Tight NBBO Matters to Off-exchange Trades Too

A tight and actionable NBBO creates a lot of positive benefits for investors – lower trading costs, more liquidity and lower costs of capital – something economists sometimes call “positive externalities.” 

As we discuss today, that’s even true for those who don’t trade on exchange.

The mid isn’t always the middle

We have already found that a better odd lot quote is very likely for high-priced stocks with multiple ticks in between the NBB and the NBO. 

We also know that getting ticks right can improve valuations while eliminating round lots would tighten spreads, which should reduce trading costs for investors.

A recent academic study compared round lot and odd lot quotes in a different way and found that often, the true mid-point in the market isn’t the middle of the NBBO. For example, in the market snapshot below (Chart 1), the NBBO midpoint is $3,112.29, and a number of odd-lots are actually bid higher than that, which means midpoint fills during that time may cross the true market. 

The study used this detail to find that the odd lot quotes were also better at predicting future prices than the NBBO. Then they used this knowledge to simulate a simple trading bot that was able to earn positive returns with that knowledge.

Chart 1: Odd-NBBO can be very different and directionally more accurate versus NBBO midpoints

As an aside, the way this paper showed the snapshot of the market is also revealing. With real odd lot quotes resting inside the NBBO, colored by exchange (Chart 1), we can clearly see a problem with the new MDI rule for dynamic round lots. Under that rule:

• A stock above $1,000 would have 10-share round lots (dashed blue line), designed to ensure the protected quote adds to $10,000 at least while trying to solve the round lot problem and compress spreads.
• Each exchange will also aggregate their odd lots into the “worst odd price” round lots, so orders smaller than 10 shares still count (eventually).

However, as Chart 2 shows, with 16 exchanges all showing different odd lots at different prices, even the aggregated 10 shares from each exchange could bear little resemblance to the real best prices in the market for a $10,000 order to trade with. In the example above, the new NBO could be either the:

• Offer at $3,113.65, twelve levels, and 46 cents, from the inside odd lot at $3,113.19.
• Or the three combined small lots on EDGX (green bars) that would set an NBO at a slightly better $3,113.57

But NMS-II is not what we’re talking about today.

Mid is often far from the middle

Knowing that the odd midpoint could be different from the NBBO midpoint, we looked at the average difference between these two prices over time for each stock. The results in chart 2 show that there are sometimes quite large average differences.

In the chart below, we look at the absolute value of the difference and average that over time. So, a 10-cent difference doesn’t mean the NBBO mid is persistently rich or cheap – just that the average difference between the odd and round lot midpoint prices averages 10 cents. We also note that this data is from before a number of recent stock splits.

Using the same color scheme we have used in the past for our perfect stock price and odd lots and multi-tick studies, we see that stocks with “too many ticks” (darker grey dots) are also the ones where mid-point difference is more likely to be larger (dots higher on the chart), regardless of the stock’s price (the color change is horizontal).

Chart 2: We see odd-BBO-midpoints on some stocks are persistently very different from NBBO-mid

Importantly, for orders routed to exchanges, a limit order would not “trade through” a hidden midpoint order on the same exchange but would generally execute against it. That also means the odd lot won’t set a new odd-lot market that crosses the mid-point. 

Although more fragmentation makes this problem more likely.

NBBO matters for dark spread crossing trades too

An interesting question is: “How many trades happen at the official midpoint or far touch (NBBO) off exchange?”

Looking at TRF fill prices, we see that:

• Off-exchange Midpoint: Executions add to around 8%-9% of total traded volume across the market cap spectrum (black in Chart 3).
• Off-exchange NBBO: Executions add to around 8% of total traded volume in mid- and large-cap stocks and even more in smaller stocks (orange in Chart 3).

Chart 3: All off-exchange orders rely on an efficient and competitive NBBO

Some would suggest that a wider spread gives sophisticated broker algorithms an opportunity to capture some better spreads. However, research suggests that large institutional orders need to cross spread 20% more often than they capture it (Chart 4) in order to complete trades adding to a net positive spread cost.

Consequently, fills at the far-touch cost investors more than near-touch fills save. That is true whether on or off exchange when spreads are wider.

Chart 4: Even with sophisticated algorithms to work orders, investors mostly net pay spread

Others might say that mark-outs are lower in dark pools, meaning there is more spread capture when you do get a near-touch fill. Otherwise known as adverse selection, this occurs when trades occur just as the old NBBO changes, resulting in a loss on the fill to a two-sided market maker.

Looking at fills in the microseconds around a quote change (Chart 5), we see that around 38% of value traded on lit venues occurs as a quote changes. 

However, data shows that doesn’t only happen to exchange quotes. We also see a spike in executions in the TRF (dark line) as quotes change on exchange. In fact, the total TRF executions in the chart, a 15-millisecond window around a quote change, represent 28% of all TRF value traded.

Interestingly, we see dark activity (or off exchange) increases immediately before and after quote changes. The rise in the dark line just before the quote change might be algorithms searching for liquidity before sweeping lit markets. While the fact dark trading remains elevated after the quote change might instead show some reporting of dark fills has more latency than we see on exchanges.

Chart 5: Proportion of volume occurring around a quote change

We’re not saying every order from every algo should route to exchanges; there are a lot of trade-offs that traders need to make when working orders, including minimizing signaling. 

But we are saying reducing spreads should help reduce trading costs for investors, even those trading off exchange. A better NBBO could easily save investors billions of dollars each year.

Why is this important

A lot of venues trade off prices derived from quotes provided by others on exchange.

For large investors, who often have one-sided trades to work, a tight spread and accurate mid-point price become even more important to ensure they don’t add to their transaction costs.

That’s just another reason why getting incentives to quote publicly, tick sizes and round lots right is important.

Phil Mackintosh

Nasdaq

Phil Mackintosh is Chief Economist and a Senior Vice President at Nasdaq. His team is responsible for a variety of projects and initiatives in the U.S. and Europe to improve market structure, encourage capital formation and enhance trading efficiency. 

Read Phil's Bio