RRegular readers know that we spend a lot of time researching how to make trading more efficient so investors keep more of their returns and stock valuations improve.
One of the ways we can make changes ourselves is through our own listing rules and market incentive programs. In fact, we’ve shown that our level playing field, with incentives available to all Nasdaq members, creates a more competitive market. Our data shows that it is better for investors and issuers to have multiple buyers and sellers working on either side to tighten spreads.
Simple averages miss the point
However, because we spend so much time studying trading, we know you can’t easily compare one stock to another.
Things like market capitalization cause spreads and liquidity causes spreads to vary from less than 0.01% to well over 1.00%. It is also well known that growth and small cap stocks have higher “beta”, which is good when markets are up, but also – by definition – means they will have higher volatility while it happens.
In short, when comparing stocks and markets, simple averages are just bad math.
Consider, for a first example, the data in Chart 1 below, which compares stock size and spreads by trading venue. Almost all stocks (dots) form along a diagonal line, which shows that market capitalization is very important for spreads. In fact, for stocks above $1 trillion, the widest spread is around 9 basis points (bps) (for GOOG, before its recent split), which is cheaper than the best spread for stocks below $1 billion.
Chart 1: Spreads are determined more by stock characteristics like company size than by listing location
Coloring this table by listing the places highlights that the points overlap almost completely. Far from a clear winner – when stocks are compared correctly, the differences are measured in basis points.
It would be easy to average “all stocks” by color. In fact, even though there are virtually no gray dots below $200 million in market cap in Chart 1, the gray dots that do exist confirm that the expected trend (that smaller stocks continue to have wider gaps) continues.
Obviously, the simple average would be different by a factor, not a fraction. It’s because:
- The spreads on small stocks are well over 100 basis points – 100 times more than for large stocks – mainly due to the size of the company. Thus, small stocks drive up an average quite quickly without saying anything about the relative performance of the market.
- To emphasize this, consider that large cap stocks also trade more because they have more value invested in them. If we calculate the weighted average cost of the spread, which is closer to the costs for real investors, we see that the spread is closer to 9 bps, whereas a simple average of all stocks, the spread is around 75 bps. It’s over 8 times higher, simply because it counts smaller companies that trade less in the same way as large companies that trade a lot.
As anyone can see from the chart, using a single average without considering trading patterns is like comparing apples to oranges.
Understanding cross-sectional trends is also important
The composition of the stocks that make up the blue dots and the stocks that make up the gray dots are also very different.
We already know that market cap matters. But within the market capitalization there are also differences. For starters, almost all companies over $1 trillion are blue. They make up about 8.5% of all value traded each day, but when using a simple average, they only account for 0.05% of the outcome.
Even though the Nasdaq has many more quotes, it turns out that the representation in institutional benchmarks, like the Russell 3000, is almost exactly equal (blue area in Chart 2). However, the composition of the blue areas is different.
But what makes the averages even less relevant is the fact that the Nasdaq has almost twice as many total listings. This means that micro-cap stocks are 48% of the Nasdaq average but only 10% of the NYSE average.
Chart 2: Nasdaq lists many more growth companies
The fact that the Nasdaq has so many micro-cap stocks proves it is the preferred listing venue for new growth companies. This is evident from the data in Chart 3, which shows two things:
- The average age of companies is more affected by market capitalization than by listing location.
- But at all levels of market capitalization, Nasdaq quotes are even younger.
Chart 3: Nasdaq lists more younger and growing companies
Having younger new economy stocks also affects the sector composition of each trading venue. Not surprisingly, new companies tend to be more common in sectors like healthcare, technology and biotechnology than in financial services and utilities.
However, these industry differences are also important for understanding how averages can be misused. Average intraday volatility is higher for newer, higher growth sectors (Chart 4). This makes sense considering other research we have done which shows that the outlook for these companies can vary significantly as these companies expand into their new markets – in part due to less diverse and partly due to a lack of long-term financial comparables with which to use to forecast cash flows.
Understanding the volatility of stocks and the wider industry is another important factor in considering market quality – as it is widely recognized that volatility also adds to spreads, even when all else is equal.
Chart 4: New economy stocks tend to have greater intraday volatility (regardless of where they are listed)
There are advantages to being a new high-growth company. Usually they have higher valuation multiples and higher beta. All of that is usually good when stocks go up, but by definitionthis means that the stock also tends to have more volatility.
Again, we see that the average P/E multiple changes more as market capitalization changes – even more than the difference we see within each group due to listing location.
Chart 5: Nasdaq lists more fastest-growing companies
How do statisticians compare different populations?
The fact that many stock-specific factors affect things like spreads, trading, and closing volatility doesn’t mean it’s impossible to compare data between sites. Sometimes something as simple as grouping by one of the major factors (as we do with market cap above) will yield a much fairer result.
However, statisticians still have better ways of doing it mathematically. For example, in this research paper, academics use panel regressions to compare closing auctions between Nasdaq and NYSE. They took into account stock characteristics such as size, spread, stock prices, turnover and daily volatility (see page 14 of the document).
After that, they found that “price spreads are 1.2 basis points higher for NYSE auctions.” This represents about 18% of the average price gap (8.1 basis points) – a significant fraction of the average but not a multiple.
According to them, one of the reasons could be the advantages that d-quote offers to some participants over others. However, another paper found that NYSE auction order imbalances and indicative prices are also less accurate, which can make NYSE auctions less efficient. In short, an uneven playing field adds up to a lack of supply and demand transparency, reducing competition for closing liquidity.
We replicate their findings more simply, simply grouping stocks by market capitalization. What this shows is consistent with the research above. The discrepancies are much more due to size than location, but across all categories the Nasdaq close is better.
Figure 6: Closing auction dislocation (after controlling for firm size groups)
It is more difficult, but not impossible, to compare AAPLs to oranges
What all of this shows is that using something as simple as an average to compare all stocks in the market is like comparing AAPLs to oranges – they can both be fruits, but everyone rest is very different.
But it is not impossible.
There are well-known statistical ways to do this accurately, as well as easy ways to compare stocks that are more similar in the first place.
You just need to understand how trading trends work in the first place.