The strategies employed by algorithmic traders on the Tel Aviv Stock Exchange and the connection between them and indicators of trading quality
Full research (Hebrew)
In recent years, securities trading algorithms have been developed in Israel and abroad, which have allowed automated high-frequency trading (HFT) without human intervention. Research conducted by Orry Kaz and Dr. Roy Stein of the Bank of Israel Research Department examined the activities of securities trading algorithms (high-frequency trading—HFT) on the Tel Aviv Stock Exchange (TASE), based on TASE data. The main findings of the research are:
a. HFT in Israel submit 90 percent of the orders, relatively high compared with other countries, but the share of transactions that they actually execute is markedly lower than the share worldwide. This indicates that their activity in Israel is broad and unique.
b. HFT activity in Israel is focused on providing liquidity through market making (simultaneously placing buy and sell orders), and is therefore important. This strategy, based on advanced market making technology, improves the liquidity and is coordinated with a decline in spreads and intraday volatility, and with an increase in the speed of price discovery. However, the analysis conducted found that market makers decrease their activity on noisy days, indicating that they probably create phantom liquidity, so that there is less liquidity precisely on the days it is needed more.
c. The activity of HFT that does not function as a market maker is not aligned with an improvement in indicators of quality and at times is even aligned with a negative impact on them—indicated, for example, in examining balanced HFT, algorithmic trading tools that do not change position at the end of the trading day compared with that of the previous day.
d. Official market makers, trading accounts that the TASE appointed to act in financial assets with low trading volume, do in fact reduce the spread to an extent. However, they are characterized by high positive serial autocorrelation relative to other strategies, a phenomenon that becomes worse on noisy days. This indicates that they are not efficient as market makers.[1]
e. In contrast to HFT using other strategies, balanced market-makers are significantly positively autocorrelated. This correlated activity is liable to increase the vulnerability of the secondary market and to increase the systemic financial risk.
In view of the findings, consideration should be given to limiting the activity of HFT in all traded financial assets, even if it involves some decline in their liquidity, as this will contribute to reducing the risk of failures and manipulation. It is therefore important to establish close oversight of trading activity—oversight that monitors it in real time, among other ways though innovative means.
Background and motivation for the research
Over the years, the activity of HFT worldwide and its impact on securities trading have been the focus of attention both of capital market participants and of the supervised entities. Algorithmic trading can contribute to trading quality because they integrate new information rapidly, reducing the information asymmetry and increasing the speed of the price discovery process, increasing assets’ liquidity. However, in contrast, algorithmic trading could exploit its ability to respond with high frequency and to adopt trading strategies that adversely impact the “slow” traders. It is also liable to increase the information asymmetry and negatively impact assets’ liquidity. The awareness of the risks incorporated in HFT increased markedly due to occurrences such as the Flash Crash (May 2010), Knight Capital Group (August 2012), and the sharp fluctuation that took place on October 15, 2014 in the price of US government bonds. These incidents led to expanding the research focused on monitoring them and studying their effect.
The Bank of Israel researchers based their work on unique Big Data that include activity of various traders at a frequency of a thousandth of a second, as the HFT at times carries out tens of transactions in the same asset within a second. In order to identify and characterize the activity and strategies of HFT, the researchers used advanced technological and statistical tools. The sample period spans two years, beginning in January 2014, from when there has been an obligation to record HFT in TASE accounts, and it ends in December 2015. The sample was examined over time and by division into “regular” days and “noisy” days, meaning days with relative volatility in asset prices.
[1] This finding shows, more than anything else, that market makers for small-cap stocks act without regulated securities lending, which increases the risk of their position and adversely impacts their activity as market makers.