AlphaSift skill
Stock screener for AI agents: picks shares on seven markets with YAML strategies and checks saved runs against prices.
- 473 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Requests go to public quote sources, including unofficial Yahoo and Naver endpoints.
- With a model enabled, the candidate list is sent to the chosen LLM provider.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/alphasiftDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Requests go to public quote sources, including unofficial Yahoo and Naver endpoints.
- With a model enabled, the candidate list is sent to the chosen LLM provider.
This is third-party code. Review the repository files before installing.
What it does
AlphaSift scans every China A-share, or the larger listings of six other markets, screens them with strategies written as YAML files and saves each run. Through `SKILL.md` an agent calls the `alphasift` commands, receives a JSON table of candidates and analyses it further. Hard filters, scores and protective thresholds run without an API key. A language model joins only if you configure one, and it re-ranks the shortlist. `portfolio_profile.mode: hard_cap` limits how many picks one sector can take in the top list. Later `alphasift evaluate` checks saved runs against price movement: the return N trading days after entry is compared with the market index.
Who it is for. Private investors and analysts who look for candidates on the China, US, Hong Kong, Japan, Korea, Taiwan or India markets and want to test their selection rules against saved runs.
Good fit when
- You need a list of China A-shares with low PE and PB, with at most two names per sector.
- An agent builds a JSON table of candidates for further review on another market, for example Hong Kong.
- A week after a run you want to know whether the picks beat the index.
- You want your screening strategy in a readable YAML file and compare changes through a diff.
Not a fit when
- You look for a backtester that models dividends, trading suspensions and position limits.
- You need to place trades: the tool sends no orders.
- You need deep analysis of one stock rather than a broad market screen.
- You need candidate-level news and fund flows outside China.
Example request
Run the dual_low strategy without an LLM for the China market, show five picks and save the run so the result can be evaluated later.Limitations
AlphaSift selects and ranks candidates. It is neither a backtester nor an execution system, and the check after N trading days ignores dividends, suspensions and position limits. Markets other than China run through one unofficial Yahoo source with fewer fields, and Korean PE and PB come from an unofficial Naver API. Candidate-level news, announcements and fund flows exist only for China A-shares. Data and model output can be delayed or wrong, so the result is research material, not investment advice.
How to disable. Remove the skill folder and uninstall the `alphasift` package from the Python environment. Saved runs live in the `ALPHASIFT_DATA_DIR` directory (default `./data`) and are deleted separately.
FAQ
Does AlphaSift need an API key?
No. Screening, scoring and protective thresholds run without one, and model ranking is switched on separately through `GEMINI_API_KEY`, `OPENAI_API_KEY` or `DEEPSEEK_API_KEY`.
Which markets does AlphaSift support?
China A-shares through the main source chain, plus the US, Hong Kong, Japan, South Korea, Taiwan and India through the Yahoo screener. The `--market` flag selects the market.
How do I check whether a run worked?
Save the run with `--save-run` and a few days later execute `alphasift evaluate <run_id> --with-price-path --explain`. For an N-trading-day window use `evaluate-strategies --window 5,10,20`.
Can I write my own strategies?
Yes. Filters, factor weights, protective thresholds and sector buckets sit in YAML files in the `strategies/` directory, and the repository's strategy guide has examples.
Does AlphaSift send trades to an exchange?
No. It only screens and ranks candidates, while decisions and execution stay with the user.
Discussion
No questions or reviews yet.
Related
SkillsFour skills for China's A-share market: quotes and indicators, trend-pullback signals, a simulated account and backtests
A skill that queries Arkvol for China, Hong Kong and US market data and retells it with the date, indicators and limits of use
InvestSkill gives an assistant 34 ready frameworks for analysing US stocks, from company valuation to portfolio risk
A skill that finds bottlenecks in an industry's supply chain and shortlists A-share, Hong Kong and US stocks with data checks
A set of skills for stock and options analysis via the AlphaGBM API: quotes, strategy scoring, Greeks, volatility and research notes
A set of installable finance skills for the agent: news, stock data, sentiment, forecasting, signals, reports and search
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| AlphaSift (this card) | no | 473 | ||
| stockscreen-mcp | yes | not required | 417 | serves daily results of ready-made screeners through MCP; AlphaSift runs its own YAML strategies and evaluates saved runs |
| Claude Trading Skills | no | 3,800 | a skill set for traders with market overview and trade journal; AlphaSift focuses on full-market screening and run evaluation |
README in short
The repository describes layers L1 (hard filters and scores), L2 (model ranking) and L3 (post-analysis), plus step-by-step checks of saved runs. Data sources are ordered by health, and a last-good snapshot marked as stale is used when every live source fails.
