability.
• Cross-sectional design prevents causal inference; selection effects (e.g., inherently higher-testosterone founders advancing further) cannot be ruled out.
• Afternoon sampling across eight hours may introduce diurnal variability for testosterone and cortisol despite randomization.
• Company stage classification relied on external databases and may contain misclassification; stages themselves are ill-defined.
• Convenience sample of YC-affiliated founders may not generalize to broader founder populations or other ecosystems.
• Analyses focused on male founders; female data were collected but not analyzed here, limiting scope.
• Age normalization used a proprietary reference dataset, which may introduce bias if reference population differs from founders in unmeasured ways.