Service Performance

Data Quality Framework

Following the inaugural meeting of its Data Quality (DQ) Committee on 14th May 2026, ETS Connect UK has established its Data Quality Framework, comprising a suite of baseline and advanced outlier detection methods designed to identify potentially erroneous trade reports. These controls form a core component of the UK bond Consolidated Tape Provider (CTP) service, supporting the delivery of accurate, high quality post-trade data to the market. 

The Data Quality Framework combines a set of baseline threshold-based controls for prices and notional amounts, targeting common input errors, along with a more advanced suite of statistical outlier methods. Together, these controls are designed to systematically identify and flag anomalous trade reports, supporting the integrity, reliability and usability of the consolidated tape as a trusted source of post-trade transparency. 

Identified anomalies will be reported to Contributors and the FCA in a timely and transparent manner, enabling effective remediation and continuous improvement in data quality across the market. 
 

Baseline Data Quality Checks

The baseline data quality checks applied by the CT are grouped into a number of categories. These categories reflect regulatory and operational considerations and provide a transparent and proportionate framework for the handling of post-trade transparency reports. These categories are as follows: 

• FIX Message Formatting checks 

• Regulatory Compliance checks 

• Advisory checks 

• Reference Data checks 

• Economic Outlier checks 

 

Advanced Statistical Outlier Methods

Details of the statistical outlier methods implemented on the CT are as follows: 

1. Geometric Brownian Motion 

The Geometric Brownian Motion model assesses how statistically likely a reported price is given the expected range of price movements over time, based on historical data. It is time- and volatility-aware, meaning that price movements are scaled based on the time between trades and the level of market volatility. The model is driven by historical price behaviour, the time interval between observations, the estimated volatility and contextual limits based on recent median price ranges to differentiate transient spikes from sustained price changes. 

2. Isolation Forest 

The Isolation Forest model evaluates how atypical a trade is relative to the broader population of observed trades, identifying instances that are distinct from typical patterns. Its specific focus is on detecting rare or isolated patterns, without relying on predefined assumptions about how prices should behave. The model is driven by parameters including derived characteristics such as price changes and volatility dynamics. 

3. Exponentially Weighted Moving Average 

The Exponentially Weighted Moving Average model assesses whether a price is consistent with recent market conditions, by comparing it to an expected range derived primarily from the most recent observations. The primary focus is on capturing short-term dynamics and shifts in market conditions. The model is driven by parameters including recent price behaviour, dynamically updated volatility estimates, weighting factors determining how quickly older data loses influence and contextual limits based on recent median price ranges to differentiate transient spikes from sustained price changes. 

 

In line with prudent practice for new statistical models operating on live market data, parameters will initially be set with wide ranges. This approach minimises false positives while the initial dataset is established. Model parameters will be recalibrated as meaningful production data becomes available. 

The Data Quality Framework is subject to a formal governance process involving review by the CTP’s Consultative Committee and the DQ Committee, with final approval by the CTP Board. This governance process has been agreed with the FCA, and is designed to ensure independent oversight, transparency and on-going refinement of the framework. 

ETS Connect UK will publish further updates on this page, as the framework develops.