--- name: washout-stock-screener description: Use when screening US or China stocks for a Washout Pattern, 主力洗盘 style setup, healthy pullback after a strong uptrend, bull flag, fake breakdown recovery, trend-continuation technical candidates, or daily Top 5 stock candidates with score, pullback, volume contraction, trend integrity, restart signals, sector heat, float size, and quality classification. Allows both profitable growth stocks and high-growth unprofitable stocks when revenue growth, gross margin, and balance sheet quality support the setup. Do not use mystical market-maker language; translate 洗盘 into verifiable trend, liquidity, and volatility structure. --- # Washout Stock Screener Use this skill to screen for short-to-medium-term trend continuation candidates after a strong advance and a controlled pullback. The system does **not** infer hidden intent by market makers. Treat “主力洗盘” as shorthand for: - strong prior trend - normal-to-sharp pullback - declining sell pressure - intact medium-term structure - possible liquidity re-accumulation before continuation Always frame outputs as technical candidates, not predictions or investment advice. ## Required Stance Avoid language such as: - 庄家控盘 - 主力故意洗盘 - 即将暴涨 - 必涨 - 吸筹阴谋 Prefer: - 趋势结构仍完整 - 回调期间抛压减弱 - 出现缩量整理 - 存在重新启动可能 - 资金行为接近趋势延续结构 ## Data Freshness When the user asks for current picks, daily screening, latest candidates, or market data, use up-to-date market data. Prefer reliable sources such as Yahoo Finance, Finviz, TradingView, Alpha Vantage, Polygon, IEX Cloud, Financial Modeling Prep, Tiingo, exchange data, ETF data, or the user's provided dataset/API. If live data is unavailable, say so and provide either: - a reproducible screening recipe, or - an analysis of user-provided tickers/data only. ## Market Scope Default to US equities unless the user requests A-shares or another market. US market preference: - better data quality - stronger liquidity - clearer fundamentals - cleaner technical structures A-share usage: - suitable for high-sentiment themes such as AI, semiconductors, robotics, biotech, low-altitude economy, and new energy - apply stricter liquidity and breakdown filters because false breakouts and liquidity traps are more common ## Hard Filters Apply these before scoring: | Filter | Default | | --- | --- | | Market cap | 300M to 20B USD | | Preferred market cap | 1B to 10B USD | | Quality gate | A-class profitable growth or B-class high-growth unprofitable | | Float size | < 300M shares preferred | | Aggressive float size | < 100M shares | | Average daily dollar volume | > 10M USD | | Prior 3-month performance | > +30% | | Stronger trend confirmation | 6-month performance > +50% | | Recent pullback | 10% to 30% from recent swing high | | Preferred pullback | 15% to 22% | | Volume contraction | recent 10-day avg volume < previous 30-day avg volume | | Preferred volume contraction | > 20% | | Trend integrity | price > 50-day MA or 50-day MA slope positive | | Stronger trend integrity | 200-day MA trending upward | ### Quality Gate Do not use `PE > 0` as the only quality filter. Many strong trend candidates in AI, biotech, SaaS, robotics, and space are not currently profitable. The system should allow two classes: **A-class: profitable growth stocks** - PE > 0 - Optional stronger band: 5 < PE < 60 - Best fit: semiconductors, software, AI infrastructure, mature growth stocks - Typical behavior: cleaner institutional trend, lower relative volatility **B-class: high-growth unprofitable stocks** Allow missing or negative PE only when quality data supports the growth story: - revenue growth > 25% - gross margin > 40% - cash > debt Best fit: AI growth, biotech, high-growth SaaS, robotics, space. B-class candidates are often more volatile and sentiment-driven, but they can produce classic washout structures because market expectations, theme heat, and liquidity matter more than current accounting profit for this specific setup. If one or more B-class metrics are unavailable, label the quality gate as `Incomplete` and either downgrade the score or place the ticker in an observation bucket instead of treating it as fully qualified. ### Float Size Float is an important washout-pattern factor: - Preferred: float < 300M shares. - Aggressive: float < 100M shares. Smaller floats can produce sharper volatility, fake breakdowns, and fast reclaims, but they also increase liquidity and manipulation risk. Always combine float with the dollar-volume filter; do not accept thinly traded micro-liquidity names simply because the float is small. Preferred sectors: - AI software - AI infrastructure - semiconductors - biotech - cybersecurity - cloud - robotics - space - fintech Generally avoid unless the user requests otherwise: - utilities - banks - traditional energy - defensive staples ## Washout Structure Checks Evaluate each surviving stock for these structures: 1. **Recent Pullback** - Pullback should be 10% to 30%. - Under 10% is often too shallow. - Over 30% may indicate trend damage. 2. **Volume Contraction** - Recent 10-day average volume should be below the previous 30-day average volume. - A contraction above 20% is preferred. - Interpret as reduced sell pressure, not hidden intent. 3. **Trend Integrity** - Price above 50-day MA is preferred. - If price briefly loses the 50-day MA, require fast reclaim or positive 50-day MA slope. - Prefer rising 200-day MA. 4. **No Heavy Distribution** - Penalize large red candles on unusually high volume. - Exclude obvious event-driven collapses, earnings breakdowns, or repeated high-volume selloffs. 