Enhanced Studies — Real-World Cycle-Log Aggregates

Anonymised aggregate statistics derived from 487 self-reported cycle logs across 50 performance-enhancement compounds, harvested from public bodybuilding and harm-reduction forums and reduced to derived statistics only. Per compound: gender split; dose distribution (median/min/max, grouped by unit); duration distribution normalised to weeks; co-reported compounds (alias-matched); source-domain provenance; a self-reported outcome split (worked / mixed / stopped-or-regretted); and a tally of explicitly mentioned side effects. Every sparse field carries an explicit coverage percentage. Dose, duration, gender and stack extraction are fully deterministic (regex). A language model is used at exactly one step — classifying each log's free-text outcome and tallying reported side effects — constrained to the log text. Read this before you cite. These are aggregated self-reports from public forums, not clinical data. Small samples (typically 6–12 logs per compound). Every figure is indicative, not representative. Selection bias is real. Testosterone reads as mostly “mixed” because its logs are dominated by TRT bloodwork-troubleshooting threads — not because Testosterone does not work. Side-effect counts reflect what users mentioned, not incidence. “Demand” is search volume, not usage. Doses are the figures stated in the logs, not frequency-normalised weekly totals. Where a signal is missing it is left null, never guessed. Coverage percentages are published alongside sparse fields. Privacy. Derived counts and distributions only: no usernames, no verbatim post text, no source URLs. An automated leak-check runs over the outputs. The raw corpus is deliberately withheld — it consists of posts by identifiable people who did not consent to republication — so the method is auditable but the pipeline is not reproducible end-to-end from the published artifacts. Educational and informational only. Not medical advice, and not an endorsement of use. Code: github.com/enhanced-studies/open-data · Method: https://enhancedstudies.com/methods/