There’s been a lot of discussion lately around market structure regulatory reforms and what they mean for market participants. One topic that stood out was the role of the National Best Bid and Offer (NBBO) at the recent Order Protection Rule roundtable.
As part of that conversation, several participants raised questions about locked and crossed markets — what they are, why they happen, and why they matter? In this blog, we’ll break down these concepts and explain their significance to investors.
What? When? Why?
A locked market occurs when the best bid and best offer are equal across different exchanges. For example, a buyer on one exchange wants to pay $10, but a seller on another exchange is willing to sell for $10, and yet no trade is happening.
A crossed market happens when the best bid is higher than the best offer. For example, a buyer on one exchange wants to pay $10, but a seller on another exchange is willing to sell for $9.99, and still no trade is happening.
Rule 611 of Regulation NMS, also known as the “order protection rule,” requires exchanges and other trading venues to prevent trade throughs. That is meant to ensure all investors trade at the best prices available in the market, regardless of which exchange has that quote.
In addition, Rule 610 requires exchanges, which advertise prices, to avoid locking and crossing markets with their displayed quotes.
However, that doesn’t mean the NBBO itself can’t appear locked or crossed.
In fact, when we look at S&P 500 stocks, we see that, on average, each stock is locked for around 2.5 seconds each day. Although, importantly, markets are crossed far less – with an average of just 4.2 milliseconds each day.
Chart 1: Locked and crossed markets are rare and last for a small fraction of the trading day
The spread matters!
Importantly, not all stocks are average!
In fact, what the data shows is that it is much more likely that a tick-constrained stock will be locked more often. As the data in Chart 2 also shows, lower-priced stocks are more likely to be tick constrained.
That should be no surprise. After all, a tick-constrained stock is already all but 1 cent away from being locked – and typically has even more depth and longer queues than other stocks.
In contrast, the data shows that higher-priced stocks usually have wider spreads with multiple cents (ticks) between the bid and offer. As traders can improve prices by several ticks without hitting the far touch, it is much less likely that a quote improvement will lock the market.
Chart 2: Bid-ask spreads are intimately correlated to locked and crossed quotations. The tighter the spread, the stronger chance to observe a locked or crossed quotation.
Another conclusion we can draw from this data is that the U.S. Securities and Exchange Commission’s (SEC) proposed minimum tick size change (where tick-constrained stocks would have half-cent ticks) should reduce the number of locked and crossed quotes in the most affected stocks
How long do locked markets stay locked?
The data above shows the total duration of locked and crossed markets per stock. In short – this proves locks and crossed markets do happen.
But from the perspective of whether OPR works (or is still needed in an all-electronic market), it is also important to understand how long each one typically lasts.
The data shows it is brief, but not “deminimus.” On average, the locked markets last 5.5 milliseconds, while crossed markets exist for an even shorter time, just 0.8 milliseconds.
Chart 3: Locked & Crossed are corrected very quickly
Some locks might only happen on the SIP
It’s also important to understand that messages take around 0.2 milliseconds to travel from one venue to another. That means even though the SIP shows a lock or cross, it might not be true in other locations.
For example, locked markets seem to indicate that arbitrage profits are available. However, academics who accounted for how long it would take even the fastest traders to route orders to capture the trades find that crossed markets are mostly gone before their order can trade.
Holsten, Pierson and Wu (2023) looked at this in detail. Using exchange timestamps from the SIP, they reconstructed what the NBBO would have looked like in Mahwah, Secaucus and Carteret after adjusting for latency. They found that once latency is factored in, trades matched to locked spreads drop by about 80% (blue line below). That suggests that geographic latency may actually explain a lot of the locked and crossed quotes we see here today.
Chart 4: The number of locked and crossed quotes drops significantly once we account for geographic latency
What does this mean?
With the SEC thinking about the economics of the NBBO, and some institutions even suggesting that we might not need an NBBO at all, data like this is important.
This analysis shows that markets do lock (and sometimes cross), but current rules mostly make the NBBO return unlock much faster than a human can blink. We also have evidence that market makers and arbitrageurs act very quickly to uncross markets.
What we can’t know is what happens if the Order Protection Rule goes away. Locks could become more common, especially if traders want to avoid take fees, but retain queue position on resting orders. A secondary question is whether, in a market like that, midpoint orders, price improvement and 605 metrics still make sense.
These are important questions for market participants and the SEC to weigh as they rethink Regulation NMS.