The print nobody meant to make

You are not at the screen at 11.04. The position is a small midcap, bought two days ago at ₹296, with a stop-loss order resting at ₹291 — comfortably below a fortnight of lows, chosen deliberately so that ordinary noise would not reach it. The phone buzzes at 11.06 with a fill: sold, 640 shares, ₹271. You open the chart and the morning looks exactly as you left it, a quiet range between ₹297 and ₹303, except for one candle with a spike of a wick hanging off the bottom of it, down to ₹268, on a body that begins and ends around ₹299. The stock is at ₹300 while you are looking at it. Nothing was announced. Nobody sold anything. For about four seconds, one order took the price down twenty-nine rupees, and in those four seconds it collected your stop on the way past.

This is a freak trade, and the first thing to understand about it is that it is not an error in the ordinary sense. No system malfunctioned. The exchange matched an order against the orders that were sitting there to be matched against, in the order the rules require, and the trades that resulted are real trades: they settled, somebody paid for them, and they are in the exchange’s own record of the day. The chart is not lying to you. The chart is telling you the truth about something that should not have happened.

How one order gets to a price nobody was offering

A [[Market order]] does not have a price. It is an instruction to trade the quantity at whatever prices are available, and it is matched against the resting limit orders in the [[Order book]], best price first, until the quantity is exhausted. In a liquid name the book is thick — there are thousands of shares bid within a few paise of each other — so a market order for 5,000 shares clears against the top few rungs and moves the price a little. In a thin name the book is a ladder with most of the rungs missing. A market sell order for 5,000 shares meets 300 at ₹297, then 200 at ₹294, then nothing at all until somebody’s speculative bid at ₹268 that has been sitting there for a week hoping for exactly this. The order walks all the way down and prints every rung it touches.

• The size was a mistake, or the account was. A quantity typed with an extra zero, an algorithm with a bad parameter, or a client whose position was being liquidated by a risk system that was told to sell now rather than sell well.
• The book was empty for a legitimate reason. Market makers pull quotes around scheduled events, and in an illiquid name there was never much of a book to pull. The same order that would be absorbed at 2.30 in the afternoon walks the ladder at 9.20 in the morning.
• Somebody wanted the print. A price touched briefly at a chosen moment can trigger a settlement value, a stop cluster, or a level a lot of people are watching. This is a market-abuse question rather than a charting one, and it is not for you to adjudicate — but the resulting bar behaves identically to the accidental kind, which is the only part that concerns your indicators.

What stops it, and why what stops it is not enough

Exchanges do not let an order walk indefinitely. Most securities carry a daily price band — a fixed percentage away from the previous close, beyond which orders are simply not accepted — and securities that do not carry a fixed daily band instead sit inside a [[Dynamic price band]], a narrower operating range around a reference price that the exchange can widen in steps once it sees genuine two-sided interest at the edge. Brokers layer their own, tighter, order-value and price-range checks on top. Between them, these controls put a floor under how far a single order can travel.

Nor should you count on the trade being undone. Exchanges do have a route by which a trade can be annulled, but it is exceptional, it has to be applied for within a tight window, and the presumption throughout is against interfering with a concluded trade — because somebody on the other side of it has a perfectly good bargain that they are entitled to keep. Plan on the working assumption that the trade stands, the fill is yours, and the print is in the historical series permanently.

Which of your numbers eat the wick, and which do not

This is the practical half of the lesson, and it is much simpler than it sounds. Every indicator you use takes some subset of the four numbers in a bar. If it only ever reads the close, a wick cannot touch it. If it reads the high or the low, the wick goes straight in.

Telling a fold from a real reading

You cannot act on any of this until you can distinguish a freak trade from a genuine violent move, and there is no single test that settles it. There are four, and they agree with each other far more often than not.