Proof · past record as of 21 Oct 2026
Where Zeno earns trust.
Historical tests, separated by field, skill and research package.
- fields
- 24
- scored past forecasts
- 3.7M
- ranked forecasters
- 1,474
three skills
Field by field.
Weather vs airports
- Corrects forecastsv0.80.2 beta
- Proven
- Spots a forecaster about to be wrongv0.82 tree experiment
- Promising
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 22,187 labelled forecasts · 5 forecasters. Not the evaluation sample.
Evaluation interval: not measured here. Airport outcome timing is assumed; inspected-period re-checks are retrospective.
Beta package and verdictEnergy demand
- Corrects forecastsv0.80.2 beta
- Proven
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 2,667 labelled forecasts · 1 forecaster. Not the evaluation sample.
Evaluation interval: not measured here. Single-forecaster field; no evidence of consensus dependence.
Beta package and verdictWeather vs analysis
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Promising
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 48,708 labelled forecasts · 6 forecasters. Not the evaluation sample.
Evaluation interval: not measured here.
Beta package and verdictSea
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Promising
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 12,096 labelled forecasts · 24 forecasters. Not the evaluation sample.
Evaluation interval: not measured here.
Beta package and verdictCrypto
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Promising
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 83,504 labelled forecasts · 3 forecasters. Not the evaluation sample.
Evaluation interval: not measured here.
Beta package and verdictCOVID-19
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Promising
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 6,365 labelled forecasts · 16 forecasters. Not the evaluation sample.
Evaluation interval: not measured here.
Beta package and verdictRSV
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Promising
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 2,891 labelled forecasts · 6 forecasters. Not the evaluation sample.
Evaluation interval: not measured here.
Beta package and verdictPhiladelphia Fed (points)
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Promising
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 4,190 labelled forecasts · 197 forecasters. Not the evaluation sample.
Evaluation interval: not measured here.
Beta package and verdictECB
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Promising
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 1,098 labelled forecasts · 116 forecasters. Not the evaluation sample.
Evaluation interval: not measured here.
Beta package and verdictForecastBench
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Promising
- Spots the whole panel about to be wrongnot measured
- Not yet
Evidence
historical: two sealed test periods; warning: historical: development and later confirmation slices of past data.
Corpus inventory: 108,699 labelled forecasts · 806 forecasters. Not the evaluation sample.
Evaluation interval: not measured here.
Beta package and verdictSea-level pressure
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Wind
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Urban heat
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Single-run weather
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Air quality
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Farming rain
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Ensemble weather
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Space weather
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Prediction markets
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Market traders
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Cross-venue markets
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Flu
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Philadelphia Fed (ranges)
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
Numinous
- Corrects forecastsv0.80.2 beta
- Not yet
- Spots a forecaster about to be wrongv0.82 tree experiment
- Not yet
- Spots the whole panel about to be wrongnot measured
- Not yet
| Field | Corrects forecasts | Spots a forecaster about to be wrong | Spots the whole panel about to be wrong |
|---|---|---|---|
Weather vs airportsEvidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 22,187 labelled forecasts · 5 forecasters. Not the evaluation sample. Evaluation interval: not measured here. Airport outcome timing is assumed; inspected-period re-checks are retrospective. Beta package and verdict | Provenv0.80.2 beta · past data | Promisingv0.82 · past data | Not yet |
Energy demandEvidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 2,667 labelled forecasts · 1 forecaster. Not the evaluation sample. Evaluation interval: not measured here. Single-forecaster field; no evidence of consensus dependence. Beta package and verdict | Provenv0.80.2 beta · past data | Not yet | Not yet |
Weather vs analysisEvidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 48,708 labelled forecasts · 6 forecasters. Not the evaluation sample. Evaluation interval: not measured here. Beta package and verdict | Not yet | Promisingv0.82 · past data | Not yet |
SeaEvidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 12,096 labelled forecasts · 24 forecasters. Not the evaluation sample. Evaluation interval: not measured here. Beta package and verdict | Not yet | Promisingv0.82 · past data | Not yet |
CryptoEvidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 83,504 labelled forecasts · 3 forecasters. Not the evaluation sample. Evaluation interval: not measured here. Beta package and verdict | Not yet | Promisingv0.82 · past data | Not yet |
COVID-19Evidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 6,365 labelled forecasts · 16 forecasters. Not the evaluation sample. Evaluation interval: not measured here. Beta package and verdict | Not yet | Promisingv0.82 · past data | Not yet |
RSVEvidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 2,891 labelled forecasts · 6 forecasters. Not the evaluation sample. Evaluation interval: not measured here. Beta package and verdict | Not yet | Promisingv0.82 · past data | Not yet |
Philadelphia Fed (points)Evidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 4,190 labelled forecasts · 197 forecasters. Not the evaluation sample. Evaluation interval: not measured here. Beta package and verdict | Not yet | Promisingv0.82 · past data | Not yet |
ECBEvidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 1,098 labelled forecasts · 116 forecasters. Not the evaluation sample. Evaluation interval: not measured here. Beta package and verdict | Not yet | Promisingv0.82 · past data | Not yet |
ForecastBenchEvidencehistorical: two sealed test periods; warning: historical: development and later confirmation slices of past data. Corpus inventory: 108,699 labelled forecasts · 806 forecasters. Not the evaluation sample. Evaluation interval: not measured here. Beta package and verdict | Not yet | Promisingv0.82 · past data | Not yet |
| Sea-level pressure | Not yet | Not yet | Not yet |
| Wind | Not yet | Not yet | Not yet |
| Urban heat | Not yet | Not yet | Not yet |
| Single-run weather | Not yet | Not yet | Not yet |
| Air quality | Not yet | Not yet | Not yet |
| Farming rain | Not yet | Not yet | Not yet |
| Ensemble weather | Not yet | Not yet | Not yet |
| Space weather | Not yet | Not yet | Not yet |
| Prediction markets | Not yet | Not yet | Not yet |
| Market traders | Not yet | Not yet | Not yet |
| Cross-venue markets | Not yet | Not yet | Not yet |
| Flu | Not yet | Not yet | Not yet |
| Philadelphia Fed (ranges) | Not yet | Not yet | Not yet |
| Numinous | Not yet | Not yet | Not yet |
Historical research, not a live guarantee. Correction: v0.80.2 beta versus its calibrated reference. Forecaster warning: v0.82 boosted-tree experiment, not the neural run. Passing one skill does not validate the others.