The IMD Wrap: Déjà vu as exchange data industry weighs its options

These columns usually serve up my own thoughts, but this week, I’d like to showcase some recent reporting by my London-based colleague Emma Hilary Gould, who has shone a spotlight on two issues currently affecting the US options markets, both of which are being driven by data—and specifically, the rapidly growing amount of market data being generated by these markets.

First, Cboe has proposed an amendment to the plan governing the Options Price Reporting Authority (Opra) that would allow options exchanges to sell their proprietary feeds—and options trading firms to subscribe to them—as a substitute for the full Opra datafeed. In Cboe’s case, it would promote its Cboe One feed as an alternative to the Opra feed. The exchange states that this will make options data more widely available to investors, while reducing the cost burden to the industry, and is merely clarifying existing rules.

Of course, it also potentially generates more revenue from proprietary feed sales for exchanges, including Cboe, and—as Nasdaq noted in its comment letter objecting to Cboe’s proposal—might divert funding from Opra and make it harder for the organization to do its work. On the one hand, there appears to be demand for more competition in the data market, and alternatives to Opra. And exchanges could potentially offer more innovative versions of consolidated tapes, segmented by the symbol or sectors that a trader covers, rather than the full firehose of all options symbols, which would be more voluminous and more expensive.

As US options data has grown in volume and cost, Cboe says changing the public feed’s governing document would make way for more competition from private alternatives, including its Cboe One Options Feed, launched in 2023.

So, yes, there are both explicit and implicit ways this could save money. On the other hand, if options data is permitted to fragment, the need for a centralized processor to aggregate the data becomes more important to create a complete view of the market, rather than individual micro-views based on data from individual exchanges. In the equities market, the Securities and Exchange Commission saw fit to establish a Consolidated Audit Trail (CAT) to ensure there was a definitive record of trading activity so that, for example, if a flash crash or fat-finger error occurs and sends markets into free-fall, it’s (in theory) simple to trace the event back to its cause, instead of having to pore over trade records from each exchange and match them with each other.

Were that to happen, which regulator(s) would be responsible? In the US, options are regulated by separate entities, depending on what type of option they are. The SEC and the Financial Industry Regulatory Authority (Finra) regulate options based on equities and indexes, while the Commodity Futures Trading Commission (CFTC) and National Futures Association (NFA) regulate those based on currencies, commodities and futures. As far as Cboe’s proposal goes, the SEC is the regulating body, and the Opra plan falls under Regulation NMS, but clearly options regulation is as complicated as the options themselves.

Cboe’s proposal isn’t entirely without precedent. It launched the Cboe One feed in 2023, but its heritage dates back 10 years to the launch of the BATS One feed of data from BATS’ marketplaces prior to its acquisition by Cboe in 2017. There’s also the Nasdaq Basic feed, launched by Nasdaq in 2009. Both were touted as affordable alternatives to the Consolidated Tape Association feeds from all US equities markets.

Over time, as exchange groups have grown by acquiring other exchanges, the proprietary data from within their various marketplaces becomes more representative of the market as a whole, and for some functions at least, can be used to replace official, industry-wide consolidated feeds of equities and options data. In Cboe’s case, the Cboe One Options feed provides quote and trade data from Cboe’s four options markets—the Cboe Options Exchange, C2 Options Exchange, Cboe BZX Options Exchange, and the Cboe EDGX Options Exchange—though not options markets operated by other exchange groups, such as Nasdaq.

Cboe may be appealing to other exchanges who see the potential to increase proprietary data revenues at Opra’s expense (despite Nasdaq’s objection), and to those unnerved by recent technical issues at Opra—or to cost-conscious firms who find the prospect of a “good-enough” feed with a lower price tag attractive. What’s interesting is that Opra’s securities information processor (SIP) technology is run by the Securities Industry Automation Corp (Siac), a subsidiary of Intercontinental Exchange via its acquisition of NYSE, while Cboe oversees Opra’s administrative functions.

But historically, money-saving promises about data rarely deliver savings, because not only does the price of datasets always continue to rise, but consumers feel compelled to continue subscribing to the old consolidated feeds as well as buying the stripped-down, higher-performance and cheaper proprietary alternatives.

In fact, this was the Opra members’ own interpretation following the Equivalent Access Provision amendment to the Opra plan in 2001, which allowed exchanges to distribute proprietary datafeeds directly to trading firms in addition to Opra. Before that, Opra had sole responsibility for the collection and distribution of US options data.

This, Cboe now says, was not the original intention of that provision. Rather, it was intended that firms could subscribe to feeds directly from exchanges and obtain Opra data on-demand. But the majority of Opra members disagreed on this interpretation, and even a lawyer retained by Opra concluded that the provision required firms to subscribe to Opra as well as any direct feeds. So, in 2003, Opra’s management committee voted to agree with the decision. But now, two decades later, Cboe’s proposal is attempting to correct what it sees as a decision that was “legally and factually flawed”.

As a result of that decision, similar to in the equities markets when low-latency data became a competitive advantage, firms ended up with two distinct infrastructures: one that supports the original consolidated feed for compliance purposes, and another that requires a separate infrastructure for capturing and consolidating competing proprietary feeds from multiple exchanges in-house—itself a more costly proposition, on top of their existing expenses.

