Training Turk

Political Expert — Elections Forecasting & Political Risk

$150–200/hr · Mercor · Hourly, 20 hours a week

You forecast elections and political risk from fixed evidence cutoffs, grading AI models on the judgments seasoned analysts would make.

What you would do

  • Author political forecasts and risk reports constrained by stated evidence cutoffs, without access to hindsight data
  • Make and document judgments about polling shifts, win probabilities, and market-implied odds in real time
  • Articulate the causal mechanisms linking political events to expected market and political consequences
  • Evaluate and grade model-generated political analyses against your own professional standard
  • Flag instances where analyses use data, information, or hindsight outside the allowed evidence window

Who they want

  • 8+ years of live judgment experience in polling, campaign strategy, election forecasting, or political-risk analysis
  • Background from leading national polling organizations, election forecasting shops, or political-risk teams at major institutions
  • Demonstrated ability to make probability judgments about elections, polling, or financial markets in real-time environments
  • Strong understanding of polling methodology, statistical inference, and causal reasoning in political contexts
  • Willingness to work from strict evidence cutoffs and document reasoning without material non-public information

Main skills

Elections forecastingPolling methodologyPolitical risk

What the interview asks about

  1. 1.Point-in-time forecasting discipline

    Making judgments at a fixed moment in time, before outcomes are known, demands rigor that retrospective analysis never does; the interviewer checks you distinguish live judgment from hindsight commentary.

    For example: “Describe a specific election or polling question where you made a probability estimate before the outcome was known. What data drove your forecast, and how did the actual result compare?”

  2. 2.Polling literacy and interpretation

    Polling is central to this role; you need to read polls critically, understand sampling and weighting, and avoid common misinterpretations that trap less rigorous analysts.

    For example: “A candidate's approval rating rose 3 points in one poll but fell 2 points in another released the same week. How would you reconcile these, and what would you tell a client about the true trend?”

  3. 3.Causal reasoning under uncertainty

    Political impact analysis requires articulating plausible causal chains; interviewers want to see whether you can reason through mechanisms, not just cite correlations.

    For example: “A central bank unexpectedly raises rates two weeks before an election. How might this affect voter sentiment, turnout, or market prices, and what additional information would help you sharpen your forecast?”

  4. 4.Probability judgment in live markets

    This role involves evaluating odds and win probabilities in real-world political markets; demonstrating sound judgment under uncertainty is essential.

    For example: “A prediction market prices a candidate at 65% to win. What pieces of information would make you confident this is mispriced, and what would you recommend to a decision-maker?”

  5. 5.Model evaluation against professional judgment

    You'll grade AI analyses; this requires knowing what sound political analysis looks like and spotting where models fail to reason causally or miss nuance.

    For example: “An AI model forecasts a tight Senate race based on polling averages alone. What limitations do you see in this approach, and what factors would a seasoned political analyst layer in?”

A task you may get

Point-in-time exercise: Given an election scenario, specific polling data, and a stated evidence cutoff date, produce a brief forecast of the winner and explain your reasoning, including uncertainty and key assumptions.

How to prepare

  • Study recent elections and polling data from the actual time period, not retrospective analyses, to understand how forecasters reasoned in real time
  • Review case studies of forecast errors and successes in published political-risk or polling research, focusing on causal mechanisms
  • Practice articulating multi-factor reasoning: how different data streams (polling, economic indicators, turnout proxies) interact in your thinking
  • Research major political-risk frameworks used by banks, funds, or consulting firms that combine structural factors with real-time judgment

The facts

Pay
$150–200/hr
Hours
Hourly, 20 hours a week
Where
Remote
Field
Life, Physical, and Social Science
Project name
Jade
Posted
9/14/2026
Places left
47

We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.