Methodology
The original experimental formula, with a separate CMC price source and outcome record.
Market data
The dashboard compares prices from multiple providers and shows their display average and chart. Forecast inputs and outcomes use only CoinMarketCap BTC/USD quotes, independently of that display average. CMC display quotes refresh approximately every ten minutes; collection around publication and target times supports accurate timestamps. Stale or unavailable model input prevents publication.
Unchanged formula
The previous POC formula retains its fixed log-return mean and standard deviation: 0.0005 and 0.026 for 24h; 0.0035 and 0.069 for 7d; 0.015 and 0.145 for 30d. A lognormal distribution produces medians, direction probabilities and 50%, 80% and 95% intervals. These assumptions do not adapt to current volatility.
Experimental status
No dependable predictive advantage has been established. Probabilities are outputs of fixed assumptions, not measured chances guaranteed to occur. Free and Plus use the same experimental model; subscriptions add monitoring, alerts and reviews, not a validated prediction advantage. This is not financial advice or a trading signal.
Publication and outcomes
A daily edition is normally issued shortly after 00:00 UTC. Each target is exactly 24 hours, 7 days or 30 days after issuance. Immutable editions resolve against CMC observations within 90 seconds of their target; missing observations are marked invalid.
Separate track record
The CMC edition starts a new outcome record. The previous multi-provider results do not validate the source change. No EUR conversion, raw-data API or bulk download is offered in this release.
Improvements in testing
Adaptive volatility and fitted candidates remain private. The latest fitted CMC candidate failed its held-out comparison and was not promoted. Replacements require frozen evaluation rules, matched baseline comparisons and new forward evidence; no automatic promotion occurs.