Eligible Universe · 基础股票池

策略并不扫描全部美股，而是只扫描约 1,850 只市值、股价和流动性达标的普通股。微型市值和低流动性股票容易因异常成交、巨大价差和少量交易扭曲技术指标，因此默认排除。
Strategies do not scan every US-listed security. They scan about 1,850 common stocks that meet market-cap, price and liquidity thresholds. Micro-cap and illiquid stocks are excluded because abnormal trades, wide spreads and thin volume distort technical signals.

1. 美国交易所上市的全部证券约 12,800 只 · All US-listed securities, about 12,800
2. 排除 ETF、ETN、基金、权证、优先股、SPAC 等非普通股；在美上市的外国公司 ADR（如中概股）、REIT 和合伙制公司保留 · Exclude ETFs, ETNs, funds, warrants, preferreds and SPACs; foreign-company ADRs, REITs and partnerships are kept
3. 保留满足上述市值、股价和成交量门槛的股票 · Keep stocks meeting the thresholds above
4. 需有约一年以上的交易历史（新上市股票暂不纳入），且有最新交易数据（停牌、退市、被收购的股票排除） · Require about one year of trading history (recent IPOs join later) and current price data (halted, delisted or acquired stocks are removed)
5. 最终约 1,850 只进入每日扫描 · About 1,850 stocks are scanned daily

Multi-Factor Models · 多因子模型

Discover 策略通常结合趋势、动量、成交量、波动、价格结构、相对强弱和背离等多个信号，而不是依赖单一指标。
Strategies generally combine multiple independent technical dimensions rather than relying on a single indicator.

Top Picks · 精选结果

每个策略可能筛出更多符合条件的股票，但用户端每日最多展示强度最高的 Top 5。若当天只有 2 只真正符合要求，则只展示 2 只，不会为了凑数而降低标准。
Each strategy displays at most five top-ranked qualified stocks per day. Fewer than five may be shown when fewer stocks meet the strategy requirements.

Strategy Confluence · 策略共振

共振基于各策略的全部合格结果（Qualified Matches）计算，不仅限于用户端展示的 Top 5。看多与看空策略分别计算，不将方向相反的信号混合计数。
Confluence is calculated from all qualified matches across independent strategies, not only the displayed Top 5. Bullish and bearish signals are evaluated separately.

Outcome Tracking · 结果追踪

策略筛选结果会持续追踪后续市场表现，用于展示 5D、10D、20D、60D 等真实历史结果。样本量会同时展示，历史结果不代表未来表现。
Strategy signals are continuously evaluated against subsequent market outcomes. Sample size is shown alongside historical performance.

Proprietary Parameters · 核心参数

页面公开策略目标、方向和主要技术维度，但具体阈值、内部权重和核心组合规则属于 20cm.ai 的策略模型，不公开展示。
Strategy objectives and major technical dimensions are described publicly, while proprietary thresholds, scoring weights and internal model rules remain private.

Continuous Improvement · 持续优化

随着更多市场数据和 Outcome 样本积累，股票池过滤条件、模型结构和策略排序方法可能持续优化。
Filters and methodology may evolve as additional market data and outcome evidence become available.