| Abstract: |
We analyse the intraday behaviour of high-frequency traders (HFTs) and its
impact on aspects of market quality such as liquidity, price discovery and
excess volatility. For that, we use a unique transactions data set for four UK
stocks, over the period of a randomly selected week. Our data identifies the
counterparties to each transaction, enabling us to track the trading behaviour
of individual HFTs. We first find that HFTs differ significantly from each
other in terms of liquidity provision: while some HFTs mostly consume
liquidity (ie trade more ‘aggressively’) by primarily executing trades via
market orders, others mostly supply liquidity (ie trade more ‘passively’) by
primarily executing trades via limit orders. To examine how trading behaviour
is related to these patterns of liquidity provision, we split the HFTs in two
groups, according to their trade aggressiveness, and examine the behaviour and
impact of each group separately. We find that the ‘passive’ HFTs follow a
trading strategy consistent with market making and as such their trades have
alternating signs and are independent of recent (ten-second) price changes. By
contrast, ‘aggressive’ HFTs exhibit persistence in the direction of their
trades and trade in line with the recent (ten-second) price trend. We then
explore the relationship between HFT activity and market quality. We find that
both higher price volatility and lower spreads cause HFT activity to increase.
We suggest a number of reasons as to why this might be so. Finally, we use a
tick time specification to examine the impact of HFT activity on price
discovery (ie information-based volatility) and noise (ie excess volatility).
We find that while HFTs have a higher information-to-noise contribution ratio
than non-HFTs, there are instances where this is accompanied by a large
absolute noise contribution. |