Is the conventional wisdom regarding high-frequency trading's contribution to 'flash crashes' wrong?
Is it unfair to lay the blame of the worst market-structure characteristic at the feet of high-frequency traders?
According to the authors of a recent research note published by the UK's Financial Conduct Authority, HFT firms are not the one whom most likely fuel mini-flash crashes, but agency brokers that provide direct market access for their clients.
The study did not examine the significant price moves that were on the same scale as the original 2010 flash crash or the whiplash that happened with sterling's flash crash in October 2016.
Instead, the authors decided to investigate incidents where a price movement exceeded the average realized variation of the previous 20 trading days by 300% that reverted by 50% within 30 minutes as well as triggering a high level of trading volumes, such as more than the top 5% percentile of the distribution.
The authors studied the trades of the FTSE350 in the UK secondary equities market from January 2014 to June 2015 and found that the hybrid broker-dealers tended to compound the shock by aggressively trading in the direction of the price movement. The HFT firms, on the other hand, initially tended to "lean against the wind," before ultimately trading in the same direction as the price shock.
The hybrid firms and HFTs continued to provide liquidity to the market, but not a the rate to meet the appetite of liquidity consumers.
The authors noted that the HFTs commonly withdrew their orders from the top of book and placed them further down in the book on the same side as the price movement, which further intensified the price movement, and took longer to return to top-of-book trading.
The authors have an interesting conclusion, but there's a question of whether someone is placing his or her thumb on the scale.
If the hybrid firms provide DMA services to their clients, how many of those clients are HFT firms using active trading strategies compared to the mix of active and passive order flow coming originating from the rest of the broker-dealer's clientele? Without the data, the contribution of hybrid firms to feeding flash crashes likely will be overstated.
