Kit generates multiple future financial candlestick paths accurately
KiT: A Foundation Model for Financial Time-Series Forecasting using DiffusionTransformers
Machine LearningComputer Vision and Pattern Recognition
Summary
Predicting future prices in financial markets is very hard because the data is noisy and varies a lot across different markets. The researchers created a new model called KiT that looks at the history of price data and generates several possible future scenarios instead of just one prediction. KiT was trained on a huge amount of data across many markets and time lengths, making it stronger and more general than earlier models. It performs better at forecasting than both specialized financial models and general time-series models.
What this means in practice
- •For quantitative trading teams: Generate multiple plausible future price trajectories to improve algorithmic trading decisions under uncertainty.
- •For risk management teams: Model a variety of potential future market movements to better assess financial risks and portfolio volatility.
Authors
Boyu Zhang, Haorui Li
Abstract
Financial candlestick forecasting is fundamental to quantitative investment, yet it remains exceptionally challenging due to extremely low signal-to-noise ratios and vast heterogeneity across markets and instruments. Existing approaches have largely attempted to introduce deep learning to capture hidden temporal features, but most adopt an auto-regressive formulation, which leads to error accumulation during inference. Meanwhile, general-purpose time-series foundation models are not tailored to the unique structure of k-line data and yield unsatisfactory performance on downstream candlestick forecasting tasks. To tackle these problems, we introduce KiT, a K-line Diffusion Transformer foundation model, and reformulate future prediction as conditional path generation via flow matching: given a historical context window, the model generates an ensemble of plausible future OHLCV trajectories. We pre-train KiT at multiple parameter scales on billions of candlestick bars spanning multiple markets and timescales. Across three markets and seven resolutions, KiT attains a mean return RankIC of 0.057 and a mean volatility RankIC of 0.66, leading at every timescale and outperforming both task-specific financial forecasters and general time-series foundation models. Code will be available at: https://github.com/Luciferbobo/KiT.