dsh-a-share-screener

Scanning five thousand-plus A-share tickers for flat-base-after-drawdown or faded-volume-heavy-limit-up patterns by hand is impractical. dsh-a-share-screener decomposes those technical setups into fifteen reusable atomic filters—deep drawdown, low percentile, flat base, platform breakout, MA stabilization, volume limit-up, cooldown pullback, industry clearance, market-cap band, and more—that you combine with AND, OR, and NOT into a single screening strategy. One natural-language prompt triggers a full-market scan and returns candidates with per-gate quantified evidence.

Three things set the plugin apart. First, three built-in strategies ship ready to use, yet the JSON predicate DSL lets you assemble any filter composition on the fly without writing a line of screening code. Second, all price math runs on a chained daily-return index, so splits and dividends never fake a crash or a bottom, and limit-up thresholds adapt per board (10 percent main, 20 percent ChiNext and STAR, 30 percent BSE). Third, data comes from Sina (default), Eastmoney, and Tencent—three free, token-less endpoints—with a local disk cache that keeps subsequent scans incremental and a cooperative cancellation path should you need to abort mid-scan.

It is aimed at individual quant hobbyists and small research teams who prefer local, privacy-preserving workflows over SaaS subscriptions, and at developers who want the atomic-filter-plus-expression-tree abstraction as a starting point for their own screening pipelines. Implementing the DataSource interface is all it takes to plug in a new vendor. The plugin is a technical screening aid: it surfaces candidates and near-misses with per-gate diagnostics; it does not provide investment advice.