A-Share Candlestick Pattern Matcher

Problem

In A-share technical analysis, the hard part is comparing a recent candlestick move against similar historical segments across the whole market. Manual screening is slow and hard to quantify, especially when estimating follow-through probability.

k线形态选股器 converts the target move into a daily return sequence, searches local data for similar historical windows, and outputs statistical references.

How it works

• Manual return input: use --returns to provide daily percentage changes for flexible full-market pattern search.
• Stock code input: read the last --window trading days for one stock, then match against the full market or its own history.
• K-line screenshot extraction: extract the return sequence from a daily chart image to create a pattern query quickly.

Results include similarity, average return over the next --after days, and up-move probability, with visual comparison output. The main script uses East Money local data; the older script can use tushare, subject to API rate limits.

Boundaries

Pattern matching is probabilistic, not deterministic. Higher similarity gives stronger reference value, but a 70% up-move probability still leaves downside risk. Short windows have more noise, while 30-60 day windows are better for trend-level judgments. Suspended, recently listed, or data-thin stocks may be skipped; ST stocks may still appear in full-market searches.