A Reminder Of A Topsy Turvy Madness In Small Cap Valuations

29th Mar '17 by Edward Croft

30 comments 17387 reads

Here’s a small mind experiment. I’ve got two investments to offer you. They are almost identical in what they do and priced at a similar valuation, but the first (company A) is very profitable and spits off lots of free cashflow, while the second (company B) is losing money and requires constant cash funding through debt and share issues. Which company would you rather invest in? Clearly, only a madman would prefer to invest in company B. But I hope to show you that there are very many madmen around…

Quality & Value - uneasy bedfellows

Any rational investor should prefer investing in higher quality companies to lower quality companies. So he’s likely to bid up the price of higher quality companies to a premium valuation, while lowering his bids on the lower quality companies to a discount. In theory this is the way the market should work.

In order to investigate whether the theory holds true I’ve created a few scatter plots using the Stockopedia StockRanks. On the horizontal axis I’ve used the Value Rank, while on the vertical axis I’ve used the Quality Rank. These ranks score every stock in the market as a percentile from zero (worst) to 100 (best). So the cheapest stocks have a Value Rank of 100, while the most expensive stocks have a Value Rank of 0. Each blue circle on the plot represents a single company, plotted along each axis according to its own pair of rankings.

If we plot Value against Quality for UK mid and large caps (greater than a market capitalisation of £350m) we see a picture emerge.

I’ve regressed a line on the above scatter plot to show the trend. It’s a weak trend, but it does back up our general thesis - that higher quality companies tend to be more expensive, while lower quality companies tend to be cheaper. The real holy grail in investing is finding high quality companies that are cheap (up in the top right corner of the plot), but the market is pretty smart and doesn’t give you too many opportunities there. In fact plenty of very successful value investors (such as Terry Smith of FundSmith fame, or the legend Warren Buffett himself) have learnt that paying up for Quality is one of the only ways to buy the best stocks in the market.…

Unlock with Email

Unlock with Google

Already have an account?
Login here

JMLDutch

29th Mar '17

For the purpose of these regression analyses, it would also make sense to weigh companies by market cap. It doesnt make sense that a 3M market cap gets the same weight as a 100M market cap.

Trigger14

I agree. The regression lines look quite spurious as the errors are enormous. I think it makes perfect sense for there to be a large number of low Q & V rank shares amongst the small cap stocks. Pre-revenue speculative blue sky stocks are going to measure as low Q & V regardless of whether they are fairly priced or not.

If you just look at the highest quality stocks, e.g. above 75, the top graph is more clustered on the left (high Q is expensive) while the bottom graph is more clustered on the right (high Q is cheap). Interesting though I'd be wary of reading anything into it.

mojomogoz

Nice piece. That there is no meaningfully positive regression between quality and value is evidence enough that the market isn't "efficient" in pricing smaller caps.

The analysis would be more robust if bottom lefthand side of small cap dots were removed (ie low quality, high value). As well as angel dust companies there will be a fair smattering of distressed cos that have high valuation due to damaged earnings (although maybe your value rank adjusts for this with EV/sales etc type measures...not a great valuation metric in isolation though - I should look before commenting as don't use the Stockopedia ranks).

In small caps I think there is a tendency to worry more about the sustainability of margins. This is as the fragility of market place they are in is often much more evident than for a bigger company. The very fact that smaller caps are more understandable leads to easier skepticism. This is particularly true when sales are not spectacular.

peterg

Pre-revenue speculative blue sky stocks are going to measure as low Q & V regardless of whether they are fairly priced or not.

That is a good point, Trigger

Edward Croft

I think many may be missing the real insight of the scatter plots. The real significance is not that there's a low R-squared with their trend lines, but how far these plots are from what we should expect.

As lavinit explains.... "That there is no meaningfully positive regression between quality and value is evidence enough that the market isn't "efficient" in pricing smaller caps"

The really significant point is that these scatter plots are not significantly downward sloping from top left to bottom right. As I tried to explain in the intro, if the market were efficient, then good companies would be at premium valuations, and junk companies would be at discount valuations - i.e. downward sloping. We do see a very weak correlation with this idea for mid/large caps (R2 = 2%), but a stronger correlation with the completely opposite idea in small caps (R2 = 20%).

Over the last 4 years we've seen good/cheap stocks outperform expensive/junk stocks. Efficient market theory says the market should have adjusted to wipe out this spread. But it hasn't.

58dbe4bf8b44aStockRanks_QVM_Performance_

If the market were fully efficient, this 'spread' between high and low ranking shares would disappear and you wouldn't see the outperformance of the higher ranking shares.

I'm sure me banging on about this stuff will help the market tend towards efficiency in some tiny way.... perhaps in 10 years time as the robots take over I can post another scatter plot that shows a downward sloping trend line with a very high R-squared. Perhaps the statisticians will then approve... but will they have missed the boat ?

Rusty2

It is an interesting article. I have just looked at the top QV ranked UK shares and to be honest I don't fancy many of them (I think QM is better), how can Bonmarche Holdings (LON:BON) be so highly ranked, I really have to question SR figures there. I suppose Purplebricks (LON:PURP) could be your company B, loss making, needing funding, complete junk yet the share price is up over 100% in the last year. Bonmarche Holdings (LON:BON) is down 57%. So I think you are correct more people probably would buy Purplebricks (LON:PURP) than Bonmarche Holdings (LON:BON).

I am a bit confused if the market has it so wrong why are SR 90+ outperforming the lower ranks on average?

