Every NSE terminal shows the 5-best-price window: five bid levels, five offer levels, with quantities and order counts. Read naively it misleads; read properly it is a running poll of nearby intention.
Each side shows price, total quantity resting at that price, and the number of orders composing it. Quantity ÷ orders gives average order size — 5 orders totalling 50,000 shares is a very different crowd from 500 orders totalling the same. The totals row aggregates all five visible levels per side.
"More total bids than offers = it will go up" fails for two reasons. First, the book shows only passive intention — the aggressive orders that actually move price are invisible until they execute. Second, visible size is optional: participants can display a fraction of their true order or park misleading size they intend to cancel. Depth is evidence, not testimony.
Behavioural reads outperform snapshot reads: size that holds and reloads when repeatedly hit (genuine interest — the visible edge of absorption); size that vanishes on approach (decoration); spread widening under stress (liquidity withdrawing). The full DOM ladder makes these behaviours much easier to track than the 5-level snapshot.
The standard retail data product carries 5 levels per side. Fuller depth exists at the exchange; 5 levels is what standard feeds distribute.
Weakly at best — it shows passive, cancellable intention only. Trades (executed aggression) move price, and those never appear in the book until done.
It reveals crowd composition: many small orders versus a few large ones at the same price carry different information about who is present.