Statistical Outliers in the UK Bond CT – From Calibration to Implementation

Introduction

 

On the face of it, at go-live, much of what the UK bond Consolidated Tape Provider (CTP) does is operationally straightforward. It consumes transaction data from multiple contributors, consolidates it into a single feed, applies data quality checks, and distributes transactions that meet the prescribed thresholds — all within a one-second regulatory timeframe.

 

Improving Data Quality

 

However, a key area of development lies beyond initial go-live. As part of its data quality framework, the CTP is committed to introducing three advanced statistical models to identify, amongst other things, potentially erroneous prices in reported transactions. Given that erroneous prices identified by the CTP form part of its regulatory reporting obligations, ensuring the robustness and reliability of these models is critical. While the methodology has been defined at a high level, calibration and validation can only be meaningfully undertaken once sufficient live data is available.

 

For that reason, the CTP’s approach is to:

 

• Deploy standard validation and rule-based checks at go-live
• Observe and analyse real-world data behaviour in the immediate post-launch period
• Calibrate statistical models using observed market conditions before implementation

 

This staged approach is intended to maximise accuracy and minimise false positives from day one of statistical model deployment.

 

Calibration Before Implementation

 

The earlier availability of these models within Plus is intended to support users who wish to:

 

• Explore the impact of statistical outlier detection in advance of CTP implementation
• Develop internal tooling or workflows aligned to the forthcoming methodology
• Access more granular analytical outputs alongside the standardised CTP feed

 

Once the CTP statistical models are formally implemented, Plus will remain complementary by continuing to provide enhanced analytics and visibility into underlying model behaviour where the CTP itself will simply flag a transaction as ‘Potentially Erroneous’ or ‘Suspicious’ if ultimately a reported price is considered unreliable.

 

Transparency through Publication

 

In parallel, and consistent with its commitment to transparency and non-discrimination, the CTP will publish details of the statistical methodologies.

 

A Measured Path to Higher-Quality Data

 

 

The introduction of statistical outlier detection represents an important evolution in the CTP data quality framework. By calibrating models on real-world data, publishing methodologies to the market, and enabling early exploration through Plus, the overall objective is to ensure that implementation is both robust and broadly understood.

If you are interested in ETS Connect Plus, or would like to discuss the statistical methodologies in more detail, please contact us at [email protected]