BofA Flags Algo Improvement as it Adds VWAP to Bloomberg
Posted by Colin Lambert. Last updated: June 18, 2024
Just over a year since it upgraded its algo tech stack, Bank of America has published a study of TWAP and VWAP algo performance, pre- and post-upgrade, ahead of the bank making its recently-launched VWAP algo available on Bloomberg.
The study, Getting the Basics Right – TWAP and VWAP Algo Performance, is the bank’s latest report in its Microstructure series from its e-FX trading and quantitative services teams, and finds that TWAP slippage performance improved by 22%, from an average of 0.31bp slippage to 0.24bp. It also says child orders now collect more spread on average. The analysis is based upon G7 TWAPs since January 2022, with no limit price, and at least $1 million in notional for at least one-minute duration.
In the study, BoFA observes that a “good” TWAP will dynamically adjust child order size and placement to maximise spread capture and minimise slippage to the reference mid-rate on the prevailing exchange. The enhanced algo provides this ability, as well as new ways to interact with hidden liquidity. “There is an intelligence to how large each child order should be to maximise the spread capture of each fill while minimizing slippage to the TWAP benchmark,” the study states. “At one end of the scale, if the order is divided into many small slices, the algorithm will be able to match the benchmark rate, but will be unable to fill passively, due to external venue minimum sizes, and cross into an internal risk price – paying child order spread.
“Conversely, if the child orders are too large and infrequent, the algorithm might capture spread, but end up losing against the TWAP benchmark by not sampling the market with enough trades,” it adds.
The enhanced algo optimises across both metrics to calculate the optimal child slice size based on market conditions and then places and updates these orders based on a real time-to-fill signal and execution risk tolerance rather than on a regular time clock. “These features help minimise slippage to the benchmark, maximise passive fills and reduce market signalling and impact,” the study states.
The upgrade to the algo tech stack has also allowed BofA to offer VWAP execution, like the TWAP, this algo also takes in a start and end time as well as notional, but instead of trading at a constant speed, it adjusts its schedule based on the historical and observed market volumes. The VWAP allows the bank to offer a strategy that removes one of the challenges of TWAPs, over- and under-participation.
The study highlights how a TWAP will over-participate in Asian hours, relative to market volumes, and under-participate during European hours, leading to unnecessary market impact in the first case and lost opportunities in the second. The new VWAP algo uses multiple sources that provide market volumes, but has been enhanced with market volume signals, that adapt to changing market conditions.
“In the case where the execution period can extend across multiple hours and changing liquidity conditions, the VWAP is ideally structured to manage market impact and ensure participation is correctly sized for market conditions, when run over a long enough duration,” the study states.
The VWAP algo is being released on Bloomberg as the latest strategy from Bank of America, “This is the latest step in ensuring our clients can access our full suite of products across all platforms,” says Tan Phull, head of FX algo trading at the bank.