Steven Poser is Vice President in the Strategic Analysis and Market Data Group at NYSE Euronext. Steven is responsible for providing analysis for...
In Part 1 and Part 2 of the series, we spent some time looking at day-to-day changes in major indices as a way of estimating whether or not the markets are more volatile now than they were in the past. We found mixed evidence using measures such as daily trading ranges, likelihood of a large (3% or greater move), VIX® and historical volatility. In short:
This week, we look at intraday volatility as measured by five minute trading ranges. We were surprised to find:
NYSE Euronext tracks the average five minute trading range throughout the regular hours (9:30-16:00) trading day. Specifically, we calculate the high price minus the low price as a percentage (actually basis points) of the mid-point of the high and low price. This kind of information is more likely to highlight market microstructure differences over time. Our data starts in 2002, so we cannot compare pre-decimalization to post-decimalization. However, we can see the impact of more recent developments such as HFT and Reg NMS.
The chart below shows for several years since 2002 the percent of issues that had an average five minute trading range in various buckets, starting at less than 10 basis points (0.1%) and rising to 100 (1%) or more basis points for a given trading day. We clearly see that even though 2002 took place as the market bottomed from the Internet Bubble, five minute ranges are somewhat more skewed to wider trading ranges now than they were in the first half of the 2000s.
Probably the most surprising conclusion from this set of data was that, on average, stocks have had more issues trade in the tightest trading band during 2011 than 2010 (through September). Trading ranges year-to-date in 2011 versus 2007 for NYSE issues are slightly worse, but this is likely due to the large number of financial issues listed on NYSE, which have been directly impacted by the European debt crisis, the U.S. credit downgrade, and the WWE Smackdown-esque battle in Congress over a U.S. government shut down.
Given that 2007 had far less in the way of macroeconomic “excitement”, and even included several months pre-Reg NMS, this is a surprisingly positive outcome.
One possible conclusion from the data is that HFT is actually helping market quality overall. Intraday volatility data are equivalent or better than 2003, although 2005-2007 showed more low volatility trading. This is likely at least partially due to financials concentration among NYSE listed issues.
The British government recently commissioned studies that reviewed the impact of HFT. The report concluded that HFT improved market quality. However, the research also pointed out that during periods of stress, there might be a tendency for high frequency firms to step back from the market. We find that for high volume NYSE-listed issues, about 1.6% of all stocks in 2011 had a 1% range in the last five minutes (2.1% in 2010). For the 2004-2006 period, this number never exceeded 1.0%. We are not convinced that this is a sign of HFTs stepping away and may very well be due to increased market fragmentation.
Late Day Volatility
We also reviewed end of day trading ranges. Monthly averages indicate that until volatility picked up in August, end of day five minute trading ranges were in line with pre-Reg NMS (2005-2006) levels. We have observed that primary exchanges tend to achieve higher market shares as the official close approaches. This may help alleviate some volatility, as liquidity becomes less fragmented and depth improves substantially as end-of-day approaches.