Live BTC/USDT Forecast Dashboard
Upside Probability (Next 24h)
0.0%
The model's confidence that the price in 24 hours will be higher than the last known price.
Volatility Amplification (Next 24h)
16.7%
The probability that predicted volatility over the next 24h will exceed recent historical volatility.
24-Hour Probabilistic Forecast
The chart below shows the historical price (blue) and the probabilistic forecast (orange). The orange line is the mean of multiple Monte Carlo simulations, and the shaded area represents the full range of predicted outcomes, indicating forecast uncertainty.
Methodology Overview
This demo showcases the forecasting results of Kronos, a foundation model pre-trained on the "language" of financial markets. The predictions are generated using the following process:
- Model: The `Kronos-small` (25M parameters) model is used to autoregressively predict future K-line data.
- Data Context: The model uses the last 360 hours (~15 days) of BTC/USDT 1h K-line data from Binance as context for each new prediction.
- Probabilistic Forecasting: We employ Monte Carlo sampling (N=30 paths) to generate a distribution of possible future price trajectories, not just a single point forecast.
- Derived Insights: The resulting distribution is analyzed to produce the mean forecast (solid line), the uncertainty range (shaded area), and the key probability metrics shown above.
About The Kronos Project
Kronos is the first open-source foundation model for financial candlesticks (K-lines), trained on data from over 45 global exchanges. It is designed to serve as a unified model for diverse quantitative finance tasks.