Delivery percentage explained

Two stocks trade 10 lakh shares today and both close up 4%. One of them delivered 22%, the other 71%. I want to know which is which before I look at the chart, and for years I had no screener that would let me ask.

Delivery percentage is the fraction of a day's traded volume that actually moved between demat accounts instead of being bought and sold back inside the session. NSE publishes it for every stock, every day. Most screeners park it in a column you can sort by, and leave it there.

Go back to the two stocks. In the 22% name, nearly four-fifths of the volume was intraday churn: positions opened and squared off before the close. In the 71% name, seven of every ten shares traded went home in somebody's account overnight. The price move is identical. The two days are not the same day.

Read it against the stock's own norm

My first version of this scan used an absolute threshold, delivery_pct > 60. I ran it every evening for about a fortnight and it kept handing me the same sleepy large caps, which is when I worked out why. Liquid large caps routinely run 40 to 60% delivery on an ordinary day. Some F&O-heavy names live near 20% and never leave. Sixty per cent is not a high reading, it is just a number, and whether it means anything depends entirely on the stock.

So every delivery scan here is built on deviation from the stock's own baseline instead:

where delivery_pct > 1.5x avg(delivery_pct, 20)

A move from a usual 35% to 70% is an event. So is 15% to 30%, in a name that normally delivers nothing. The relative form catches both and the absolute one catches neither properly.

The four patterns I keep

A delivery surge on a day the price also moved is the plain confirmation case: Delivery surge.

Rising delivery while the price does nothing is the one I find most interesting, because somebody is quietly taking size home over several sessions and the chart has not admitted anything yet, which is the only moment in the whole sequence when you are early rather than informed. That is Quiet accumulation. It also has the most false starts of anything here, since "nothing is happening" describes a great many stocks that are simply dull.

High delivery on a large turnover day is a size filter as much as a conviction filter. Seventy per cent of a ₹50-crore session is not one trader with a hunch: Institutional-grade delivery.

Then the two that read the other way. Delivery tells you how committed the trade was and says nothing about direction, because determined sellers deliver too. Heavy delivery on a down day is the distribution profile, which is Distribution warning. A price spike on unusually low delivery is the churn signature: Low-delivery spike.

What the number will not do for you

In stocks with liquid derivatives the cash market is a smaller window into the real activity, because arbitrage and hedging flows pass through it. Delivery reads cleanest outside the F&O universe, or against the stock's own baseline.

The data arrives with the exchange's end-of-day reports and is missing for older history. There is no intraday version and there cannot be.

One high-delivery day is one day. The patterns worth trading are sustained ones, which is why three days of rising delivery and five days above 55% exist as separate scans.

Test the folklore

Delivery analysis is heavy on received wisdom and light on published evidence, which makes it good material for replay. Open Quiet accumulation and read the hit-rate panel: every session in the last 250 that the scan matched, and what those matches did over the next 1, 5 and 20 days. It is a sketch, close to close, with no costs or slippage in it. It still settles the argument faster than another article will. The rest of the family is at the delivery scans hub.