Probabilities, not predictions dressed as facts
Outcomes are expressed as ranges and likelihoods, with uncertainty kept visible.
A Bitcoin forecasting experiment
A forecast should earn your attention.
BTC Horizon is building transparent probability forecasts for Bitcoin—and recording every result. The model is being tested privately before the forecasts become a public product.
No subscription yet
No trading signals
Results kept visible
Why BTC Horizon
Less certainty.
More accountability.
A permanent track record
Forecast editions are frozen at publication and measured after their horizon matures.
Simple models remain competitors
A more complicated formula is useful only when it repeatedly beats transparent controls.
Failures stay part of the story
Rejected horizons, missed forecasts, and weak performance are evidence—not content to hide.
The approach
Built to be challenged.
BTC Horizon separates model construction, locked historical evaluation, and daily forward testing.
Form a hypothesis
Price behaviour, long-term structure, currency context, and market sentiment become versioned inputs—not an endlessly changing formula.
Test unseen history
Training, validation, and test periods stay chronological. The candidate must beat a price-only control on declared metrics.
Let time decide
Qualifying editions run unchanged each day. Forward results—not a persuasive backtest—decide whether the model deserves a public role.
Current status
Research first.
Release later.
Formula v3’s 24-hour edition cleared its locked historical gate by a small margin. That is a reason to test it forward—not a reason to claim predictive skill.
✓
Historical research locked
Formula parameters and evaluation rules are frozen.
→
Daily 24-hour forward testing underway
The sample must grow before performance can be interpreted responsibly.
—
Public forecasts not released
No launch date or paid plan is being promised while validation and data rights remain unresolved.
Research guides
Understand the evidence.
Bitcoin probability forecasts, explained
What a Bitcoin probability forecast says, what it does not say, and why ranges are more honest than a single price target.
Read guide →Why backtesting is not enough
A practical explanation of chronological backtests, locked tests, and why a forecasting model still needs forward evidence.
Read guide →Brier score for Bitcoin forecasts
How the Brier score evaluates probability forecasts and why confidence is punished when the outcome goes the other way.
Read guide → Follow the experiment
Receive the research updates—and a beta invitation if the model earns one.
Join the early list for occasional validation updates. There is no paid plan, no trading advice, and no promise that the research will become a product.