5. **Stabilization or Restart Signal** - shrinking-range consolidation - long lower wick near support - support undercut followed by quick reclaim - fake breakdown recovery - high-volume green candle after contraction - breakout from flag/handle range ## Pattern Labels Use the clearest matching label: - Bull Flag - Cup and Handle - High Tight Flag - Fake Breakdown - Shrinking Base - Support Reclaim - High-Base Consolidation - No Clear Pattern If a chart is ambiguous, say so. Do not force a named pattern. ## Sector Heat Prefer strong stocks in strong sectors. Estimate sector heat from available data: - sector ETF performance - sector ETF relative strength vs SPY/QQQ - sector ETF volume expansion - fund flow data if available - recent news/theme momentum Useful US proxies: | Theme | ETF proxies | | --- | --- | | Broad market | SPY, QQQ, IWM | | Semiconductors | SMH, SOXX | | Software/cloud | IGV, WCLD | | Cybersecurity | CIBR, HACK | | Biotech | XBI, IBB | | Robotics/automation | BOTZ, ROBO | | Space | ARKX, UFO | | Fintech | FINX, ARKF | ## Scoring Score each candidate out of 100: | Category | Points | Guidance | | --- | ---: | --- | | Prior trend strength | 20 | 3M/6M performance, relative strength, MA slope | | Pullback health | 25 | pullback depth, volume contraction, controlled selling | | Trend integrity | 20 | 50/200 MA structure, higher lows, support retention | | Stabilization/restart | 20 | reclaim, fake breakdown, volume green candle, tight range | | Sector heat | 15 | sector ETF strength, theme/news/fund-flow support | Suggested interpretation: - 85 to 100: high-quality continuation candidate - 75 to 84: strong watchlist candidate - 65 to 74: acceptable but needs confirmation - below 65: usually exclude from Top 5 ## Feedback Rules From 2026-05-26 Review The 2026-05-26 Top 10 review showed a mixed but useful signal: equal-weight performance through 2026-05-29 was about +1.0%, with strong winners such as DAVE, VRNS, and INGM, but breadth was weak and AMSC suffered a large intraperiod drawdown. Use these lessons to tighten future ranking: - Prefer concentrated Top 5 output over broad Top 10 output. - Rank higher when strong individual setup aligns with strong sector or theme heat. - Rank higher when the stock has small-to-medium float, meaningful dollar volume, and clear volume contraction during the pullback. - Require stronger confirmation for candidates from weak sectors, defensive sectors, or unrelated themes. - Downgrade non-core-theme stocks even when their technical score is high; materials, cyclicals, defensive consumer, and other non-preferred themes need stronger sector confirmation before entering the final Top 5. - Downgrade candidates with nearby earnings, event risk, heavy intraperiod drawdown history, or obvious news-driven gap risk. - Exclude or heavily downgrade candidates with earnings or known binary events within the next 10 trading days unless the user explicitly asks for earnings-event trades. - Do not let a technically clean pullback override weak sector participation or missing restart evidence. - When earnings/event filtering and non-core-theme downgrades leave fewer than five high-quality candidates, return fewer than five instead of forcing weak fillers into the final list. - Apply a broad-market regime filter: if SPY, QQQ, and/or IWM are in clear pullback mode, below key moving averages, or showing weak risk appetite, shrink the final list aggressively and allow zero candidates instead of forcing exposure. ## Output Format For daily screening, return a Top 5 table: | Ticker | Sector | Score | Pullback | Volume Change | Trend Status | Pattern | | --- | --- | ---: | ---: | ---: | --- | --- | | RXRX | AI Biotech | 87 | -18% | -35% | Intact | Bull Flag | When data is available, add quality and float columns: | Ticker | Sector | Quality | Float | Score | Pullback | Volume Change | Trend Status | Pattern | | --- | --- | --- | ---: | ---: | ---: | ---: | --- | --- | | RXRX | AI Biotech | B-growth | 93M | 87 | -18% | -35% | Intact | Bull Flag | Then add a compact note for each candidate: - **为什么入选**: trend, pullback, volume, sector, restart evidence. - **当前风险**: volatility, earnings proximity, weak sector, failed reclaim, liquidity, news/event risk. - **质量分类**: A-class profitable growth / B-class high-growth unprofitable / Incomplete. - **结构判断**: 健康洗盘 / 高位震荡 / 出货风险 / 二次启动候选. End with a short caveat that these are technical screening candidates and require risk controls. ## Risk Filters Before ranking highly, check: - broad-market regime: SPY/QQQ/IWM trend, pullback state, and key moving averages - earnings within the next 10 trading days - major financing/dilution risk - recent regulatory or legal event - stock below key moving averages with no reclaim - repeated high-volume breakdowns - dollar volume below threshold - float too small relative to dollar volume - B-class quality gate missing revenue growth, gross margin, or cash/debt support - extreme gap risk after news Mention important risks rather than burying them. ## Optional Advanced Factors Use when data is available or the user asks for a deeper version: - relative strength vs SPY, QQQ, and sector ETF - short interest and days-to-cover - float size and free-float market cap - revenue growth, gross margin, cash/debt, cash runway - options flow, unusual call activity, IV expansion - earnings proximity - ATR-adjusted stop distance - distance from VWAP or anchored VWAP - institutional ownership or fund-flow data ## Implementation Notes If building code or a screener: - Make filters configurable. - Separate hard filters from scoring. - Store intermediate metrics so the user can audit why a stock passed. - Prefer structured financial APIs over scraping when available. - Use adjusted daily OHLCV for technical calculations. - Calculate pullback from a recent swing high or rolling 20- to 60-day high; state which method is used. - Use dollar volume, not share volume alone. - Implement the quality gate as two explicit paths: A-class profitable growth and B-class high-growth unprofitable. - Treat missing B-class fundamentals as incomplete evidence, not automatic pass. - Capture float size when available and score it together with liquidity. - Return rejected candidates with reasons when debugging.