What "dark" means here.这里的「暗池」指什么。 The share of a stock's daily volume reported to FINRA's off-exchange facilities (dark pools + broker internalisers) rather than the lit exchanges. This is settled end-of-day tape — NOT live prints, and NOT direct dark-pool trade records. "Off-exchange share" = FINRA-facility volume ÷ consolidated day volume (exact date match).个股当日成交量中上报至FINRA场外设施（暗池+券商内部化）而非公开交易所的比例。此为结算后的日终成交数据——非实时成交，亦非直接的暗池成交记录。「场外占比」=FINRA设施成交量÷综合日成交量（精确日期匹配）。

Why there is no direction call.为何这里没有方向判断。 Off-exchange volume hides who is trading and which side initiated, so it cannot call direction on its own — and roughly half of it is short-marked by default, which is a feature of how market makers hedge rather than evidence of bearish intent. This desk therefore reports what was observed: how far a name's off-exchange share ran from its OWN norm, how long the run has lasted, where it printed, and what the price did over the same stretch. It does not tell you which way the stock goes next.场外成交隐藏了交易方身份与主动方，因此本身无法判断方向——且其中约一半默认被标记为卖空，这是做市商对冲机制的特征，而非看空意图的证据。因此本数据台只报告观察到的事实：个股场外占比偏离自身常态的幅度、已持续多久、成交发生在哪里，以及同期股价的表现。它不预测股价接下来的走向。

How a name gets on the board.个股如何进入焦点板。 Its off-exchange share must be both well above its own norm (≥1.5σ against a trailing baseline that EXCLUDES the session being measured) and material (≥40% of that session's volume). Ranking is by distance from each name's own norm, not by raw level — a stock that always prints 55% off-exchange is not news, while one that normally prints 28% and printed 42% in the measured session is. Names without enough comparable history show no score at all rather than a made-up one.其场外占比必须既明显高于自身常态（相对于一个不含本次测量交易日的滚动基准 ≥1.5σ），又足够显著（≥该交易日成交量的40%）。排序依据是偏离自身常态的程度，而非绝对水平——一只长期场外占比55%的个股并非新闻，而一只常态28%、本次测量交易日却达42%的个股才是。历史可比数据不足的个股不显示评分，而非给出杜撰的数值。

Dark pools are the smaller half.暗池只是较小的一半。 Off-exchange volume is registered dark pools (ATS) PLUS wholesaler internalisation — broker-dealers filling order flow, largely retail, against their own book. The second is the bigger half: across large names only about 17–31% of off-exchange volume actually reaches a dark pool. Each card shows that split, so "trading in the dark" is never confused with "institutions are accumulating".场外成交由注册暗池（ATS）与批发商内部化两部分构成——后者是券商以自有账簿承接订单流（多为散户订单）。后者才是较大的一半：在大型个股中，实际进入暗池的场外成交通常仅占约17–31%。每张卡片都显示这一构成，以免将「暗池成交」误读为「机构在吸筹」。

Who was on the other side.成交对手方是谁。 Off-exchange volume that does not reach a dark pool was internalized by a firm, and FINRA names that firm. We map each firm to its PRIMARY BUSINESS — retail wholesaler, retail broker, or bank/institutional desk — which separates retail order flow from institutional risk transfer. This is a labelled heuristic on the firm, not a measurement of who was actually trading, and never a direction: large firms run several businesses at once. FINRA reports small firms only as one "De Minimis" aggregate (about a third of internalized volume market-wide), and that plus any firm we do not classify is shown as unattributed rather than pushed onto a side.未进入暗池的场外成交由某家机构内部化完成，FINRA会公布该机构名称。我们按其主营业务将其归类为散户批发商、零售券商或银行／机构交易台，从而区分散户订单流与机构风险转移。这是针对机构主营业务的标注式启发规则，并非对实际交易者的测定，也绝非方向判断：大型机构往往同时经营多项业务。FINRA仅将小型机构合并为一个「De Minimis」汇总项（约占全市场内部化成交的三分之一），该部分及任何未归类机构均标注为「无法归类」，而不会被归入任何一方。

Venues.交易场所。 FINRA OTC Transparency weekly per-ATS breakdown. The publication lag is measured and printed next to the week label rather than assumed — it runs longer than the nominal 2–4 weeks. Bar length is each venue's share of that week's total ATS volume; the arrow is the week-over-week change in percentage points.FINRA场外透明度每周各ATS分项。公布延迟为实测值并标注于报告周旁，而非按名义值假定——实际延迟长于名义的2–4周。条形长度为各场所占该周ATS总成交量的比例；箭头为周环比变化（百分点）。