Skip to content
All library documents

Algorithmic Quoting and Market Making in Bond Markets

Article Quant Q&A · Author: sean Guo
Read the original

Summary

The note describes two uses of algorithms by bond market makers. A system can maintain required bid and ask quotes across bonds that a desk may not actively want to trade, adjusting them as the best market quotes move. It can also generate indicative prices for rarely traded bonds as a starting point, with a human trader checking whether those prices make sense when a customer shows interest.

Quote models may draw on parsed news, historical relationships between bonds and other market prices, and trade flows that reveal supply and demand. These examples show how automation can support quoting in markets where many instruments trade infrequently. The discussion is qualitative: it provides no model specification, execution results, or measured performance. It also does not describe how a desk controls inventory or quote risk, and its examples should not be taken as a complete account of bond market-making practice.

Key ideas

  • Algorithms can maintain bid and ask quotes across bonds that traders do not actively favor.
  • Required quotes may be positioned away from the best market and updated as market quotes change.
  • Algorithms can produce indicative prices for infrequently traded bonds for later human review.
  • Quote models may use news, cross-bond relationships, other market prices, and trade flows.
  • The note gives examples but no model details or performance evidence.

Tags

Fixed incomeMarket makingExecution
Full text
# Just wondering any algo strategy popular for vanilla bond trading?


# Just wondering any algo strategy popular for vanilla bond trading?












I got extensive experience on algo trading for cash equity, FX, so just wondering any algo strategy popular for vanilla bond trading?

## Answer by Dimitri Vulis (score 1)

https://quant.stackexchange.com/a/58959

Most bond market makers have some sort of algo that does some or even most of the boring market-making for some or all of their bonds, which computers can do better than human traders. The two main motivations I can immediately think of are:

- sometimes bond market makers are compelled (by a trading venue or by regulators) to publish a bid an ask for many bonds in which they're really not interested in. A computer program can publish bid and ask that 1) are not so unrealistic that they won't be accused of mocking the system 2) are, as much as possible, not the best. So the program finds the best bid and ask and publishes its own that are not as good, but also not clearly ridiculous, quickly changing them if the best quotes move. These obligatory quotes are very unlikely to be hit/lifted.

- sometimes bond market makers use algo to publish indicative bid or ask for seldom-traded bonds, as a first approximation, and adjusts them for changing market conditions. If someone actually expresses interest, then a human trader verifies whether the computer's quotes actually make sense.

The models that generate these bid and ask can be quite interesting - looking, for example, at parsed news, at historical correlations of bonds (especially corporates) with all sorts of other market quotes, and at trade flows (supply and demand).

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.