A Bank of England working paper finds that aggressive buying among HFT firms usually results in a lasting change in stock prices

The normal debate over high frequency trading (HFT) is whether it adds to price discovery and market liquidity, or simply puts a drag on earnings through unnecessary intermediation. But the quant crash of 2007 and the flash crash of 2010 show how correlations between automatic trading can pose a real threat to market stability, at least in extreme cases.

A recent Bank of England working paper written by Evangelos Benos, James Brugler, Erik Hjalmarsson, and Filip Zikes looks at how correlated HFT strategies really are on a normal day and the effect that they have on prices.

“We find that instances of correlated trading by HFT firms are associated with a permanent price impact whereas correlated trading by investment banks is associated with only a temporary price impact,” they write. “In other words, HFT firms appear to be reacting simultaneously and quickly to new information as it arrives at the market place, which makes prices more efficient.”

HFT trade correlation

IB trade correlation

Aggressive buying correlated across HFT firms

The researchers restricted themselves to the ten largest HFT firms, which account for 98% of HFT volume in the UK, and used trade activity for the twenty largest stocks on the FTSE over a three month period. To determine whether HFT firms were actually moving together, they sampled aggressive buying and selling (orders to be filled at the best available price) and looked for Granger causality (a measure of how well one time series forecasts another one).

They found that aggressive buying by one HFT firm is Granger causally connected to buying by other HFT firms. Investment banks, which the researchers used as a proxy for well informed investors in general, have the opposite behavior. When one IB is aggressively buying, others are more likely to be aggressively selling.

HFT price movements tend to stick

Next, the researchers looked at whether price impacts reversed within the next five minutes (ie long-term in HFT land). They found that price movements caused by HFT trading tended to be ‘permanent’ while price movements caused by IB trading often reversed within the five-minute span.

“These results suggest that HFTs’ correlated trading is likely the result of HFTs trading on the same ‘correct’ information. In contrast, the correlated trading of IBs is associated with price reversals, suggesting that the correlation in IB strategies is less informationally driven” they write.

This doesn’t mean that correlations between HFT can’t create major problems when you get a negative feedback loop like in previous flash crashes. But it does imply that HFT is improving price efficiency the rest of the time.