Drop Readiness in one sentence
Drop Readiness is a single 0–100 score that estimates how close an AI model is to shipping, so a builder can read one number instead of reconciling a dozen rumours, changelogs, and market tickers by hand. It is computed from the formula DR = .45 ODDS + .25 INTEL + .20 DEADLINE + .10 VOLUME and refreshes hourly.
The point of the score is altitude. Any individual signal — a leaked codename, a market that ticked up, an expected window on the calendar — tells you something, but none of them tells you where to spend your week. Drop Readiness flattens those signals into one sortable value so a forecast board ranks itself: the releases worth preparing for float to the top, the noise sinks. That is the whole job. The exact weighting and a worked calculation live in the full Drop Readiness methodology; this page is the builder's-eye view of why the number exists and how to use it.
The four inputs, and why they are weighted that way
Drop Readiness blends four independent signals because no single one is trustworthy alone: blended Polymarket and Kalshi odds carry 45%, intel recency 25%, deadline proximity 20%, and market volume 10%. The weights reflect how much each signal moves the truth — odds lead, volume merely confirms.
Odds dominate because public prediction markets are the fastest consensus on timing; they reprice within minutes of a credible leak or a denial, faster than any editorial calendar can update. Intel comes next because corroboration matters — a claim echoed by a changelog and two well-placed posts is worth far more than one anonymous screenshot. Deadline proximity encodes ordinary common sense: a model expected this month is readier than one penciled in for next quarter. Volume earns the smallest slice because it is a confidence dial, not a direction — deep trading makes the odds more believable without telling you which way they lean.
| Input | Weight | What it captures |
|---|---|---|
| Odds | 45% | Blended Polymarket + Kalshi implied probability that the model ships in its window |
| Intel | 25% | How fresh and how corroborated the supporting evidence is |
| Deadline | 20% | Proximity to the expected release window |
| Volume | 10% | Depth of market activity behind the odds — a confidence modifier |
The weights never drift; they read identically here, in the methodology spec, and in the glossary. For the normalisation details and an end-to-end arithmetic example, see the full Drop Readiness methodology.
What the score means: imminent, expected, rumoured
Drop Readiness maps onto three stages that summarise both the band the score sits in and the quality of the signal behind it: imminent, expected, and rumoured. The bands below are illustrative groupings, not hard cutoffs — a thin-volume release can read lower than its odds alone would suggest.
- Imminent — high readiness, near-term window. Markets are confident, the deadline is close, and intel corroborates a ship soon. Illustratively, this is the top band — the releases you act on now.
- Expected — a credible window with solid signal, but further out or carrying more uncertainty than an imminent drop. Worth tracking; not yet worth blocking a sprint for.
- Rumoured — early or single-source signal and low readiness. On the radar, but short of the corroboration or market depth to rank higher.
Each stage is defined precisely, alongside the coined terms it relies on, in the stage and signal definitions.
How to read a Drop Readiness score
Read Drop Readiness as a probability-weighted planning signal, not a promise: an illustrative score around 80 says the evidence strongly favours a near-term ship and it is reasonable to start integration scaffolding, while a score around 40 says watch, don't plan. The numbers here are examples to show the shape of the judgment, not live readings.
The most useful move is to watch the trend, not the snapshot. Because the score refreshes hourly, a release climbing from the 50s into the 70s over a few cycles is telling you that markets and intel are converging — that is your cue to prepare. A score that spikes on one rumour and then sags back is telling you the opposite. A high reading is never a guarantee that a model ships on a given date; it is a calibrated estimate that the available evidence points that way. The odds half of that estimate is the part most people misjudge, so it helps to understand how the market-odds input works before you bet a roadmap on it.
Drop Readiness is a forecasting signal for planning around AI releases — not betting, gambling, or investment advice. Prediction-market odds enter the score only as a forecast of timing. A high score raises the odds you should be ready; it does not promise a launch.
Where the inputs come from
The four inputs draw on public, citeable signals: the odds come from Polymarket and Kalshi, and the intel comes from official changelogs, public posts on X, and deltas in public model-catalog listings, with each item source-linked so you can judge it yourself. Polymarket and Kalshi are the only data sources we name, and they are sources we build on rather than rivals.
Intel is graded before it counts toward the score — a verifiable changelog entry weighs more than an unconfirmed post, and a market move weighs differently again. The grading scheme is its own short read on how intel is graded, and the full freshness-and-provenance policy lives on the where the inputs come from page. To see all four inputs resolved into live scores, watch Drop Readiness on live cards on the home forecast board — static pages like this one explain the method and ranges, never current numbers.