Setting the Table for a Challenging 2021

lmost a year ago, we looked at 20 things that might be important to watch for in 2020. What we didn’t expect was a pandemic!

In reality, though, despite the pandemic dominating markets and the economy, regulators still proposed big changes. However, as we walk backward through proposals in 2020, we see a lot of fuzzy logic. It was, in hindsight, a year of conflicting messages, and not just about whether we should trust in masks and science!

That is especially important to note now, as the SEC appears likely to approve at least some aspects of its NMS II proposals in a meeting on Dec. 9.

That’s likely to mean more fragmentation and more infrastructure, less actionable and less meaningful quotes, with little that attempts to make markets more attractive to issuers.

So 2021 is shaping up to be a year where we start to deal with the consequences of conflicting rules raised over the past 12 months.

Should we quarantine the market from COVID or not?

As we watched the market selloff in March, we saw four Market-Wide Circuit Breakers (MWCBs) trigger in the fastest bear market in history. The first two happened within minutes of the open. Then, the third one happened in the first second of trading. Moreover, that quick stop was fully expected, given the SPY ETF was trading down more than 7% pre-open. That led some to claim that MWCBs shouldn’t be triggered so quickly in the day, at least when ETFs were trading down lower already, and that markets should be allowed to stay open.

Chart 1: MWCBs did no harm and reopens were orderly

S&P 500 MWCB minute-by-minute returns

Source: Nasdaq Economic Research

However, that misses the point of MWCBs, which are mostly designed to stop feedback loops and mispricing.

We also need to remember that the market had seen only one MWCB in the previous 30-plus years, which occurred during the 1997 Asian financial crisis. Given how few we see, and how they work only on days of panic, it is important to keep the rules and math around MWCB’s simple. That way all participants can have confidence in at least one thing: how the MWCB will work.

Finally, we shouldn’t forget that these rules have been honed over the years, having missed many other dramatic selloffs, including the credit crisis, 2010 Flash Crash and the messy Aug. 24, 2015 open.

  • After the flash crash, where markets fell 9% in minutes on no news, the MWCB was tightened from 10% to 7%.
  • On Aug. 24, 2015, the MWCB should have triggered around two minutes into trading. Instead, we saw a feedback loop that mispriced ETFs and increased single stock halts.

Both created feedback loops that exaggerated already weak markets. Given what we found in 2020, the MWCBs stopped this from happening again, as they were intended to do.

But while some were calling to keep markets open longer on MWCB days, others were calling to shut markets down entirely. Research shows markets have weathered similar volatility many times in the past. In fact, prior shutdowns are almost always due to operational problems, not returns.

In hindsight, the liquidity proved important to investors and issuers, including Moderna, which tapped the market for funds to develop its COVID-19 vaccine.

We want to encourage lit bond prices but not stock prices

One thing that also became obvious in the dash for cash during the COVID-19 selloff was that the total lack of transparency into bond market trading left regulators and investors unable to act effectively. That’s led to a renewed call to create a more “equity-like” market structure.

Before the coronavirus selloff, moves by FINRA to release their TRACE for treasuries data were halted. But months after the selloff, the New York Fed published a report suggesting the lack of bond trading data affected trading costs for investors. That was consistent with another study that found that liquidity did, in fact, evaporate from bond markets during March, with quoted spreads widening, leading to potential mispricing for investors and portfolio managers, especially in off-the-run bonds.

Following that revelation, the SEC announced plans to force large Bond ATSs to allow fair access and report more details of their operations and order types.

But while all that happened, the access fee pilot proposal lingered. Its design reduced incentives to provide lit quotes. Our data shows those incentives add significantly to market quality, with better depth and more consistently tighter spreads.

Chart 2: Rebates created better market quality (and better fills for everyone, even off exchange)

Average percent time at best quote

Rebates are important incentives that exchanges use to attract lit quotes. Not only do they help offset providers directly for their adverse selection, they help to compete against the advantages of hidden trading venues who can also pay for order flow, segment customers and price improve.

Clearly, the lit spreads from maker-taker markets also result in better fill prices even for the hidden orders trading off-exchange. But by focusing on the explicit costs of take fees being paid, regulators ignore the benefits that lit quotes provide to the market. This is the opposite of what we’re finding works for bond markets.

We want one SIP, in multiple locations, that is only best-ex sometimes

Of course, while all that was going on, the SEC was also working on some big additions to Reg NMS.

In a proposal dubbed NMS-II, the SEC proposed creation of a single SIP governance plan (in January) but then said (in February) that there should be multiple independent consolidators in many locations. We went from a single, public, “golden source” NBBO to many commercial (M)BBOs, depending on your location and provider.

Although it might not matter much whether your provider has the same BBO as others in this new world. That’s because proposed changes to round lot sizes added odd lots to the BBOs without protecting them. So investors often aren’t guaranteed a fill, even if they’re resting at the BBO.

It was a complicated proposed policy shift. While giving the industry more centralized control over SIP plans with an expansion of core data mandating consumption of all quotes, the SEC simultaneously proposed moving to a new laissez-faire world of competitive BBO compilation where not all quotes matter.

Chart 3: New rules add odd lots to the BBO but didn’t protect them

SIP qualifying BBO vs protected quote vs minimum quote

Sadly, one of the few rules that wasn’t touched in the proposed revamp was Rule 605. That might have been useful to quantify the costs of trade-throughs and missed fills for investors, allowing brokers to prove that the new system would cause investors no harm.