README badge
Original SKILL.md
--- name: alphasift description: "全市场选股 Skill。Use when: 用户要按策略筛选 A 股、美股、港股、日股、韩股、台股或印度股票，列出可用策略，运行双低/放量突破/全球低估值等选股，或保存运行并做 T+N 后验评估。通过 alphasift CLI 或 Python 接口输出候选股票列表。" --- # alphasift — 自动选股 Skill 按策略筛选、评分并排序候选股票：A 股，以及美股、港股、日本、韩国、台湾、印度。 定位：全市场候选发现与横向排序引擎。它站在 `daily_stock_analysis` 这类单股深度分析服务上游；DSA 只是可选 L3 后置分析器，不是主筛选依赖。 ## Use When - 用户要列出当前可用策略 - 用户要按 `dual_low`、`volume_breakout` 这类策略筛选 A 股，或用 `global_value` 筛选其他市场 - 用户要拿到结构化 JSON 结果，供后续 agent 继续分析 - 用户要保存选股运行，并在之后用最新快照评估结果 - 用户第一次接触本项目，想用 `alphasift quickstart` 一键看到全市场→候选→排名的最小闭环 ## Preconditions - 默认 `market="cn"`；`market` 也可以是 `us` / `hk` / `jp` / `kr` / `tw` / `in`（Yahoo 筛选器快照 + yfinance 日 K，代码为 Yahoo 代码如 `0700.HK`；无 hotspot/board_heat/tushare/候选级新闻等 A 股专属数据），需使用 `market_scope` 包含该市场的策略：`global_value`（us/hk/jp/kr/tw/in）或 `us_momentum_quality`（us）。韩国 PE/PB 来自 Naver Finance - 需要先在仓库根目录安装包：`pip install -e .` - 如需 LLM 排序，可设置 `LITELLM_MODEL`、`LLM_CHANNELS`、`LITELLM_CONFIG` 或旧变量 `LLM_API_KEY/LLM_MODEL/LLM_BASE_URL` - 可直接复用 `daily_stock_analysis` 的 LiteLLM 配置字段，包括 `OPENAI_*`、`GEMINI_*`、`DEEPSEEK_API_KEY`、`OLLAMA_API_BASE` - 策略 YAML 可通过 `scoring_profile`、`risk_profile`、`portfolio_profile`、`scorecard_profile` 覆盖默认规则 - 策略 YAML 可通过 `event_profile` 配置偏好/规避事件、公告类别和候选上下文来源权重 - LLM 会输出候选行业/主题标签；如候选提供 `industry/concepts/board_heat_score/board_heat_trend_score`，会作为 LLM、主题热度因子与组合分散层锚点；history sidecar 可回填持续性、降温和状态字段。默认组合分散层会用这些标签映射风险桶，降低同一拥挤交易重复占位 - L3 默认启用本地 `scorecard` 后置评分器，也可追加 `dsa` 或 `external_http` - 如需 DSA 后置分析，设置 `DSA_API_URL`，默认调用 `POST /api/v1/analysis/analyze` - 依赖日 K 的策略会在 L1 后自动对 Top N 候选做日 K 增强 - 日 K 数据源 `DAILY_SOURCE` 支持 `akshare`（默认）、`baostock` 或 `auto`，`auto` 会在 akshare 失败时自动降级到 baostock 作为免费兜底 ## Operations ### 1. 列出策略 ```bash alphasift strategies ``` ### 1.1 一键演示（无 API key） ```bash alphasift quickstart