jev_stock
Experiment that builds past-only Hong Kong market state and asks Jev for an up, flat or down forecast, with a standalone HTML report.
Overview
This repository is an experiment in forecasting the next Hong Kong stock-market direction with TypeSafe Jev. You enter a stock code; the program fetches the stock plus the Hang Seng Index and Hang Seng TECH Index through AKShare, drops an incomplete current trading day before 16:30 Hong Kong time, aligns dates, builds a past-only state, and asks Jev one independent question per requested horizon for an up, flat or down call. Responses are validated and can be rendered as an HTML report showing the cutoff date, forecasts, probabilities and state summary.
The author is notably cautious. They see Jev as a disciplined reasoning layer that gives one repeatable interpretation of mixed weak signals, not a proven trading model. They stress that a probability like down 0.68 is not a historical win rate and that calibration would need many timestamped forward predictions. A retrospective backtest script is included for debugging but is not a clean test because the model may already know historical events.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Will this Hong Kong stock move up, flat or down over the next horizon? | up · flat · down |
How it works
- Normalise stock code and fetch stock, HSI and HSTECH data
- Drop incomplete bars, align dates, reject stale data, set as_of
- Build past-only state and ask one question per horizon
- Validate response and optionally render an HTML report
Strengths
- Code owns data, leakage checks and scoring; the model only classifies
- Explicit leakage and staleness checks
- Transparent about the limits of model probabilities
Caveats
- Experimental; not investment advice
- Probabilities are not calibrated win rates
- Backtests may be contaminated by the model's prior knowledge
At a glance
- Type
- Project
- Status
- research
- Language
- Python
- Runtime
- CLI
- Has code
- yes
Use cases and integrations
How curated lists describe it
Short-term forecasting: experimental Hong Kong stock framework that turns structured market state into a Jev decision on price direction, with a backtest script for the first trading day.
yibie/awesome-jevAn experimental JEV-powered framework for forecasting short-term stock price direction from structured market data.
hellogumbo/awesome-jevExperiment in forecasting Hong Kong stock direction: it builds a past-only state from market data, asks for an up, flat or down call, and renders a standalone report.
MrJev/awesome-jev
Frequently asked questions
Does this give trading advice?
No. It shows raw model output and the author explicitly does not turn it into a recommendation.
What data does it use?
The target stock, the Hang Seng Index and the Hang Seng TECH Index fetched through AKShare.
Can the backtest be trusted?
Only for debugging; the author notes the model may already know the historical events.
Related projects
Local China A-share technical analysis workbench where Jev assesses eight indicator dimensions and returns Buy, Hold or Sell with evidence.
ProjectbetaPython, TypeScriptStock terminal that gathers live evidence via Valyu and asks Jev for the probability a stock trades higher in 30 days.
ProjectprototypeTypeScriptExperimental Polymarket bot: a web-search model writes an evidence brief, Jev prices P(YES), and code trades only on a configured edge.
ProjectprototypePythonTerminal agent for Polymarket 5-minute BTC markets that asks Jev up or down and buys once only above 0.90 confidence; dry-run default.
ProjectprototypeTypeScriptPolymarket BTC 5-minute up/down agent: code prices with a volatility model, Jev scores regime and vetoes anomalies; paper by default.
ProjectprototypePythonPaper-only research tool for Polymarket BTC 5-minute Up/Down markets: Jev scores the direction, separate code decides entries.
ProjectresearchTypeScript