Product · Forecasts
Financeβ calibration periodA forecast is scored only when something can settle it.
Below is a forecast published from our own workspace, read live. Before the probability was written, the forecast was tied to a named public series, an explicit threshold, a settlement rule and a deadline. Those four fields are not metadata added afterwards: the binding is constructed with the forecast, and the lane that issues forecasts cannot construct one without it.
That rule did not reach backwards. Forecasts issued before it are on the record, cannot be scored, and are published as unscoreable rather than dropped — most of the record is in that state. Accuracy counts the scoreable and the unscoreable live.
read from the published set 2026-10-21 09:37 UTC · issued 2026-09-18 07:04 UTC · forecast ffd5b7bd
The question
What was asked, and what would answer it.
This is the question text stored on the binding, not a restatement of it. It names the series, the threshold, the window and what counts as YES.
Probability that total USD stablecoin market capitalization reaches $325B by Dec 17, 2026 (DeFiLlama total). Binary outcome: resolves YES if the threshold is crossed at any point between forecast issuance (Sep 18, 2026) and the deadline, NO otherwise.
The binding
The part that makes it scoreable.
On the resolution date the series is read, the rule is applied and the outcome is recorded — whatever it says. Nothing in this table can be edited after issue.
- Bound series
- defillama_total_stablecoins_peggedUSD
- Threshold
- 325,000,000,000
- the value the series must reach for the outcome to be YES
- Settlement rule
- threshold_before
- the only rule the lane writes: the threshold crossed at any point before the deadline settles YES
- Issued
- 2026-09-18 07:04 UTC
- Deadline
- 2026-12-17 23:59 UTC
resolution binding as stored with the forecast · issued 2026-09-18 07:04 UTC
The binding is a frozen record, and its four fields — series, threshold, rule and deadline — are required: none of them has a default, so the lane cannot construct a spec without all four. That is why a forecast issued under this rule can be scored later, and why the score is published whether it flatters us or not. The company record — what is scoreable, what is scored, and what can never be scored — is on Accuracy.
The estimate
A point estimate, and the interval around it.
The published payload carries the point estimate and its interval. It does not carry the sampled distribution, so no distribution is drawn here.
70.51%
Point estimate, p(YES)
probability the threshold is crossed before the deadline
69.62–71.40%
95% interval
monte_carlo_sampling_95
price_band
Forecast type
sector: finance
point estimate and 95% sampling interval · issued 2026-09-18 07:04 UTC
What the interval is. monte_carlo_sampling_95 — a 95% interval on the Monte-Carlo sampling of this probability, computed from the estimate and the number of samples drawn. It states how tightly the sampler has pinned down the model's own answer. It is not a credible interval over the world and it carries no claim that the model is right: a narrow interval around a wrong probability is still a wrong probability. That is what the Brier score is for.
No probability bins are published with this forecast, so no posterior histogram appears on this page. Where a binned distribution appears elsewhere on this site, it is labelled illustrative.
The public scoreboard and the commit–reveal ledger are readable without an account. Forum Forecast inside the product is a Team feature.
Start freeSee the ledgerA modelled likelihood during the calibration period — not a statement of fact, and not investment, legal or financial advice. See the methodology and the public record.