And that cost—already an additional infrastructure cost in addition to a subscription cost—will also continue to rise. The need to process larger amounts of data without introducing latency has led to higher costs for hardware, ticker plants, and feed handlers (and, sometimes, FPGAs and other low-latency hardware devices engineered to be dedicated ticker plants or feed handlers). Additionally, the cost of bandwidth to accommodate datafeeds from multiple sources is growing, and the capacity planning required to accommodate any rises in traffic or unexpected data spikes over those feeds adds complexity. More options quoting and trading generates more data, which means more messages per second, requiring more bandwidth capacity to handle that data without clogging the pipes so that data may arrive delayed, out of sequence, or not arrive at all.

Which of these scenarios is worse depends on who you talk to. Slow data negates any competitive advantage you may have gained from having fast, proprietary datafeeds and from your investments in low-latency technology to process it, and leaves you at risk of trading on stale data and missing profitable trades (or ending up trading at a loss). If the data never appears—liken that to getting stuck in a pipe until it can be unclogged—then you have no view of the market, and can’t trade.

The danger here is not in taking on new positions, but being unable to unwind existing positions in a timely or efficient manner—again, risking trading at a loss. But for some, the scariest prospect is when data arrives, but—unbeknown to the recipient—some data packets may have been delayed by bandwidth constraints while others are not, resulting in data arriving all mixed up and in the wrong order—for example, appearing to show the market moving in one direction when in truth it is moving in the other direction.

So, ensuring that you have sufficient bandwidth is a costly endeavor—and an ever-increasing cost, at that, as data volumes naturally continue to rise. In 2004, shortly after adopting the Equivalent Access Provision, Opra increased the capacity of its feed to 75,000 messages per second (mps). By 2008, the projected requirement had rocketed to more than a million mps and was still rising fast. And by the start of this year, Opra was predicting that by July, datafeed recipients would need to handle almost 12 million messages per 100 milliseconds—equivalent to 120 million mps (Opra doesn’t even quote requirements in mps anymore, preferring to give more granular figures for 10 millisecond and 100 millisecond intervals to account for short peak bursts).

Concerns over creaking infrastructure in US options markets are fueling talk of measures to limit the exponential amount of data being generated.

For years, the industry thought it had solved its options bandwidth crisis. But with options trades almost tripling to 11 billion in 2023 compared to a decade ago, data volumes continue to rise to unprecedented levels. And there are signs that market participants are straining under the ever-growing cost burden.

Emma Hilary also recently reported on industry attempts to mitigate the number of strike prices being executed on options exchanges, which has increased in recent years, in part due to the popularity of short-dated options. For exchanges, there is less incentive to limit the number of strike prices, because more prices encourage more trading. And if a client wants to transact at a certain strike price, the exchange is unlikely to reject that business.

“Exchanges are not in the business of saying no. They’re in the business of ‘yes’ to their customers,” Steve Sosnick, chief strategist at Interactive Brokers, told Emma Hilary. “It’s much easier for the number of strikes to grow than it is for the number of strikes to shrink.”

Though the burden of responsibility for this problem should fall on those generating the volumes of data—i.e., the exchanges—Emma Hilary notes that some exchanges are nervous that any cooperation to solve the problem may be viewed by regulators as collusion.

And if any of this feels familiar, it may be because we’ve seen it all before: warnings of data tsunamis, capacity increases, investment in infrastructure upgrades, and efforts to shrink the amount and size of data being generated. This time around, though, the industry may still have a sour taste in its mouth from its efforts to adopt the FIX-FAST protocol in the late 2000s.

FAST (FIX Adapted for Streaming) was a data compression and decompression protocol developed to solve the growing data volume and bandwidth problem at the time. After many exchanges and trading firms began using it, an obscure company called Realtime Data (also known as IXO) filed a lawsuit against every company it could identify to be using the protocol, alleging that it already owned a patent on this technology. With the collective sums of money at stake, the industry as a whole dropped FAST like a hot potato. Bandwidth, participants reasoned, was becoming cheap enough that they could bear the costs rather than get tied up in expensive litigation, while exchanges such as CME and the International Securities Exchange (now part of Nasdaq) pursued binary data protocols to aid with bandwidth reduction.

Industry-wide problems like this require industry-wide solutions, run by industry-wide organizations. And, ideally, with the involvement and authority of regulators to ensure progress. Industry-led initiatives, unfortunately, tend to become mired in delays and competitive rivalries without someone with the power to cattle-prod them along—a case in point is the SEC’s CAT, which is now run by Finra. With the complicated regulatory environment governing US options trading, and with some of those exchange groups’ equities markets and regulators at legal loggerheads on other issues, cooperation between regulators and industry bodies (such as FIX Trading Community and the Financial Information Forum) may prove challenging—and yet, it may be necessary if the industry is to develop and implement a long-term solution to these problems, so they don’t continue to be a cyclical swing from boom to bust.

Agree? Disagree? Think you’re ready to build that solution? I always have the bandwidth to hear your thoughts. You can reach me at max.bowie@infopro-digital.com.