The Rise of Bilateral Markets and Trading Places First Survey of U.S. Treasury Venues
When I started trading rates there was only two interdealer venues for benchmark Treasuries: BrokerTec and eSpeed. With the exception of a few voice trades arranged by brokers, all market data was publicly available through order books. Life was simple.
Today, there are over a dozen different trading venues and bilateral liquidity providers. None of the bilateral destinations publish volumes or the number of customers. Even worse, basic information about the platforms is incomplete or wrong. In addition, all of this comes with a steep tax for participants — it can cost hundreds of thousands of dollars to integrate a new venue.
This week, Trading Places is publishing its first survey of the interdealer U.S. Treasury market. We have collected the most complete dataset ever put together on this topic based on information from venues, liquidity providers, and over two dozen customer interviews.
The origins of bilateral streaming
By 2011, Getco was the most dominant player in the interdealer market with 15% market share. However, all of Getco’s trading was confined to BrokerTec and eSpeed. Profits dwindled as faster Chicago to New York fiber came online (followed by microwave networks). Getco’s response was to do what no other high-frequency trading firm had attempted before: build a customer franchise.
The pitch was simple. Why trade on public venues (BrokerTec and eSpeed) when Getco can provide the same liquidity, protect your trading information, and charge zero commissions? Or why trade with all the sharks when you can trade with just one shark?
In reality, building a customer franchise from scratch is no easy feat, and it look a long-time for Getco to get it right. I launched the platform in 2012 with a single customer. Six years later, most of that code lives on as Virtu Fixed Income. Virtu does a lot more volume, but it only has around ten active customers.
Citi quickly copied with CitiVelocity. The thought of dealers trading with other dealers had once been unimaginable. But if dealers were willing to trade with a high-frequency trading firm, why not a dealer? Citi threw a lot more resources at the problem and signed up a lot more clients. At one point in 2014, Citi was doing 3x to 4x the volume of the successor to Getco.
After Getco and Citi, there were another half-dozen liquidity providers who wanted to jump into the game. Most of these firms were high-frequency trading peers of Getco. Jon Herrick, who had a hand in the Getco platform, led Sun Trading’s initiative. Most of these initiatives were unsustainable given the long-term nature of building a customer franchise. Moreover, the expense of integrating with each new platform was cost prohibitive.
LiquidityEdge launched in 2015 to reduce the cost of acquiring customers for liquidity providers and the cost of technology integration for dealers. LiquidityEdge aggregates a dozen direct streams under a single connection according to the firm. Dealers can pick and choose the liquidity providers with whom they want to interact.
However, since LiquidityEdge does not sit in the middle of the trade (as an interdealer broker does), each participant must sign a trading agreement with each counterparty with whom it wants to trade. This has historically created a barrier to growth. It took Jump, the largest trader of U.S. Treasuries, a long time to integrate with a critical mass of LiquidityEdge’s subscribers. BrokerTec and Dealerweb launched streaming platforms to address this problem by leveraging their broker-dealers and balance sheets to sit in the middle of transactions. (LiquidityEdge offers credit intermediation through Cantor Fitzgerald, but this option does not appear to be used by many customers.) Traction, at least initially, was limited.
The survey
Given the growth and the lack of current information, Trading Places thought it was time to try and shed some light on the interdealer U.S. Treasury market. Trading Places spoke to venues, liquidity providers, and over two dozen customers as part of its June 2018 survey. Trading Places does not have a financial relationship with any of the firms listed.
The survey’s methodology was as follows:
- For venues with public order books, we retrieved data from the venue itself and verified against a customer’s market data feed. If the venue did not provide information, we just used the market data feed.
- For dark venues, we retrieved the data from the venue itself and verified with customer interviews. If the venue did not provide information, we built an estimate based on customer and liquidity provider interviews.
- For information provided on background, we banded the volume data.
- We attempted to de-dupe volumes of liquidity providers who trade on multiple venues. For example, Virtu is a liquidity provider on LiquidityEdge, Dealerweb, and its own single dealer platform. Only Virtu’s volume on its single dealer platform is counted toward the Virtu numbers. Any volume done on LiquidityEdge would be counted as part of LiquidityEdge’s volumes.
BrokerTec: 147.2b ADV, 69.7% overall market share, 80.8% order books market share. BrokerTec remains the dominant trading venue for U.S. Treasuries, and market share has remained stable over the past six months (despite the prediction of a rival). BrokerTec is beginning to leverage its monopoly position. Nex recently notified customers that a separate subscription will now be required for its data feed: “BrokerTec US Treasury Data will now be available in a unique package and will no longer be included in the Thomson Reuters North American Fixed Income or Reuters Capital Markets Premium Packages.”
Nasdaq Fixed Income: 31.24b ADV, 14.8% overall, 17.1% order books. Nasdaq has staunched the bleeding, and its market share has been stable for the past six months. According a recent client note, Nasdaq is looking to reengineer its platform using the INET architecture. Additionally, Nasdaq is emphasizing its midpoint dark liquidity as an alternative to the bilateral streaming markets.
LiquidityEdge: 10.23b ADV, 4.9% overall, 36.4% streams. LiquidityEdge’s volume doubled from Q4 2017 to Q2 2018. Jump and XR Trading are the two largest liquidity providers on the platform and made a number of breakthroughs with customers during that time period. Competition is fierce though with a resurgent Dealerweb and LiquidityEdge’s own liquidity providers looking to trade with customers outside the platform. Dealers are also concerned that David Rutter, a founder of LiquidityEdge, is looking to sell the firm. (Nichola Hunter, CEO of LiquidityEdge, denied this but confirmed that the firm has discussed offers.)
