Bank Of England Paper on High Frequency Trading
The Bank of England has just released a new paper titled “High-frequency trading behaviour and its impact on market quality: Evidence from the UK equity market” . The paper is similar to the recent Kirilenko study in that it separates ‘aggressive” and “passive” high frequency traders and it uses data which identifies the counterparties of each transaction.
Before we get into some of the specifics of the paper, it’s important to note how the Bank of England defines HFT:
“The term “HFT” is generally used to describe any trading firm that makes extensive use of computer technology in its trading process with the aim of executing a large number of transactions within short time intervals. Furthermore, the firm normally ends the day with a relatively flat position.”
We highlight this because once again the concept of ending the day flat is part of the definition. We feel this is a key component to many HFT strategies and regret that our efforts to get this concept included in the CFTC definition of HFT were not successful.
The paper has a number of interesting observations:
– On a “per-share-traded” basis, HFTs have a much larger price impact than the other traders: whereas HFTs initiate about 12% of all aggressive volume, they account for about 40% of the total price impact.
– Given their trading volume participation of about 27%, HFTs (particularly the “aggressive” ones) can significantly amplify both price discovery and noise. Overall, HFTs have higher ratios of information-to-noise contribution than all other traders.
– The instances where HFTs contribute large amounts of both information and noise to the prices suggest that while some HFT trades have a large information content, others are entirely uninformative.
– HFT behaviour is very diverse
– HFT’s were separated into “passive” and aggressive” styles. The “passive” group trade two thirds of their shares (and also execute two thirds of their trades) by posting limit orders. The “aggressive” group on the other hand trades slightly more than half of its shares (and executes about 60% of its trades) via market orders.
– Passive HFT’s are faster than aggressive HFT’s.
– Passive HFT’s typically reverse their positions from one second to the next and their trading is price-neutral
– Aggressive HFT’s trade with the price trend. A price change over the previous 10 seconds leads to a same-direction trade which in turn leads to an additional same-direction trade in the next second.
– Both passive and aggressive HFT’s trade less when spreads widen.
– “Aggressive” and “passive” HFTs have a much larger price impact than the rest of the traders.
– HFTs are large contributors to noise. Why? HFTs end the day with relatively flat positions: If informed trading causes an HFT to build up a position during the day, the HFT will have to unload this position by the end of the day. The HFT may be forced to make a series of “uninformed” trades to simply bring her position back to her target inventory levels.
Unfortunately, the BOE paper lacks a substantial conclusion. Instead they took this approach.
“Nevertheless, the overall welfare implications of HFT are unclear; these will depend on how the marginal benefit of information at some times compares with the marginal cost of excess volatility at other times, including in periods of market stress.”