What the work actually involves

Each task hands you a named covered company, a named sell-side analyst, a catalyst (a print, a guide-down, an 8-K, a capital markets day, a deal) and a hard evidence cutoff. You forecast two things: when that analyst's note lands — same night, next morning, after the conference, or not at all — and what it contains: rating action or reiteration, price target direction and rough magnitude, which estimate lines move, and which two or three arguments carry the note. You then write the reasoning that supports it, sourced to filings, the transcript, consensus positioning, the valuation framework that analyst actually uses, and the internal mechanics of how research gets produced: when a model gets updated, when supervisory analyst review and compliance clearance constrain timing, when a quiet period or restriction list blocks publication entirely.

A second stream is review: you read model-generated analyses of the same setups and grade them against your own standard, flagging hindsight contamination, fabricated estimate detail, valuation logic that no covering analyst would defend, and conclusions that ignore how that specific franchise behaves.

What the screen is looking for

  • Whether you personally published post-catalyst research in the sector under review — not managed it, not read it. Expect follow-ups on your ticker list, your publication cadence, and your last rating change.
  • Whether you can name how specific analysts and franchises behave: who reiterates and waits, who goes out same-night with a tactical note, whose price target moves mechanically with the model versus discretionarily.
  • Point-in-time discipline. Any answer that quietly uses what happened next, or that reaches for anything resembling non-public information, is disqualifying regardless of pedigree.
  • Calibrated uncertainty. Forecasts are graded; confident answers with no stated probability or no stated falsifier score worse than honest ranges.

Logistics

Fully remote and asynchronous, contractor engagement, work drawn from a task queue rather than assigned shifts. Most contributors run 10–20 hours a week; tasks are substantial — a single publication forecast with full written reasoning can take an hour or more, and quality matters more than throughput. Pay is observed in the $150–250/hr range and tracks sector scarcity, seniority, and work-sample performance; it is not guaranteed. A paid or unpaid work sample is the real gate — brand-name experience gets you read faster, nothing more.