OWe recently reviewed the LULD rules and found that exchange-traded funds (ETFs) trigger LULD bands much less frequently than stocks. This makes sense, given that ETFs are a portfolio of stocks, representing a more diversified and theoretically less risky exposure. In fact, the data even showed that ETFs had much lower volatility than most stocks within the ETF.
Given this, we wondered if it would make sense to consider tighter LULD bands for ETFs than for stocks. Today we take a look at how this might work in practice.
We use actual ETF price returns in 2020 and 2021 and compare them to the modeled “half bands”. We also exclude the four MWCB dates as they were clearly unusual days, although around 80% of all LULD ETFs were still from 2020, with the majority in March during the Covid sell-off.
But first, let’s understand how LULD bands are calculated.
The Mathematics of LULD
Unlike MWCB bands, which are based on the previous night’s close and do not change throughout the day, LULD bands are calculated on a rolling five-minute window throughout the day. Basically, LULD uses a Easy arithmetic mean trade price, recalculated every 30 seconds for the previous five minutes.
The guardrails are then set at the LULD price ±5% for Level 1 stocks and ±10% for Level 2 stocks. Although the guardrails only move if the LULD reference price has moved more than 1% both ways.
It is important to note that a transaction during the opening auction sets the initial reference price of LULD. This allows the market to value the news overnight without triggering a LULD stop. For example, if a stock had positive earnings or M&A news overnight, the opening bid could be more than 20% higher than the previous close without triggering the first stop. However, if a stock does not have an opening bid, the LULD bands will use the previous day’s close price to establish the initial bands.
As we head towards the close, volatility can also rise at times, especially if a trader needs to complete an order before the overnight news is released. To allow this, without stopping these trades, Level 1 stocks (which have the tightest bands) switch to “double-width” bands from 3:35 p.m. Level 2 actions do not change, so they retain their original bands.
In Chart 1 below, you can see LULDs for the QQQ ETF for an entire day (dark blue lines) versus trades (yellow dots) and quotes (green lines). We also show how the open (light purple dot) places the LULD bands well above the previous night’s close (small dark purple dot).
Chart 1: The bands track prices traded during the day, only resetting when LULD rises or falls by 1% (note that the faint colors represent “half bands”, which we discuss below)
One of the advantages of the moving average trading price used by LULD bands is that they allow the market to continue trading if the stock’s price movements are “orderly” – representative of a valid price discovery or news-linked returns. For example, if a stock is actively trading, its price may drop more than 5% in the five-minute window, as new trades will reset the downward band before the five-minute window is up. This, in turn, reduces unnecessary stops and maintains liquidity in the market.
How LULDs Protect Investors
If a stock begins to rally faster than the reference price can be reset, the security enters a so-called “limit state” where the supply (in this case) is set at the upper “LULD limit”. If this order is not cleared within 15 seconds, the action halts for at least 5 minutes to allow liquidity providers time to participate in a reopening auction.
This is designed to protect investors from executing at “wrong” prices, particularly if a large market order comes in when spreads are wide or depth is shallow. For example:
- If the LULD of a Level 2 stock is $10, the upper and lower bands will be ±10%, say $9 to $11, to simplify the calculations.
- If the gap in the market is wider, say a bid of $9 – a bid of $12, a market BUY would NOT trade at $12, but instead cause a limit state, where the market buy is repriced at $11 and published in the book.
- If the NBBO does not fall back into the LULD bands (a seller hits the $11 bid) within the 15-second time frame, the action will enter a five-minute stop, triggering a reopening auction on the main exchange – which is also used to establish new LULD bands for continuous trading.
The result, in theory, prevents an aggressive market or limit order from trading at a clearly wrong price.
How LULDs for ETFs work
ETFs are split into two groups based on their liquidity – where the bands for Tier 2 (less liquid) ETFs are double the bands for Tier 1 ETFs. Leveraged ETFs have bands that are increased of the leverage multiple.
We’ve already seen that most ETFs are much less volatile than the underlying stocks in their portfolios, suggesting that perhaps we should be able to set tighter LULD bands for ETFs.
The data in Chart 2 also shows that Tier 2 ETFs (blue dots) actually have roughly the same volatility distribution as Tier 1 ETFs (orange dots). In fact, the median volatility (where the gray boxes change color below) for Tier 2 ETFs is actually lower. This suggests that to improve investor protection, LULD bands may not need to distinguish between Tier 1 and Tier 2.
Chart 2: The typical daily price range for each ETF shows that most Level 2 ETFs are actually less volatile than Level 1
More importantly, even the “half-size LULD bands” (dotted lines) are outside the average all day price range for more ETFs in each level – meaning the current bands are all good outside the average daily price ranges for each ETF.
Tier 2 ETFs trade less, but still price tight and consistently around NAV
ETFs are also different from individual stocks because they have a more verifiable “fair value” and mostly have active market makers who can arbitrate dislocations between ETFs and stock prices. This makes significant dislocations less likely to persist for ETFs, provided the underlying stocks are actively trading and creation and redemption mechanisms are available.
Examination of ETF data shows that the majority of ETFs see prices cited competitively and consistently throughout the day – closely around the underlying net asset value – even if there are no trades.
We can see this from the data in Chart 5:
- The way all the dots are to the left of the chart shows that ETF spreads are well within even half of the current LULD band levels.
- Although level 2 ETFs (blue dots) tend to have wider spreads, their spreads are almost all within half of a level 1 LULD band.
- Level 2 ETFs also have much higher quote-to-trade ratios (higher on the vertical axis). High quote-to-trade ratios indicate that even when no investor needs liquidity on these names, market makers are actively quoting all ETFs at prices synchronized with the underlying assets. In fact, the lack of trades is a testament to how ETF quotes track fair value, as if prices got out of sync there would be arbitrage trades.
Chart 3: The reason for this is that many ETFs are very actively traded, with tight spreads (but just don’t trade)
ETFs are different, but more research is needed
Obviously, ETFs are different. They are more diversified than individual stocks and can be arbitraged when prices break up.
The data seems to indicate that ETF LULD bands may not need to distinguish between Tier 1 and Tier 2 stocks, even though Tier 2 ETFs have slightly wider spreads. A more interesting question is: could ETF bands also be narrower? But that’s a question for another day.