Pricing Insights: Navigating quote data

KEY TAKEAWAYS

• Trade data is the bond market gold standard, but can often be sporadic, late, or unavailable.
• Quote data is readily available and can provide important up-to-the-moment market intelligence.
• Traders need to be able to read through the quote noise to find the trade signal.

This article was written by David Krein, Head of Central Pricing Research, CTO Office, at Bloomberg.

Price data aggregation has become table stakes for success in bond market trading. Why would traders look at this information? It helps them establish pre-trade context, provides intelligence for market engagement, and offers post-trade insights. The more pricing data that can be consumed and processed, the more likely it is that their trade expectations are well-informed and accurate.

Although such data has historically been hard to come by, traders often now have access to far more pricing data than they can digest on their own: quotes, trades, derived prices, single party streams, axes and a myriad of other types.

Not all prices are trade prices

Each type of pricing data has its nuances, but the focus here will be on quotes. Broadly, a quote is a liquidity provider’s general level of interest in buying or selling a bond. (The term is often used interchangeably with indication or contribution.) Quotes are not firm.

Why do liquidity providers offer quotes at all? It is a way of collectively signaling to the market that trading is available in that bond, provides context for what that level might be, and suggests that the specific provider can be a source of liquidity. For that last reason, liquidity providers often supply the market with levels on hundreds or even thousands of bonds a day, and update the levels regularly if not continuously.

Interpreting quotes accurately requires careful handling, especially in automated systems, even before considering the level itself. For example, the quote level can arrive in price, spread, or yield. If spread, did it come with the benchmark? If yield, did it indicate the convention? 

Further, traders would need to similarly understand how each quote handles, or whether it even explicitly provides, an associated quantity, timestamp and time zone, settlement date, and certain other important reference data fields. The presence (or absence) of this information may hint at the liquidity provider’s investment in preparing it, which is not unrelated to its usefulness.

While market convention allows us to make reasonable initial assumptions and fill in some gaps, there are enough exceptions to such guidelines that make this a greater challenge than it seems on the surface.

Once these are navigated, though, traders can more confidently turn to the levels. An individual quote can be tricky for a trader to interpret accurately, especially in real-time. However, a single bond may see 10, 20, or more quotes at a time. This set might be tightly grouped in some markets such as government bonds, or widely dispersed in others such as corporates. Some might be very far from trade levels, and a few might even be through trade levels.

Aggregating them into a more insightful datapoint ahead of market engagement is critical. In segments where trading is bilateral and discontinuous, it helps understand market structure and liquidity. 

Putting quotes to work

Let’s consider what an ideal quote (or set of quotes) would look like. Given our pre-trade objective of using quote levels to give context to trade levels, the ideal quote would be one that is as close as possible to the trade level. It should be neither too tight nor too wide, and consistently and reliably points us to where trades will occur.

With that in mind, how useful are quotes in providing context for bond trading and achieving this ideal?

We can examine this by quantifying it in a well-defined case study based on recent, aggregated historical data. Assume a hypothetical trader needs to sell $750k of a US HY bond. Using TRACE, we can identify more than 7000 client sell (dealer buy) trades in US HY bonds between $500k and $1mm that occurred in January 2025. These levels serve as our collective “ground truth.” We can now establish a universe for benchmarking by collecting all of the quote levels that would have been observable just prior to each trade observation itself. Note, certain other simplifying assumptions were made to keep the analysis succinct and viable.

The chart above captures the difference between trade levels and a set of benchmarks. The Y-Axis is a measure of that difference in price. Positive results (the upper half of the chart) suggest that a trade was tighter than (inside, or better than) the benchmark level, or conversely, the benchmark was wide. Negative numbers (the lower half of the chart) suggest that the trade was wider than (outside, or worse than) the benchmark, or conversely, the benchmark was tight. The 0 line means that a trade level was at the benchmark level. The boxplot itself captures the results as the 25th and 75th percentiles (box), as well as the 10th and 90th percentiles (whiskers).

Quote Mean – This benchmark is the simple statistical mean level from all available quotes. We observe overall results that are roughly around 0 which suggests that the Quote Mean is, on average, a good indicator of where trades occur: 58% of trades were tighter than the benchmark, and 42% were wider. However, the dispersion (the height of the box and height of the whiskers) is very large, telling us that any single quote mean level is not a good indicator of a subsequent trade level.

Quote Median – This benchmark is the simple statistical median level from all available quotes, and tells a very different story. The Quote Median is somewhat far from (wider than) the trade level: 93% of trade levels were better than (inside) this benchmark, with an average difference of 27 cents. Fortunately, the dispersion (height of the box and whiskers) is well-contained, so it can be reasonably used for pre-trade context although clearly off-market by that amount. It’s worth noting that the gap between the Quote Mean and Quote Median results suggests that most quotes are wider than trades, though there is a long, aggressive tail at the top of the quote stack.

CBBT – This product is a rules-based composite price which filters available executable quote levels in real-time to produce context for market trading. Its overall results are most similar to the Quote Median: 95% of trade levels were better than (inside) this benchmark, with an average difference of 29 cents. Viewed that way, it is also effective at aggregating quote levels when available and can also be utilized for pre-trade context. 

IBVAL – This product is different from those listed above. IBVAL Front Office, available on the Bloomberg Terminal via (IBVL) PCS and for delivery through Bloomberg’s real-time market data feed, B-PIPE, uses the same quote levels, but combines it with recent trade data and then produces its own trade level expectation. Meaning, it should be closer to the trade level than any quote-only benchmark. The results show that to be the case: trades are tightly distributed around 0, with the narrowest dispersion among the benchmarks. This would make it the most effective pre-trade tool among the set.

How can we help

Traders can use quote levels to establish reasonable pre-trade context, though not the trade level itself. However, working with the quote data directly would be cumbersome, and requires experience with and understanding all of the underlying dynamics in real-time. And although a quote-only benchmark might simplify the process, the quote data alone cannot close the gap with the trade level. For that, some combination of additional data, tools, methods, or products are required to achieve effectiveness on par with IBVAL.

Learn more about IBVAL here.