Fleas and Elephants

I first came to the thoughts in this post when I noticed that for the universe of stocks that I track, the median Quality Rank was about 63 rather than 50 as you would expect. So I decided to have a look at how the QVM ranks change with Market Cap, as my universe is limited to stocks with Market Caps over about £90m.

Jim Slater famously said that "elephants don't gallop" to describe the fact that share prices of large market cap stocks tend to move slowly. He later extended this by adding that "fleas can jump many times their own body height" to describe the behaviour of micro-cap share prices.

The implication is that small cap stocks and micro-cap stocks can provide much better returns than the blue chips. However this is a simplification of the true situation.

The distribution of Market Caps in the market: fleas, elephants and woolly mammoths

Slater defined an elephant as a stock with a market cap above about £5bn to £10bn and a flea as a stock with a market cap below about £5m to £10m.

These definitions correspond to the bottom 10% and top 5% of the UK stock market as you can see in the histogram. The elephants roughly correspond to the FTSE100.

This histogram has a log scale for the x-axis. Elephants are 1,000 to 10,000 times bigger than fleas.

The median market cap is about £100m. In other words, half of the companies have a market cap below £100m and half above £100m.

However because this is a log-normal distribution, if we ask about the balance of money across the market we get a different result.  

If we ask for the market cap where half of the total value of the market is above that value and half below that value, it is at about £28bn. To put it another way, the top twenty stocks account for half of the total value of the market and the remaining 1600+ stocks account for the other half.

So half the stock market's total capitalisation is not even all the elephants. You just need the twenty heaviest elephants - the "woolly mammoths" of the stock market absolutely dominate a weighted index.

The question to consider is whether fleas really are a better bet than elephants. So now let's look at the Quality, Value and Momentum Ranks across the whole market and compute the percentile breaks based on Market Cap.

Value versus Mkt Cap

This graph and the following two can be a bit confusing, so to get your eye in, we consider the easy case first.

What we have here is a plot showing the 10% to 90% percentile values of Value Rank for given bands of market cap. The 50% percentile (median value) is made thicker so that it is easy to pick out.

The x-axis labels are using interval notation. For example, (100,200] means the interval 100 < x <= 200. I have added panels indicating where the fleas and elephants are along the axis. As before the axis is (approximately) logarithmic (and incidentally, approximately equal sized quantiles).

Because Value Rank is largely unaffected by Market Cap, the percentile points are roughly the same as the Value Rank values. You will find an even spread of Value Ranks in any subset of Market Cap. So this is the "control" plot. This is what we see when there is nothing much to see.

Quality versus Mkt Cap

The picture changes significantly if we look at Quality Rank (QR) instead. The median line which should be at roughly the QR=50 level in fact passes through the 50 value at a market cap of about £100m. Below that size the median QR gradually falls below 50 and gets as low as 20 for the fleas.

Above £100m, the median line is flatter but there is still some increase. However for the largest subsets of Market Cap the main gain is that the lowest percentiles improve strongly. Really poor quality large caps are rare.

Momentum versus Mkt Cap

When we look at the Momentum Rank graph the overall shape is very similar to the Quality Rank case. The knee is present again at about £100m. High momentum fleas seem even rarer than high quality fleas.

Volatility versus Mkt Cap

So why do the fleas and small caps in general seem to offer the possibility of very high returns? The answer is supplied by the Volatility plot.

"Volatility" is not a ranked metric. The values on the y-axis are the uninterpreted volatility values. What we see is that beyond £100m the median volatility and other percentiles are broadly flat although it does tighten up again beyond about £1000m. However below £100m things rapidly become "interesting".

I find a volatility above 50% to be difficult to cope with. This corresponds roughly with daily swings of 5% or more and yearly swings of -30%/+50% randomly superimposed on whatever underlying trend there might be.

When volatility exceeds 100% things get really wild. (In fact when volatility exceeds roughly 80% the mathematics of stochastic processes starts to suggest that you should be trading the volatility rather than the underlying trend to maximise returns - but that is a topic for another day!)

So fleas don't just jump; they also bite. Below £30m, it is a "foreign country where they do things differently" (to paraphrase a famous quote.) High volatility needs different trading styles and strong mental fortitude.

To be honest I was a bit shocked when I saw this plot. I expected an increase in median volatility as Mkt Cap reduces, but not to this extent.

Just for completeness, here is a "ranked" version of the Volatility plot. The volatility values have been converted into a percentile value between 0 and 100 (although I have not reversed the numbers as Stocko would. So 100 is worst and 0 is best in this plot.)

I have mainly included this to make a point about ranked metrics. The same conclusions can be drawn but the plot maybe gives a different impression about how close or far away various values are. The lines are much less bunched up.

Using ranks provides a different perspective but it can be tricky to interpret correctly. This is something worth remembering when looking at Stockopedia's ranked metrics and wondering whether to use Q>75 or Q>90. A non-linear transformation has been performed on the data which might help you or mislead you!

Closing Thoughts

Most of the apparent advantages of fleas and small caps in general are due to the higher annual volatility.

• Factor investors will build in a useful edge by staying above £100m and maybe even above £200m. Bigger caps on average have higher momentum and lower volatility - what's not to like? There is no reason to apply an upper limit either. I used to avoid stocks over about £1bn but I no longer have an upper limit in my screens. If it is a good stock and passes the rules I keep it.

• Stock pickers (alpha seekers) on the other hand might prefer to look at small caps because there is a much wider range of performance on offer to be exploited. However a solid understanding of how volatility can mislead your impression of your abilities is very important in this domain. Stop losses also cease to be of much use when volatility goes over 50%.