Virtu Fixed Income: 4.5–5.5b ADV, 2.4% overall, 17.7% streams. Virtu is protective of its volume numbers, but we have noted a downward trend in our conversations with customers as competition in the space has heated up. Virtu has recently told customers it does $20 billion on high volume days. Virtu’s own presentations claim three clients do $3 billion each on big days. (As noted previously, Virtu has only ten customers). The platform peaked in profitability during Isaac Chang’s last year (2015) at $10 million.
CitiVelocity: 3–4b ADV, 1.7% overall, 12.4% streams. Citi has fallen the most of any platform. CitiVelocity used to be 2 to 3 times the size of Virtu. Ever since Mike Sung left, the platform has lacked a sponsor on the trading side of the business. “Citi is no longer in the business of buying market share.” New priorities, set by Geoff Weber and Dierdre Dunn, have curbed the ambitions of the once dominant bilateral platform. Spreads are wider, the liquidity provided is less, and customers are bewildered.
Dealerweb (CLOB): 2.74b ADV, 1.3% overall, 1.5% order books. Dealerweb’s market share has not changed over the last six months. Dealerweb is focusing its efforts on the relaunch of its streaming platform.
Dealerweb (Streaming): 2–3b ADV, 1.2% overall, 8.8% streams. Customers believe that the use of Dealerweb’s broker-dealer as a credit intermediary is the killer app for bilateral streams. At least one large primary dealer is now doing significant volume and XR Trading is by far the largest liquidity provider. Dealerweb also claims to solve the problem with aggregating streams that largely mirror one another, “Dealerweb draws out the largest size at the best price and presents that to the liquidity consumers, with sizes that optimize for full fills with a single counterparty.”
In a bit of exclusive news, Trading Places has learned that XTX Markets is having conversations with dealers about being a liquidity provider through Dealerweb. XTX is third largest liquidity provider in foreign exchange but plays a relatively small role in U.S. Treasuries. Alex Gerko, CEO of XTX Markets, confirmed the news, but insisted that they did not expect to ramp up volumes until the end of 2018 or sometime in 2019. Another employee of XTX said, “we certainly have views on the current UST market structure.”
GX2: 1.5–2.5b ADV, 0.9% overall, 8.8% streams. GX2 offers dealers Treasury liquidity (primarily through curves trades) through its single dealer platform. Growth has been steady in the last year, as a number of primary dealer customers have been onboarded. GX2 is looking to expand into Treasury algos next.
Goldman Sachs: 1.5–2.5b ADV, 0.9% overall, 7.1% streams. Goldman Sachs offers liquidity to a handful of clients including LiquidNet. We believe there is not significant management buy-in for streaming to other dealers.
Credit Suisse: 0.75–1.75b ADV, 0.6% overall, 4.4% streams. Credit Suisse is taking a unique approach to its business with a number of initiatives: it has its own single-dealer platform, it provides liquidity through LiquidityEdge, it has an internal matching network that allows customers to cross with each other, and it is offering execution algos under the AES rates brand. If the market ends up in a hybrid agency/principal model (similar to foreign exchange), then Credit Suisse could be the firm to watch.
FenicsUST: 1.06b ADV, 0.5% overall, 0.6% order books. Fenics has gone from a few hundred million of volume a day to around 1.5 billion a day in the first half of July. It is good progress, but they have a long way to go in a crowded market place.
Jump: 0–1b ADV, 0.2% overall, 1.8% streams. Jump is the largest trader on BrokerTec, eSpeed, and LiquidityEdge. However, the firm is just in the first innings of building its own customer business. Mark Bruce (recently taking over for Sam Tegel) has made a big push in this area. Jump has recently developed an API and is working through the logistics of trading with customers directly.
Hudson River Trading: 0–1b ADV, 0.2% overall, 1.8% streams. This is the old Sun Trading book. Most of Sun’s volume is being done through LiquidityEdge.
JPMorgan: 0–1b ADV, 0.2% overall, 1.8% streams. JPMorgan has at least one customer up and running on a direct API. It is unclear to many if this is a serious effort.
BrokerTec Direct: 0–0.5b ADV, 0.1% overall, 0.9% streams. BrokerTec Direct is in a dormant phase, as it retools to go after buy-side customers.
Bilateral venues will continue to grow
In recent years, bilateral trading over direct streams has exploded in the U.S. Treasury market. Nearly every single primary dealer has embraced bilateral trading. Trading Places predicts growth will continue, as larger liquidity providers make progress in developing their customer franchises and direct streaming aggregators increase in prominence.
However, the rise of dark markets does not mean the end of the multilateral order book. As Sean Murphy from BrokerTec told me, “we don’t see this as a zero-sum game.” The liquidity providers in the dark markets are some of the biggest users of BrokerTec/eSpeed and rely on a lit order books for price discovery. Meanwhile, the IDBs are not sitting still. BrokerTec is set to re-launch BrokerTec Direct soon. Dealerweb will officially relaunch its bilateral streaming platform later this summer and has already achieved results. Nasdaq is emphasizing its dark midpoint liquidity to customers. And Fenics, is now at around 1.5 billion per day (0.7% market share) and growing.
