By Phil Mackintosh, Chief Economist, Nasdaq
Single stock halts, also knowns as “Limit up/Limit down” (LULD), are one of the important market guardrails designed to stop feedback loops in today’s electronically traded markets generating erroneous prices or unnecessary volatility in stocks.
Today we summarize how they work and ask: Why are ETFs and stocks treated the same?
Market guardrails for dummies
LULDs aren’t the only protection for traders in the market.
In fact, there are three different protections designed to address different circumstances:
- Market Wide Circuit Breakers (MWCB) halt all stocks when the market suffers excessive declines in one day. Unless this occurs too close to the close, the market reopens with auctions after a 15-minute pause. The market experienced four MWCB’s during the Covid selloff in March 2020.
- Limit Up/Limit Down (LULD) is designed to stop excess volatility in a single stock.
The SIP constantly publishes prices that represent guardrails above and below recently traded prices. When the market price hits those levels, the stock is first put in a “limit state,” where additional momentum is paused, but new reversion orders can bring the stock price back within the bands. However, if the limit state persists for 15 seconds, the stock is halted for 15 minutes and reopened with an auction. - Clearly Erroneous (CE) rules allow for single executions, like fat-finger trades, to be busted.
The trade must occur with price or shares that are, using definitions below, considered clearly erroneous.
Table 1 below is a general summary of how each market guardrail works for the purpose of this discussion. For more information, click the links to each guardrail above.
Table 1: How from MWCB to LULD and CE guardrails work
How often do LULDs trigger?
If we look at the past two years (2020 & 2021), we see that LULDs don’t usually trigger that often at all, especially considering there are around 10,000 NMS securities in the market trading all day, every day.
Included within the dates we look at below is the Covid selloff in March 2020, which saw an unusually high number of single stock (LULD) halts. In fact, the four MWCB dates alone saw 3,588 LULDs (purple bars in chart 1) that accounted for 19% of all LULDs in the past two years.
Even the meme stock craze didn’t match March 2020 for LULD activity – although it did cause a spike to 461 LULDs in one week (dark blue in Chart 1).
If we remove March 2020 and meme stock week, we see a more normal week that includes an average of just 20 LULDs per day. That’s just 0.0002% of all the tickers each day. So, LULDs are rare.
Chart 1: Count of LULDS per week over the past two years (with MWCB and meme dates colored)
Which stocks see LULDs trigger most?
It is also interesting to see what stocks trigger LULDs the most.
If we start by looking at the composition of the market (Chart 2), we see that around 11% of all stocks are so-called “tier 1” stocks (light blue), making up almost one-third of all share trading. Looking at the breakdown of LULDs (final two columns), tier 1 stocks were 15% of all LULDs during the MWCB dates. However, they are typically underrepresented on normal days – adding to just 6% of LULDs on other days in 2020/2021.
In contrast, ETPs represent 25% of all NMS stocks and around 16% of shares trading. Although, ETPs were an even smaller percentage (10%) of LULDs on the MWCB days in 2020 and are typically a very small percentage of LULDs (2% of LULDs on other dates).
In fact, it’s almost not possible to see the tier 1 ETPs on normal dates – as there were only 68 in the whole period (excluding MWCB dates).
Chart 2: The composition of the market is quite different from the composition of LULDs, with very few ETP LULDs compared to their market share
ETFs are less volatile than the stocks in their basket
It makes sense that ETPs trigger fewer LULDs than single stocks. For a start, they are mostly diversified baskets of stocks. That means even if the stocks in the ETF see volatility, the ETF itself should have a lower range of returns than the most volatile stocks.
That, in fact, is what the data shows too.
In Chart 3, we look at all the stocks in the S&P 500 and compute the high/low range for each ticker each day. The chart shows the average for each ticker over the same two years we are analyzing above, plotted as turquoise lines in the chart. The box and whisker chart shows the median stock has over 2.7% range (where the grey boxes touch), with more than 75% of the stocks averaging a daily range of 2.35% (from the bottom of the darker grey box up).
In contrast, the average range for the ETF representing these stocks (SPY in this case) has an average daily range of less than 1.4% (orange bar). So, the typical range for SPY, the ETF, is around half as volatile as the median stock in the basket (1.4% vs. 2.7%).
In theory, the more diversified the ETFs, the bigger the benefit diversification should provide to the ETF.
However, the data shows the same result holds for more concentrated ETFs. For example, the SOXQ ETF, which holds semiconductor stocks, averages wider ranges each day than SPY, at over 1.75%, but the stocks in its portfolio also still see larger ranges of returns on an average day, with the median stock seeing 3.4%, still almost double.
Chart 3: ETFs have much less volatility than almost all of the single stocks in their basket, thanks to diversification
Tighter LULD bands for ETFs might protect investors more
In the real world, ETFs are significantly less volatile than single stocks, thanks to the diversification of the underlying portfolio, which lowers risk.
The data shows that ETFs should have lower volatility than the stocks they hold. Our examples above suggest the benefit of diversification could be material. The lack of LULD triggers for ETFs over the past two years seems to support that.
Based on the results above, ETPs might work fine with bands that are 50% as wide as the single stocks in the market.
However, we also need to ensure that doesn’t lead to unnecessary ETF LULD halts – that might remove critical liquidity and hedging tools from market makers just as a genuine correction occurs – in turn making a MWCB even more likely to trigger.
Clearly, this requires a little more analysis. Stay tuned.
