live · checking…
Velora
Probability cones for stock prices.
A stochastic market forecasting platform. It models a stock’s terminal price distribution with Monte Carlo simulation (up to 1,000 paths), Geometric Brownian Motion and a 3-state Markov chain, then adjusts its risk assumptions using LLaMA 3.3 sentiment on live headlines.
- the problem
- Most stock tools show a single price target, which hides how wide the range of outcomes really is.
- what I built
- A forecasting platform that simulates price paths with regime switching and headline sentiment, then reports the spread as fan charts, upside probability, VaR and CVaR.
What it does
- Live OHLCV prices from Twelve Data, plus quotes and headlines from Finnhub
- Monte Carlo engine with regime-switching volatility and jump-shock stress events
- Groq LLaMA 3.3-70B reads recent headlines and shifts the drift and volatility assumptions
- Fan chart and histogram drawn on a plain HTML canvas, no charting library
- Risk numbers: expected terminal price, upside probability, VaR and CVaR at 95%
- Scenario presets (Base, Risk-On, Risk-Off, Black Swan) and a regime heatmap with the Markov transition matrix
- Runs in demo mode on synthetic data when API keys are missing
How it's built
each layer feeds the one below
Data
- Twelve Data: OHLCV
- Finnhub: quotes and headlines
Model
- GBM Monte Carlo, up to 1,000 paths
- 3-state Markov regimes + jump shocks
- LLaMA 3.3 sentiment → drift and volatility
Output
- canvas fan chart and histogram
- VaR / CVaR (95%), upside probability
- scenario presets
ReactViteMonte Carlo