Actually looking at VM, QM and QV, QV is the worst performing and VM is the best.

iwright7

30th Mar '17

Ed,

Looks to me like the smaller and microcap stocks are “muddying the water” in the Value vs Quality <£350M Cap scatter graph. I have looked at the V vs. Q for the AIM 100 Index, the smallest companies of which have Mrk Caps >£130M.

The AIM 100 bubble graph below shows a better correlation between V vs. Q reflecting improved market efficiency and it would be interesting to see where the Regression line appears with these?

AIM 100 companies have been my happy hunting ground over the last 12-18 months and AIM 100 Index returns have been fantastic over this period. I suppose the question at the moment is whether to extend stock selection into smaller and microcap companies looking for greater market inefficiency and to risk greater lumpiness and liquidity issues, or to seek out safer small cap middle ground? Ian

Ed,

Any chance of you producing a Regression line and R2 value with the Stocko Aim 100 Index Q vs. V Data, (please)?

Nick Ray

I get this for AIM100 using ggplot2. The dark grey area is the 95% confidence interval either side.

31st Mar '17

Nick,

Great graph – It’s the same story as with Ed's <£350 Caps - Within AIM 100 higher quality companies have a weak tendency to be cheaper! I see 20% of companies fit within the 95% confidence band. Not a fantastic fit and there seems to be more of a cluster above the R line.

Sceptics may say that this is erroneous data mining, or that the Q or V formulas do not reflect real Quality and Value, but I believe that the ranking formulas are more than adequate. The graph does seem to support the "Small Cap Anomaly" whereby under researched small companies will out-preform the extensively researched big ones.

Stockrank scoring tends to pick out these types of "mispriced" Q vs V (and M) smaller companies and contribute to Stockrank out-performance over time. Ian

Orangetree

Interesting article Ed,

Is the reason why people buy low-quality stocks are being influenced by others in the bulletin boards. And the research notes by brokers who have a conflict of interest.

My theory is if you help, seek a professional. If you are ill you go the Doctor, if you want to get your teeth done, you research the best dentist in your local area.
But, if you want to make money in the stock market, seeking professional help (as I mentioned) led to a conflict of interest. And, also investing is subjective to your opinion.

The "lottery ticket" analogy can apply to the state of society today where people wanting to win big (i.e.10-bagger) are those people who put £1,000 on a stock (the majority of people). So, are not invested in doubling their money (i.e. £1,000 profit). Whereas, those who invest £100k for a 100% gain would earn them £100k in profit. A more meaningful gain.

snickers

2nd Apr '17

Here's the same data with the firms split up by size..

..looks fairly robust!

'mc100' is the name is I gave to the log(market cap), rounded down to get the firms into 9 bins (only 8 lines as one bin contains only 1 firm). mr is momentum rank. The big firms show a perfect offset, slope=-1, of quality against value. What you want to observe is whether the other lines are turning round to this same slope.

Of course maybe Stockopedia just wrangled their formulas to get this very outcome..

or you can look at it like this. . .

whatever Quality & Value are, they effectively merge, & measure the same thing, as you get into smaller firms? The markets may still be efficient! Are some factors within the definitions of the 2 terms falling out? It's an odd problem - but seems easier to fix there in the definitions, rather than saying the market is nuts. Do they have to be polar opposites?

DWit199

If you do the same thing with Growth Rank rather than Value Rank do you see the same effect?

brucepackard

3rd Apr '17

Thanks Ed, great article. This is always something that I have suspected, but never had the data to back it up.

By the way, I have met some economists who believe the EMT holds true for large caps, but admit that it breaks down where you have small, less well known companies, and market makers with 10% bid offer spreads. Most fund managers can't take advantage of this, because they are managing billions and it far better for them to get HSBC and Shell right, than worry about companies that are below £500m market cap.

But retail investors have a real advantage versus the professionals when it comes to this part of the market.

Interesting plot snickers. A clear trend there as you move to larger caps.

However, I think we have to be careful what is being measured here? These plots don't measure "quality" or "value" in an absolute sense. They are the results of calculations intended to estimate something defined as Quality Rank and Value Rank - and so must depend on the underlying assumptions made about what is value and what is quality.

Triggers point earlier in the thread seems relevant here. The smaller the mcap the more likely the company is to be an early stage one which doesn't yet produce profits. Those will tend to show up as low value, and probably low quality too. Of course some (many?) of those may never produce a profit and they may indeed be poor investments. But I think it's important to be aware of the assumptions made in potentially excluding them. There may be very valid cases for investing in some of them.

What it would be unwise to do without evidence would be to jump from an observation that in general low Quality and low Value Rank stocks perform badly to an assumption that a subset of those (the smaller cap ones) are bad investments for the same reasons as the wider population.

Peter

I'm not sure you can take the smallest companies as a special case. Sure there's a purer corollation between Q and V but the trend is there right along the size scale. Some individuality is steadily leaching out of both Q and V as you go smaller til they become the same thing. Like opera and musicals, or space and time. Who can name this deeper conception??

Ken Lowes

"Nice wallpaper" snickers!!

4th Apr '17

Is the reason why people buy low-quality stocks are being influenced by others in the bulletin boards. And the research notes by brokers who have a conflict of interest.

I think it's a big part of it - the following blogs have some more background on bulletin boards and broker research. There's lies, damn lies and statistic of course... but I do find the averages tell a good story.

Sadly we're not that sophisticated !!!

Amazing chart Snickers... you've clearly shown how this anomaly gets stronger as the market cap falls.... crazy...

A reminder of a topsy turvy madness in small cap valuations

Quality & Value - uneasy bedfellows

Unlock the rest of this article with a 14 day trial

About the Author

Edward Croft

30 comments

Unlock the metrics for all these stocks and more...

Unlock the metrics for all these stocks and more...