BMLL finds a bigger October deterioration in rates futures liquidity than in spreads

Rates futures liquidity weakened much more than quoted spreads in October, according to BMLL’s 5 November microstructure report. Four government-bond contracts recorded liquidity readings 15%–30% below their six-month weekly averages, while spreads were 3%–6% wider.[1]

The definition is consequential. BMLL’s liquidity formula relates traded volume to the magnitude of mid-price movement over 15-minute intervals. It is a volume-per-price-movement measure, distinct from the quantity resting at the best bid and offer. The percentages therefore cannot be read as equivalent falls in displayed depth.[1]

Why a tight quote is an incomplete answer

CME’s own methodology illustrates the execution distinction. Its cost-to-trade calculation asks what a specified order size would cost across the available book. If the required quantity is not available at the best ask, a purchase calculation uses offers at subsequent price levels. A quoted spread can consequently remain narrow while a larger order needs less favourable prices to obtain its full size.[2]

CME also excludes cost-to-trade calculations when the book contains insufficient orders. A missing observation is therefore not a zero-cost observation. For brokers comparing analytics suppliers, the useful comparison includes the order size, observations omitted and averaging method, alongside the spread statistic. CME’s description supplies a methodological comparison here, not an independent reproduction of BMLL’s October results.[2]

Contract selection can introduce another distortion. BMLL’s earlier research on rolling futures compared front- and next-month contracts and found that order-message activity could shift before spreads and liquidity at the touch. In its sample of 45 WTI contract pairs, covering four years, changes in insert-message activity preceded spread widening. That historical sample is useful evidence that the chosen contract and stage of its roll can change the picture; it does not establish an optimal roll date for October’s rates contracts.[3]

BMLL combines contracts by order activity to avoid inactive front months. These historical comparisons do not establish a broker’s realised execution cost or current executable prices.[1]

Report an error or submit an update