Price Bands: How Markets Catch a Fat-Finger Trade

One extra zero, and the market notices

Picture a trader trying to sell one futures contract at 9,606. She types 9,066. Or 606. On a screen with rows of numbers, that happens more often than anyone would like to admit. Somewhere between her keyboard and the exchange's matching engine, a risk engine is supposed to notice that the price is nowhere near the market — and say no.

That "say no" is what this article is about. It goes by a few names: price banding, price limits, price collars, circuit breakers. They are not glamorous. They are, however, the reason a single bad click usually becomes a rejected order instead of a market event.

The basic idea: a safe zone around the market

A price band is simply a range around a reference price. Say the reference price is $100 and the band is 5%. The lower edge (the limit down) is $95 and the upper edge (the limit up) is $105. Orders that would trade outside that zone are rejected or otherwise handled according to the exchange's rules.

That word "reference" matters because the reference isn't always the last trade. On CME Globex, for futures, the reference price changes with the market state: the settlement price during pre-open, the indicative opening price once that exists, and the last traded price during continuous trading. If no last price exists, the settlement price carries on as the anchor. The band is recalculated as prices change, which is the point — it should move with the market, not sit still while the market walks away from it.

Why bother? Two problems, one tool

Price banding exists to frustrate two different kinds of trouble.

The first is the honest mistake. A limit buy order at a price well above the market, or a limit sell well below it, can transact at absurd levels and then need to be cancelled — a chain of unintended trades set off by one typo. Banding rejects the order on entry.

The second is more uncomfortable to say out loud: a market can be pushed. Someone might deliberately send aggressive orders designed to move prices. A band doesn't catch motive, but it does make it harder to shove a price through a wide empty stretch of the book in one go. That's the "market integrity" language exchanges use, and it is the more serious half of the job.

What banding does not do

This is where people get confused, and it's worth being precise.

Banding does not stop you from placing a passive order far from the market. A buy order below the market, waiting for prices to fall to your level, is perfectly normal trading — it adds depth rather than taking it. What banding stops is the order that tries to trade at a price that looks unrealistic right now.

Banding also doesn't lock up a market during a fast move. Because the band is anchored to the last price or the best bid or offer, it travels with the market as orders come in. It follows; it doesn't freeze.

And banding is not the same as a price limit. A price limit is a hard ceiling or floor for a whole session, usually set from the previous settlement. Once a product is locked at its limit, trading may continue in the sense that quotes exist, but participants can't transact beyond that range. Where both exist, daily trading limits take precedence over banding when rejecting bad prices.

Market orders, stop orders, and other moving parts

A market order doesn't have a price to check, so banding logic doesn't block it from being entered. What it can do is bound the damage. On Globex, an unfilled market order simply doesn't push past the best bid or offer in the book — it converts into a resting limit order at that level, and the remaining quantity waits. If a market order arrives already converted into a limit order at a price that would run through the book, that order gets rejected.

Stop orders get checked against the band too. If a stop order's limit price sits further from the trigger price than the band allows, it is rejected outright. The practical effect: if an accepted stop triggers, its resulting limit price is inside the band.

Options are a different animal, and exchanges treat them that way. A single strike may barely trade while the underlying futures contract moves a lot, which can make a stale reference price meaningless. So enhanced options banding uses a theoretical options price, or a blend of last price and theoretical value, and the band width can flex with the option's delta and moneyness. Out-of-the-money options tend to get narrower bands, in-the-money ones wider.

Layered defences: bands, limits and circuit breakers

Price banding is one layer in a stack that gets firmer as the situation gets worse.

In US equities, the Limit Up-Limit Down mechanism (LULD) keeps individual stocks inside a percentage band set from the average price over the preceding five minutes. If a stock hits the band and doesn't come back inside within 15 seconds, trading pauses for five minutes. Band widths are 5%, 10%, 20%, or the lesser of $0.15 or 75%, depending on the stock's price and whether it's a Tier 1 or Tier 2 NMS stock. Tier 1 includes securities in the S&P 500, the Russell 1000 and selected Exchange Traded Products. The bands double in the last 25 minutes of the regular session for all Tier 1 stocks and for Tier 2 stocks at or below $3.00.

Above that sit market-wide circuit breakers, tied to a single-day decline in the S&P 500. Levels 1 (7%) and 13% trigger a 15-minute halt if they hit before 3:25 p.m. ET; at or after 3:25 p.m., no halt. A 20% decline halts trading for the rest of the day, at any time.

Futures markets run their own versions. CME equity index price limits coordinate with the New York Stock Exchange's circuit breaker rules. And exchanges use very short-horizon tools too: velocity logic, which pauses trading when prices move more than a set number of increments in a few seconds, and interval price limits, which impose a hold period so the market can't trade outside a narrow range. These work over seconds, not sessions.

Brokers and trading platforms add their own layer. Trading Technologies, for example, lets an administrator set pre-trade price controls by ticks or percentage, either as a symmetric range around the market or a directional one where buy orders must sit below a maximum and sell orders above a minimum. The check uses the last traded price if it falls between the bid and ask; otherwise it uses the midpoint, or the first available price among ask, bid, settlement and close. These controls are enforced per account, not inherited from a parent account — a fine detail that surprises people.

Protection order types, briefly

Some exchanges offer "market with protection" and "stop with protection" orders. Instead of a bare market order, the order gets a protection price based on the best bid or offer plus or minus a product's protection point — typically around 50% of the non-reviewable range. If it can't fill fully inside that range, the remainder becomes a limit order at the protection price. That last step runs into banding: if the resulting limit price falls outside the band in an inactive market, the exchange may reject it. And because the remainder waits as a limit order, these orders can go partly unfilled in fast markets. That's a feature of the design, not a malfunction, but it does mean "market order" no longer means "guaranteed fill at any price."

Where the rules come from, and where they differ

Price controls were introduced in the wake of the 1987 crash and sharpened after the May 2010 "flash crash," when the Dow Jones Industrial Average fell roughly 1,000 points in under ten minutes, later traced in part to a large sell order from a mutual fund. LULD was proposed by US exchanges and FINRA in April 2011 and approved by the SEC on a pilot basis in May 2012.

On the European side, MiFID II Article 17 deals with algorithmic trading, requiring firms to keep records and give regulators details of their trading parameters and limits. ESMA has since published supervisory guidance emphasising pre-trade controls including price collars, mandatory hard blocks that automatically reject non-compliant orders, and calibration documented across trading, risk, compliance and IT. The guidance also notes that firms remain solely responsible for compliance even when algorithms come from third parties — and that using AI in trading isn't currently designated high-risk under the EU AI Act, though that scope is reviewed annually.

If you trade across jurisdictions, assume the numbers differ. Thresholds, band widths, reference prices and halt durations are set per venue and per product, and they change. Check the exchange's own current documentation rather than relying on any general description, including this one.

What this means for you

None of this is a trading strategy, and none of it guarantees anything. Bands and triggers are designed to reduce the frequency and severity of disorderly episodes, not to remove the possibility of losses, gaps or partial fills. What they do offer is a useful mental model: if you send an order that is far from the market, something upstream may simply refuse it — and if a market pauses, that's a control doing its job, not a malfunction.

It's a bit like a building's fire doors. You don't think about them when you walk through, and you shouldn't test them on purpose. But it's reassuring that someone else did the measuring.