More and more firms are increasing their fixed income exposure. The evaporation of alpha for non-discretionary traders combined with thin volume has made it more difficult for buy and sell side firms alike to intermediate and profit. Additionally, the shift of treasuries, municipals (“munies”), other government bonds (“govies") and corporate bonds (“cash corps”) into the electronic market making game has made fixed income trading more attractive to more players.
With the emergence of new electronic trading venues (including those that link up buyers and sellers in almost real time as an exchange would) there is a newborn need to measure critically new statistics: Trade Away Ratios (TARs), Hit/Lift Ratios (HLRs), and overall Requests for Quotes (RFQs). These statistics provide immense value for fixed income shops and execution venues, empowering them to make decisions on making the markets appropriately.
For example, dealers on the popular execution venues need to quote a competitive spread in order to get lifted. HLRs let dealers know how many RFQs they responded to are actually picked up. As a result, HLRs are a good indication of how well dealers are quoting. High HLRs mean dealers are moving volume and their quoting is probably at the right level. Low HLRs indicate that quoting is too wide and someone can come in and beat their quote.
Similarly, an increase in the Trade Away Ratios (TARs) can be an early indication of a slowdown in business. If this metric is reported accurately in real-time, then dealers receive immediate visibility into when they start to lose business to other competitors.
Analyzing these metrics further by dimensions such as counterparty and CUSIP number or security ID provide further insight to adjust pricing, examine trading system inefficiencies, and see sources of toxicity.
In my experience, the current status quo for delivering those execution metrics remains scraping application-level logging from each individual server. The application logs record trading messages sent in either standard or proprietary protocol (many FIX-based or derivatives thereof). Typically there are manually-built, computationally-expensive scripts running on each server hourly to analyze logs, determine the trade ratios and report them. Depending on the reporting architecture, and preference of the end users, those ratios are stored in a database, presented on an internal dashboard, or emailed as top key performance indicators (KPIs) to traders and portfolio managers (PMs) alike.
The obvious problems with this approach stem from the dependency on manually maintained application infrastructure. The additional infrastructure required for application level monitoring (additional servers, applications, scripts, and databases) creates more points of failure and greater overhead. For example, if a trader doesn’t get a KPIs email on the hour, then Trade Support must troubleshoot their cobbled-together monitoring infrastructure. I have not so fond memories of writing more scripts to watch a log scraping script to ensure it is operating properly, not to mention the need to update everything as protocols, venues and traders change over time. Updates are simply more opportunities for scripting errors to creep into the works.
When discussing best practices for execution monitoring and analytics, one of our customers put it this way:
“Having a source outside of my process telling me how my process is doing is way better than just doing everything from my process [via logs]. It has definitely helped us improve a lot of things including our productivity.”
At the same time data analyzed hourly has the potential to be stale and, ultimately, unreliable. While operating on data that is an hour old may seem ok for a historically low frequency trading asset class, electronification means trading volumes will not stay this way forever. According to Reuters’ interviews with 12 large asset managers, banks and trading venues, on average almost two-thirds of government and corporate bond orders in the US and Europe are placed electronically. Greenwich Associates research shows that on any given day in the U.S. corporate bond market, over 90% of trades for 100 bonds or fewer (equivalent to $100,000 or less) are executed on electronic trading platforms.
In addition, new trading processes, enabled by technology, are in the works to make it easier for investors to trade a block of bonds across multiple liquidity providers. Just thinking about creating log monitoring scripts to deliver timely and accurate information for those scenarios makes my head hurt.
Contrasting the aforementioned approach, Corvil execution analytics sources Client Trade Away Ratios, Hit Rates, Filled Volume, and Fill Rates without the build-your-own effort to gather information from application logs (or home grown packet decoding and analysis tools). Corvil passively captures, decodes, and analyzes FIX messages off the wire, from an aggregation tap, single tap, or network SPAN ports. There is out-of-the-box support for a wide range of proprietary FIX protocols, plus monthly updates that deliver new and updated analytics to handle protocol changes.
A key benefit of this approach is real-time data delivery, so no more hour long waits for new information. The sample dashboard in the Figure below shows how business users, traders, trade support and technical users alike can immediately view metrics updated as they are calculated from network with trending indicators. The analytics also generate and update the “Top-N” or Key Performance indicators for inclusion in appropriate dashboards.
Figure: Corvil delivers key Fixed Income execution analytics and inter-dealer market insights in real time
In addition, Corvil allows users to drill into the full details of any message and also correlate message and transaction outcome information with performance, such as RFQ-to-quote times, pricing and hedging latency, etc. This allows users to not only see KPIs and trading performance in real-time, but also to identify technology infrastructure-related sources of degrading trading performance. This is the difference between information and actionable insight.
Similarly, the Inter-Dealer Market Stats table provides venue performance data, giving fixed income stakeholders a more data-driven way to systematically choose where to do business. Traders gain insight into where the most efficient hits are going to come from, while a trader may have a personal preference for a venue, real-time data can let them know that sticking to that personal preference may not yield the desired results.
Download the Best Practices for Trade Infrastructure and Execution Analytics report.