Lead-Time Ledger

Lead-Time Ledger how early our signals see a raise

This page publishes one number and holds it to a strict test: a signal counts as early only when the funding round came after we logged the signal — not after the date printed on the document the signal came from. Most trackers measure the second and call it foresight. Under that test the first results publish in November 2026, once the earliest signals have had a full 90-day window to resolve. The table below is live and will fill in as they do. Until then it shows no claim rather than an early one, which is the same rule the rest of this site runs on: every figure carries the sample it came from.

Why there is no lift figure here

A lift number needs a base rate — how often a comparable company raises in the same window — and ours would be computed over companies whose rounds we have actually recorded. We hold a dated funding round for 319 of 1051 tracked companies (30.4%). Every company whose financing we have not captured counts as "did not raise", which pushes that base rate down and any lift derived from it up. So this page publishes the share we measured directly and the sample it came from, and will publish a lift when that coverage is high enough to carry one.

The ledger share of watched companies that raised · median lead · n, by horizon

Signal	within 90 days	within 180 days	within 365 days
TLS certificate transparency private companies	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample
SEC Form D — secondary demand private companies	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample
Regulatory signals, combined private companies	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample
Regulatory signals, combined venture stage only	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample
Federal Register notices all tracked	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample
Federal contract awards all tracked	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample
Federal research grants all tracked	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample	—% 0/0 — · n=0 thin sample

Each cell: the share of watched COMPANIES followed by a raise within the window, with the hits over the sample so you can check it, and the median lead. One company counts once however many times its signal fired — a company that issues certificates continuously is one prediction, not hundreds. Samples under 30 companies are shown but flagged — read them as direction, not proof. Individual hits, with dates, are on the track record.

VIX composite score in validation until ~2027-11

The composite blends every signal into one score. Its scores are frozen as they are issued (8,353 so far) and judged only once a raise could have happened either way — so no hit rate is published for it until those outcomes mature. The signal families above are what the composite is built from, and they are measured today.

How this is measured

Universe. Private venture companies only — public incumbents, investors, people and mis-resolved names are excluded (entity-hygiene-v3-2026-10-01).
Hit. A funding round — a sourced round date or an SEC Form D filing — dated after the signal and within the window.
Matured. Only signals old enough for the outcome to be known either way are counted, so misses count as fully as hits.
One company, one prediction. A company is counted once per horizon however many signals it emitted, and counts as a hit if any of them preceded the raise inside the window. Counting each signal separately turned a handful of continuously-shipping companies into hundreds of "predictions".
No lift figure. A lift needs a base rate, and ours would rest on funding dates we only hold for about a quarter of tracked companies — see above. The share and the sample are measured directly; a lift would not be.
Reproducible. Run bt_01m4h1k2796v28b0sec48cgzxs, frozen 2026-10-09, over 3,323 signals.