Estimating Corporate Bond Prices from Dealer Runs and Market Data
Summary
The document discusses how a quantitative firm might estimate frequently updated reference prices for illiquid corporate bonds. One proposed input is dealers’ runs: messages sent to institutional clients that list bond markets. These messages can be collected and text-scraped, either by institutions receiving them directly or through data vendors. The responses suggest runs may reveal supply and demand more broadly than data from a single electronic venue.
Electronic trading platforms add another source of observations, while models can combine run information with related signals such as equity prices and credit index levels to produce pre-trade estimates. The exchange describes general industry practices and plausible modeling approaches, but provides no detailed model, measured accuracy, or evidence about any particular firm’s implementation. Coverage, data access, message quality, and the illiquidity of individual bonds limit what can be inferred from the discussion.
Key ideas
- Dealer runs can provide bond market indications that are collected and parsed from messages.
- Data vendors and electronic trading platforms offer additional sources of bond pricing information.
- Models may combine dealer indications with equity prices and credit index data to estimate reference prices.
- The discussion proposes possible practices but gives no model specification or accuracy results.
- Data coverage and bond illiquidity constrain the reliability of estimated prices.
Tags
Full text
# How could Renaissance Technologies have near real-time prices on corporate bonds and other debt? # How could Renaissance Technologies have near real-time prices on corporate bonds and other debt? - Julie Segal, What Renaissance Technologies has that you don't..., Institutional Investor, October 17, 2017. Does this just mean they are aggregating data from vendors like MarketAxess? Or is something more impressive going on? ## Answer by RWP - Down by the Bay (score 5) https://quant.stackexchange.com/a/53824 Outside of ETFs, corporate bond markets are driven by institutional flows. Over the course of a single day, multiple dealers will send multiple "runs" messages to their institutional clients. These "runs" contain markets on corporate bonds. The information from these is text scraped and stored. You can buy this information from someone like Bloomberg or you can be an institutional investor, have received the messages yourself, and do the text scraping yourself. I left this market 5 years ago so i can't currently speak for the portion of the market covered by MarketAxess, but even back then the "runs" information would present a much better picture of true supply and demand than MarketAxess. ## Answer by Brian B (score 1) https://quant.stackexchange.com/a/53847 The article implies, though does not quite say, that for quite some years now, Renaissance has been calculating up-to-date reference prices quite frequently. In the case of MarketAxess, "quite frequently" is defined as "every 15 seconds". If you are not terribly concerned with quality, even Bloomberg has provided minimal reference price capabilities for well over a decade. As pointed out by @R_down_by_the_bay, pricing in these markets has tended to be according to privately-published "runs" sent mainly to institutional clients. Aggregators like MarketAxess have often obtained their data through agreements with some of those clients. These days, the electronic trading on their platform provides an additional data stream that does not rely on partners' reporting systems. In the case of Renaissance, I doubt they are aggregating much from MarketAxess. They get the runs themselves, and have been doing so for years and years. It's easy to conceive of some big factor model where runs are constantly updating the coefficients. If part of the model has equity prices and CDX indexes in it, and good intuition was used in constructing it, it should do a decent job of providing up-to-the minute pre-trade prices even for illiquid issues. While I am sure the Renaissance model is well-constructed, they are not alone. There have been many institutions constructing pre-trade models for illiquid securities. They have no reason to tell us all about them.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.