OPR: All or Nothing or Something in Between?

On June 11, the U.S. Securities and Exchange Commission (SEC) released its new proposal regarding the Order Protection Rule, or Rule 611, in Reg NMS.

The proposal in concept is simple: Rescind rule 611 and a few other related parts of Reg NMS. Specifically, the so-called order protection rule (OPR) and Rule 610(e), which prohibits exchanges to lock or cross markets. 

The SEC said in their proposal that these rules are adding to routing costs and market fragmentation. Essentially, OPR required brokers to route orders to the venue displaying the best price, which led to excessive market fragmentation. That, in turn, led to complex order types (like ISOs), and expensive routing infrastructure that added cost without obviously benefitting investors. 

A quick recap of Reg NMS

NMS is a series of rules that set out how trading works in the U.S. Read this blog for a quick refresher of what Reg NMS does and this blog to refresh on what all the acronyms used in the market actually mean.

The NMS rules are summarized in the table below. You can see that most set the market up as fair and equal to all kinds of traders. 

• Pre-trade: Information about prices and orders and trades just done must be shared with SIPs who then share with all investors.
• Trade: Investors' orders must be displayed, and all investors must be able to access those quotes to trade.
• Post-trade: After all that, there are also post-trade calculations and disclosures to ensure investors' trades were executed well, and the NBBO was respected. 

What has changed with this proposal is shown below with strike through text. 

Table 1: Summary of currently implemented NMS rules

^*expanded

Order protection required a method for everyone to know what the protected quotes are, making the SIP a “pre-trade” tape, which, as we’ve said before, physics makes impossible. That’s because OPR essentially forces brokers to comply with the NBBO and route to markets with the best prices, regardless of whether the broker believes those orders still exist or not.

Bloomberg survey says investors prefer a NBBO

This change was not a surprise — the SEC has held roundtables with academics, traders and investors to canvas opinions. In the SEC roundtables, many investors seem to value the way the NBBO protects them from bad trade prices and lets them easily calculate their execution performance.

A recent Bloomberg paper confirmed this. It also showed that investors prefer a “Canadian” solution to the OPR problem. Rather than eliminate OPR completely, they prefer raising the bar for setting a protected quote higher. (Note that in Chart 1, the more red means more worry about that kind of Reg NMS change.)

Chart 1: Bloomberg survey of U.S. Buyside traders' views on eliminating OPR

In Canada, exchanges with less than 2.5% market share don’t qualify for the NBBO (if this happened in the USA, presumably those exchanges would also not qualify for SIP revenue, which also adds to fragmentation). 

In short, investors said they prefer a “National Big Best Bid and Offer,” which we will call an NBBBO. 

How big should the new “B” in NBBBO be?

The U.S. equity market now has 17 active exchanges. That's more than double the number from 20 years ago, and recent filings show it’s likely to rise. So, the argument goes: More venues mean more competition, more competition means tighter spreads, tighter spreads mean better prices for investors.

It's a compelling theory. But when you look at the data, the picture gets more complicated.

To be fair to all venues, the data in Chart 2 looks at continuous trading – that’s where all venues are open, and price discovery mostly happens. Although, we should also highlight that today’s exchanges’ lit trading only contributes around 30% of continuous volume – that’s because the majority of trading during the day happens off-exchange – at prices referencing the NBBO.

Chart 2: Five exchanges account for more than three-quarters of on-exchange volume

The data shows that listing exchanges lead lit volume. Nasdaq leads at 27.7%, followed by ARCA at 19.3%, and NYSE at 11.0%. 

After that, the drop-off is steep. Three exchanges sit between 5% and 10%. One clears 2.5%, while 10 exchanges fail to meet the de minimis 2.5% market share that currently exists in Canada.

How much worse would an NBBBO be?

Given the results of the survey above, it’s interesting to see just how much worse the NBBBO would get if less exchanges counted for the NBBO. 

The data in Chart 3 shows that adding more venues with 10% market share or less does result in tighter NBBBO spreads (Medium liquidity stocks trade between $300,000 and about $7mln per day):

• Liquid stocks: Spreads improve 14% (to 9.4 basis points).
• Medium liquidity stocks: Spreads improve 13% (to 50.9 bps).
• Low liquidity stocks: Spreads improve 70% (to 148.9 basis points). 

However, the data also shows that almost all of the spread improvement comes from exchanges with more than 5% market share. If we add venues with less than 5% market share, we find that:

• Liquid stocks see spreads improve from 9.5 to 9.4 basis points (just 1% of all the spread improvement).
• Medium liquidity stocks* see spreads only improve from 52.0 bps to 50.9 bps (just 2% of all the spread improvement).
• Low liquidity stocks see spreads improve from 151.9 to 148.9 basis points as you add more venues (just 2% of all the spread improvement – in the stocks where spreads are widest and competition for quotes would help those companies costs of capital the most).

The data implies that the smallest venues simply aren't adding much to the best bid or offer on liquid stock – although the data above, and SIP revenue data confirms many are copying the NBBO quote.

Chart 3: NBBO bid-ask spread at different volume market-share thresholds

Where do you draw the line?

As the SEC highlighted, every exchange added to the market adds routing complexity and, potentially, connection costs.

Brokers and smart order routers have to decide, in real time, whether it's worth reaching out to every venue on the tape. That calculation involves latency, connectivity, and the probability of finding a meaningful improvement in price. For that, there is an economic cost.

And yet, this data suggests, for many stocks there is little economic benefit.

Although perhaps the bigger question for commenters to grapple with is, what happens next? Once you remove the OPR, is the NBBO still an appropriate execution benchmark?

And for those that use the SIP for NBBO, does that even matter?