How Signalmap Collects Crypto Exchange Hiring Data: Methodology Q&A

# How Signalmap Collects Crypto Exchange Hiring Data: Methodology Q&A Transparency is core to what we do. This post answers the most common questions about our data collection and analysis methodology. ## Where Does the Data Come From? **Primary sources:** - Exchange career pages (direct scraping): Coinbase, Binance, OKX, Kraken, Gemini, Bitpanda, Bitvavo, BitMEX, Crypto.com, Robinhood Crypto - LinkedIn (via public job listings) - Indeed and other aggregators (for validation and gaps) We do NOT use: - Proprietary or non-public data - Internal company information - Leaked documents - Data purchased from exchanges All data is publicly available job listings. We are collecting the same information anyone could collect manually — we just do it systematically for all 10 exchanges simultaneously. ## How Are Roles Classified? Each job listing is classified by: **Department**: Engineering, Product, Compliance, Legal, Finance, Marketing, Sales, Operations, HR, Data, Design, Research, Corporate Development **Seniority**: Junior (IC1-2), Mid (IC3-4), Senior (IC5-6), Lead/Staff (IC7+), Management (Director+), Executive (VP+, C-suite) **Jurisdiction**: Which country or region the role is in (for geographic expansion signals) **Function tag**: More specific sub-department (e.g., within Engineering: "DeFi", "L2", "Mobile", "Backend") Classification is done via a combination of NLP keyword matching and rule-based systems, with periodic manual review. ## How Is Signal Score Calculated? Signal Score is a weighted composite of: 1. **Volume** (30%): Total active roles vs. historical baseline 2. **Velocity** (25%): 4-week change rate vs. 12-week average 3. **Diversity** (20%): Number of departments with new roles simultaneously 4. **Seniority** (15%): Proportion of Director+ level hires 5. **Geography** (10%): New jurisdiction hiring The formula is normalized so that 100 = theoretical maximum hiring signal and 0 = no hiring activity. ## How Often Is Data Updated? Data is collected and processed weekly. Signal Scores are updated every 7 days. The dashboard reflects the most recent weekly snapshot. We note that some exchanges update their job listings less frequently than others — we annotate this where it affects data freshness. ## How Are Predictions Made? Predictions are made by the Signalmap team based on: 1. **Pattern recognition**: What has this type of hiring cluster historically predicted? 2. **Cross-reference**: Does any other signal (on-chain data, news) support or contradict the hiring signal? 3. **Historical base rate**: How often do similar hiring patterns resolve positively? Each prediction includes: - Clear statement of the predicted event - Evidence (specific roles and counts that triggered the prediction) - Confidence score (our estimated probability) - Target horizon (when we expect resolution) - Outcome (publicly logged when the prediction resolves) **We never delete predictions** — all are logged with their outcome (including misses). ## What Is the Current Accuracy Rate? **83%** on 12 resolved predictions as of Q1 2026. Sample is small — treat this with appropriate statistical caution. Our methodology is designed to be conservative (we set confidence scores carefully and don't make predictions without strong evidence). ## Can I Trust the Data? We publish our methodology publicly. We don't hide predictions that turned out wrong. Our data sources are all public. We'd rather you evaluate our track record for yourself than take our word for it: see the full scorecard at [signalmap.live/predictions](/predictions). [See methodology details](/how-it-works) | [View prediction scorecard](/predictions)

Want the full picture every Friday?

More from Signalmap