Topic · Quant Research
← All topicsData Research
32 articles · 4 checkpoints · 19 deeper reads · 9 reference notes
A standalone topic: it is on no roadmap, so read it on its own terms.
Every article, in reading order
plant a flag as you finish eachRead these first
Before a dataset earns a signal-research slot, it has to survive a checklist that has nothing to do with whether it predicts returns, coverage, timeliness, survivorship, and cost all fail datasets long before the alpha question gets asked.
A dataset that costs $400,000 a year and adds a signal worth $150,000 a year to the book is a bad purchase even though the signal is real. Making that case explicitly, in dollars, is what actually gets a data budget approved or declined.
A vendor trial is a research project with a deadline and a price tag attached, so it needs the same discipline as any other, a stated hypothesis, a fixed evaluation window, and a decision rule agreed before the free data starts flowing.
Vendor data often carries a timestamp that says when a number describes, not when you actually could have known it, and the gap between the two is where a backtest quietly cheats without anyone writing a bug.
Then the rest