longbridge-ml-strategy

Original：🇺🇸 English

Translated

Machine-learning prediction strategy framework via Longbridge Securities — walk-forward rolling training with feature engineering (MACD, RSI, Bollinger Band width, volume change rate) and a scikit-learn classifier (Random Forest / Gradient Boosting); retrains every 60 days, predicts 5-day direction; buy signal when probability > 0.6, sell when < 0.4; evaluates win rate, profit factor, and Sharpe ratio. Triggers: "机器学习", "ML策略", "预测模型", "随机森林", "梯度提升", "深度学习", "AI选股", "walk-forward", "機器學習", "ML策略", "預測模型", "隨機森林", "梯度提升", "machine learning", "ML strategy", "predictive model", "random forest", "gradient boosting", "AI stock selection", "walk-forward", "rolling training", "feature engineering", "scikit-learn", "XGBoost".

3installs

Sourcelongbridge/skills

Added on2026-05-12

NPX Install

npx skill4agent add longbridge/skills longbridge-ml-strategy

SKILL.md Content

View Translation Comparison →

longbridge-ml-strategy

Walk-forward machine-learning framework for stock direction prediction. Fetches historical OHLCV data, engineers technical features, trains a rolling classifier (Random Forest or Gradient Boosting), generates probabilistic buy/sell signals, and evaluates backtest performance.

Response language: match the user's input language — Simplified Chinese / Traditional Chinese / English.

Dependencies

Requires:

scikit-learn

,

pandas

,

numpy

(usually pre-installed). Optional:

xgboost

or

lightgbm

for gradient-boosting models. If unavailable, fall back to a simpler logistic-regression model.

When to use

User asks for ML-based prediction, rolling model training, feature-importance analysis, or AI-driven entry/exit signals for a single stock.
Triggers: "用机器学习预测 TSLA 涨跌", "NVDA random forest strategy", "walk-forward backtest AAPL".

Workflow

Fetch 504 daily candles (≈ 2 years):

longbridge kline <SYMBOL> --period day --count 504 --format json

Feature engineering (compute on rolling windows):
- MACD line and signal (EMA12 − EMA26, signal EMA9)
- RSI-14
- Bollinger Band width: (upper − lower) / mid, window 20
- Volume change rate: (vol_t − vol_{t-5}) / vol_{t-5}
- 5-day price momentum: (close_t / close_{t-5}) − 1
- Label: 1 if close_{t+5} > close_t × 1.01, 0 if < close_t × 0.99, else drop
Walk-forward training:
- Training window: 252 days; retrain every 60 days
- Model:
```
RandomForestClassifier(n_estimators=100)
```
  or
```
GradientBoostingClassifier
```
- Predict probability for the current bar
Signal generation:
- prob > 0.60 → Buy signal
- prob < 0.40 → Sell/Short signal
- Otherwise → Hold / neutral
Backtest metrics (on out-of-sample predictions):
- Win rate (% correct directional calls)
- Profit factor (gross profit / gross loss)
- Annualised Sharpe ratio (assuming daily rebalance)
- Max drawdown
Feature importance: rank top-5 features by mean decrease in impurity.

Run

longbridge kline --help

to confirm flag names before calling.

CLI

bash

longbridge kline --help

longbridge kline <SYMBOL> --period day --count 504 --format json

Output

Metric	简体	繁體	English
Current signal	当前信号	當前訊號	Current signal
Signal probability	预测概率	預測概率	Signal probability
Win rate	胜率	勝率	Win rate
Profit factor	盈亏比	盈虧比	Profit factor
Sharpe ratio	夏普比率	夏普比率	Sharpe ratio
Max drawdown	最大回撤	最大回撤	Max drawdown
Top features	重要特征	重要特徵	Top features

Output: current signal box → backtest summary table → feature importance list → caveats (past performance, data snooping). Cite Longbridge Securities / 数据来源：长桥证券 / 數據來源：長橋證券.

Error handling

Situation	简体回复	繁體回復	English reply
`command not found: longbridge`	回退到 MCP 或提示安装 longbridge-terminal	回退到 MCP 或提示安裝 longbridge-terminal	Fall back to MCP or install longbridge-terminal
`not logged in` / `unauthorized`	请运行 `longbridge auth login`	請執行 `longbridge auth login`	Run `longbridge auth login`
`scikit-learn` not found	提示 `pip install scikit-learn pandas numpy` ，并改用逻辑回归降级	提示安裝，降級至邏輯回歸	Prompt install; degrade to logistic regression
Fewer than 252 candles	数据不足，无法完成 walk-forward 训练	數據不足	Insufficient data for walk-forward
Other stderr	直接显示原始错误	直接顯示原始錯誤	Surface verbatim

MCP fallback

Use

mcp__longbridge__candlesticks

with

period=Day

,

count=504

when CLI is unavailable.

Related skills

```
longbridge-kline
```
— OHLCV data source
```
longbridge-volatility-strategy
```
— vol regime as an additional feature
```
longbridge-multifactor
```
— cross-sectional factor signals complement

File layout

longbridge-ml-strategy/
└── SKILL.md

longbridge-ml-strategy

NPX Install

Tags

SKILL.md Content

longbridge-ml-strategy

Dependencies

When to use

Workflow

CLI

Output

Error handling

MCP fallback

Related skills

File layout