Most volume is noise — small orders from thousands of participants. Big-trade detection filters the tick stream for prints that are too large to be retail, then shows where and on which side they occurred.
Large participants cannot enter or exit positions without leaving marks in the tape. Even when execution algorithms slice parent orders, the child prints are frequently outliers against the instrument's normal trade-size distribution. Detecting those outliers — and clustering rapid sequences of them — surfaces the footprints of serious money.
Every NSE tick is compared against the instrument's own trade-size distribution (thresholds are per-symbol, not absolute). Qualifying prints are recorded with price, size and aggressor side; rapid same-side sequences are clustered into a single event with total quantity and notional value. The result feeds the Big Trades panel on charts and the daily big-trade statistics on VolumeLens stock pages.
Useful reads are relative, not absolute: a burst of large buy-side prints into a falling price, big trades concentrating at one level, or a session where large-trade volume skews heavily one side. These are facts about participation — most informative when combined with where they occurred and what the tape was doing overall.
It is relative: each symbol's threshold comes from its own trade-size distribution, so a big print in a midcap is much smaller than one in RELIANCE.
No. Exchange bulk/block deal disclosures are reported after the fact under regulatory thresholds. Big-trade detection works on the live tick stream during the session.
Iceberg-style execution hides intent partially, but the executed prints still hit the tape — clustering rapid sequences recovers much of the hidden size.