Competition between High-Frequency Traders and Market Quality

Johannes H. Breckenfelder
2013 Social Science Research Network  
High-frequency trading has been the subject of controversial discussions among legislators, regulators and investors alike, leading to calls for legislative and regulative intervention. The first entries of large international high-frequency traders into the Swedish equity market in 2009, using NASDAQ OMXS tick data, offers a unique chance to empirically examine how competition affects market quality. Competition among high-frequency market makers coincides (i) with an increase in intraday
more » ... se in intraday volatility of about 25%, but interestingly (ii) with no effect on interday volatility, (iii) with a decrease in order-execution time (length of time between an incoming market order or marketable limit order and the standing limit order against which the trade is executed) by about 20%, and (iv) with an increase in market share for highfrequency traders, but (v) with no significant effect on overall volume. We provide results for both entries and exits, and offer several potential explanations for this first empirical evidence on competition. for respected suggestions. I am indebted to NASDAQ OMX for providing the data as well as to Petter Dahlstrom and Henrik Talborn for fruitful discussions. The views expressed in this paper are my own and do not constitute an official position of NASDAQ OMX or its staff. All mistakes remain my own. † Swedish House of Finance, Drottninggatan 98, 111 60 Stockholm, Sweden. Email: established banks or hedge funds that are also significant players in the American equity market. We observe 228 entries and exits, measured by actual trades, for each individual stock and trader. 5 Contrary to previous literature, we can observe trader identities and therefore distinguish between different HFTs. 6 Our findings suggest unequivocally mixed results regarding market quality. First, intraday hourly volatility increases severely by an average of over 25%, five-minute volatility 15% and maximum intraday volatility about 15%. Interday volatility, both measured from opening to closing and closing to closing price, however, shows no sign of a significant increase or decrease. These results hold for both entries and exits, noting that, for the latter, the intraday volatility decreases. Second, order-execution time, defined as the length of time (in seconds) between an incoming market order or marketable limit order and the standing limit order against which the trade is executed, decreases in its median by about 20%, which is also reflected in a significant reduction of its standard deviation. Finally, even though the HFTs' proportion of total volume increases and decreases significantly after entries and exits respectively, there is, unexpectedly, no significant effect on total volume and the turnover of stocks. Granting these findings about competition and market quality, there are several plausible interpretations. First, competition increases intraday volatility since HFTs compete for the same trades. We find that HFTs in competition trade on the same side of the market in two-thirds of the cases (Figure 8 ) and have a correlation of 0.35 between their inventories. Second, HFTs trade more quickly and therefore significantly reduce the time for which limit orders wait to be executed. Third, there is no effect of competition on overall volume. While HFT volume indeed increases, as suggested by theory (Li (2013)), from an average of about 10% to 20%, there is likely to be a crowding out of other investors such as non-high-frequency market makers. Our findings of decreased order-execution time and the increased HFT volume could be related to a crowding out story of slow investors such as traditional market makers. These slower traders that are crowded out are likely to leave the market eventually. 7 Since HFT market making can respond more quickly 5 Throughout the rest of the paper, when referring to entry or exit, we will use the terminology in the following sense: entry represents the change from HFT monopoly to HFT duopoly within a specific stock, and exit the change from HFT duopoly to HFT monopoly within a specific stock. 6 See, for example, Brogaard, Hendershott, and Riordan (2012) or Hasbrouck and Saar (2012) , who work with a NASDAQ dataset that flags messages from 26 HFTs and has been the most comprehensive HFT database available to researchers in recent years. 7 A famous example is LaBranche Specialist, a long-time specialist on the NYSE, that exited the market in 2010 as new rules and technology made profitability difficult.
doi:10.2139/ssrn.2264858 fatcat:372eqpl4djaulgnx5j3j4qwfri