Geographic latency is re-approved and should be removed

Of course, a large driver of the competing consolidators proposal was to remove geographic latency from the SIP—which it does.

Although that completely contradicts the SEC’s one millisecond (ms) de minimis rule and the linked approval of speed bumps that (by definition) add virtual geographic latency to the SIP.

It’s also confusing in light of the decision months later to approve IEX’s D-limit order type, where lit orders will be pegged to prices on other (remote) exchanges. That not only endorsed geographic latency, but data already shows it is creating what some have called “phantom quotes” for takers.

As a result of the SEC’s proposed actions, the U.S. is moving away from an equal, actionable, accessible market. It will become more similar to Europe, where segmentation and latency are so prevalent that, even though public data is available, each investor’s view of what they can access is different, making it hard to agree on what quotes and trades should be used to benchmark their executions.

The faster SIP should be slowed down?

Going against all these efforts to speed the SIP up are plans to add depth data to the SIP as well.

By our estimates, adding depth to the SIP increases the data to be ingested by more than 10 times. Especially in times of quote changes and peak messages, this will slow the updating of the SIP down, giving back a time advantage to those ingesting more streamlined feeds.

One commenter even made the point that these proposals clearly weren’t reviewed by any actual engineers.

Chart 4: Adding depth increases the SIP more than 10-fold

Average Nasdaq quote message records

Source: Nasdaq Economic Research

A bigger question is who needs depth? 

While a loud coterie of financial players says they need it to make decisions, it turns out that almost 99% of trades happen at or inside the NBBO. In fact, we found that of the 1% that do trade through the top of book; 50% of that volume is traded off-exchange and usually within a couple of cents of the market’s best prices anyway.

Chart 5: Depth is used by just 1% of all trades, and half of those trades are off exchange

Volume distribution

Source: Nasdaq Economic Research

That means the NBBO is all that most investors really need to trade and that it is often just a reference price. In fact, more trading already happens at hidden prices than lit prices. So adding lit depth, to any trader at human speed, would seem to be a marginal benefit.

Everyone should be a sophisticated investor

Proposals to add depth and speed to the SIP come from people who say the “two-tier” market we have is inherently unfair.

The problem with that debate is that not every investor is equal. In fact, most investors are humans. For a human, with reaction speeds of 250ms and unsophisticated internet connections, additional investments in SIP speed are totally unnecessary. For Main Street investors who are price improved, off-exchange, versus the NBBO, depth is irrelevant. For these retail investors, paying for other professional investors to have government-mandated depth and low latency is unfair.

Ironically, what is most important to retail investors is the golden source NBBO and protected quote, neither of which may survive to 2021.

We also found is that additional exchanges add much less to the NBBO than people expect and seem less valuable to investors than the SIP revenues they collect.

Chart 6: Most investors are humans who don’t need to invest in faster NBBO or depth

SIP customers vs Nasdaq depth customers

It’s clear that different investors need different amounts of data. Choice is actually important for them to only pay for what they will use.

From that perspective, the new “infrastructure” proposals are designed for the 1% of traders who are sophisticated already – but are missing any indications that retail investors won’t be forced into paying for a share of the changes.

We’ve also shown before they do that depending on their needs. Being forced to consume all data from all providers will further support the economics of fragmentation.

Something missing from this debate seems to be that sophisticated traders do already get depth feeds and direct feeds, often from independent microwave providers or data aggregators, or slower cheaper aggregators like Bloomberg. The “infrastructure” is already there. Perhaps it is the SEC’s rules about pricing everyone equally that needs reconsideration.

If all investors need direct feeds, why don’t they take them?

Speaking of prices for direct feeds, one of the most interesting studies we did this year looked at the prices and consumption of SIP and direct feed data.

Conventional wisdom is that all sophisticated investors “need” direct feeds. Further, so the theory goes, once you take one direct feed, you “need” to take all of them. The same theory then supposes that venues can (and do) charge what they want.

Actual data shows none of that is true.

Using the new Form ATS-Ns, and price sheets for direct feeds (which are both public) shows that, in fact:

  • Many ATSs chose to work with just SIP feeds.
  • Others substitute SIP for direct feeds from venues that add little to NBBO quality.
  • This results in prices for venues discounting as market share (and quality) goes down.
  • Despite that, some venues can’t even give free direct feeds away.

In short, the free market for prop data shows a lot of customer choices. That, in turn, leads to prices that better reflect the value of each exchange. And even then, consumers aren’t forced to consume data they don’t see a cost-benefit for.

Chart 7: Direct feed prices vary in line with market quality, but even then some venues can’t give free direct feed data away

Showing how price discounts and usage of direct feeds add up

Source: Nasdaq Economic Research

Ironically, the expanded core data definition would mandate the consumption of all direct feeds, something that free markets don’t do now.

That would increase the subsidization to new venues, adding to fragmentation and increasing industry fixed costs, even more than the SIP does now.

Did data really solve any problems in 2020?

All the data and research that went into the development of a COVID-19 vaccine would suggest, undeniably, yes.

But for many of the very smart ideas for our stock market, the data would suggest, maybe not so much.

Despite what we’re talking about today news last night makes it look like the NMS II ship is set to sail. That means 2021 could be a year where we see more market fragmentation, new segmentation and quote deterioration.

As we’ve seen many times, economics will drive what matters to the market. It comes down to working out what features we want to reward.