2
Aug
2017

The wisdom of clouds

Small details can lead to big differences.  Often it’s not even the details, just things you can’t quite put your finger on.  I’ve been looking at a couple of companies that do the same thing, are the same size, operate in same industry but one is 6 times more valuable than the other.  I’ve been messing around with a programming language called R, to see if there are clues that the computer can pick up in the text, which I might miss.

I took the two companies management statements, and started by taking out all the numbers. Then used “word clouds”, where the size of the word reflects its frequency to look at what the companies say about themselves.  That is, the bigger words appear more often.  Below are the word clouds.  

(Click on them to enlarge)

But from this it’s not apparent which company is worth 6x more than the other one.  TrakM8 in red emphasises “revenues” and “telematics”, while Quartix in blue emphasises “fleet” and “insurance”.  From this, with a bit of imagination you might be able to guess that both companies provide vehicle tracking systems, either to companies with large fleets of vehicles, or to the insurance sector – reducing insurance costs for individual drivers.  The idea being that if you have a tracking device in your vehicle, the technology can tell how aggressively you are driving, which in turn predicts both your fuel consumption and your probability of having an accident.

To be honest though word clouds are not much good for giving you that insight.  You could have just read it yourself from the company’s home page.  But perhaps there are other things that this analytical approach to the text will reveal.  Below is a “commonality cloud” which reveals all the words in common.  

From this you can see both companies are optimistic, “increased”, “growth”  as well as “sales”,  “fleet”, “telematics” and “technology”.  The sharp eyed might be surprised that there is no revenue appearing in this cloud, when it should be there, probably the biggest, most frequent word of all.  This is my fault, still getting to know the R programming language I had trouble stemming “revenues” with “revenue” (one company uses the singular, one company uses the plural) because the computer counts them as two separate words, but they mean exactly the same thing.  I also wanted to stem “sales”, which is effectively same thing as revenue, to create one big word.  However my programming skills weren’t up to it.  So bear in mind that “revenue” would probably be the largest word in cloud.

But we still haven’t answered which company is more valuable and why.  One last go, with a couple of graphics.  TrakM8 is the company in red, which emphasises “revenues”, “telematics” and “new” (contracts and to a lesser extent new products).  While Quartix emphasises “fleet”, “insurance”, “revenue” and “subscriptions”.  And I’ve put the same information in a chart side by side. Again you’ll have to click on them to see clearly.

OK, I think we can safely say that Charlie Munger and Warren Buffett are not going to be losing sleep that I can replace their hugely valuable qualitative judgement.  To be honest, it’s early days for me using this programming language.  I haven’t programed since I was about 11 or 12 years old.   One thing that I’ve notice though is that it’s so much easier now, because I make a lot of mistakes and nowadays the error messages are much easier to understand. Plus when something still isn’t working you can look it up on the internet.  Or ask a friend who knows far more than I do, and often can instantly see what I’m doing wrong (thanks Tom!).  None of which I could do when I was programming in BBC BASIC.  The only other language option I had was assembly language – which is hard for a 12 year old. Actually assembly language is hard even if you have a computer science degree. So I lost interest.   I probably discovered girls were a more interesting challenge than computers (they still are).

 Value that can’t be touched

Even if the word clouds aren’t much help, I should explain why the companies are valued so differently.  This type of knowledge you only get from reading Annual Reports, understanding accounting and understanding management.  This is the valuable knowledge human fund managers possess, and it’s difficult, expensive (and often tedious) knowledge to acquire.

Thanks to those clever chaps at stockopedia, we can see that TrakM8 is trading on 1x sales, and 1.5x book value.  Whereas Quartix is on 7.5x sales and 10x book value (or a whopping 54x book value if you don’t count intangible assets.)  I’m not sure why my screen shot from stockopedia isn’t very clear – but if you click on it the resolution is much better.

Despite having the same amount of revenue, Quartix is far more profitable.  TrakM8 struggled to report a profit last year, whereas Quartix reports Return on Capital of 37% and an operating margin of 28%.  Those are really impressive numbers.

Actually, the financial gap between the two companies is even bigger.  The low level of profitability flatters TrakM8.  TrakM8 struggles for free cashflow at the moment, it looks poor.  The reason is because it is hard to value money spent on developing technology, particularly software.  For instance, am I wasting my time with this programming language R?  Or am I going to be able to develop it into something really quite valuable? No one knows, but when accounting for costs you are supposed to assume all the money is wasted, until it is clear that it hasn’t been wasted.  Much of the modern economy functions like this, from brand value and advertising spend to Paul Pogba (who developed as youth academy player at Manchester United, but who left without signing a new contract, but then was bought back for a world record €105m).   I digress, but the point is there is still judgement involved when spending money to develop any asset where the future value is unclear (be it software, a brand, or a football player).  

That is why I chose these two companies to compare: TrakM8 and Quartix take a different approach to how they treat their research and development costs.  

TrakM8 spent a total of £4.5m on research and development, only 30% (or £1.3m) it treated as a cost and reduced profits.  The rest it treated as an asset, that has recognisable value.  In contrast Quartix spent much less £1.4m, and expensed it all though the cost line of the p&l.  And remember Quartix is trading on 54x book value, if you don’t count intangible assets.  Or put another way intangible assets are 98% of Quartix’s value.  TrakM8 is trading on 10x tangible book, so 90% of it’s value is intangible assets.

I knew this beforehand, but wanted to see if the computer would notice it.  Or if it would throw off any qualitative insights about whether the money that TrakM8 is spending will turn out to be wasted or valuable.  I think it’s too early to say.  Both whether I am wasting my time, and whether TrakM8 are wasting their money.  But I have bought shares in TrakM8,  I think they are right to be spending the money, investing in the business.  The word clouds of both companies seem to agree that future prospects for telematics are positive. But it’s hard to know….if I’m wrong then I will be wasting both my time and money.

5 Responses

  1. Alan

    I think you mean TrakM8 here “Quartix is trading on 1x sales, and 1.5x book value”.

    Thanks for sharing the blog generally, I enjoy reading it, particularly how you blend financial ideas with more abstract ones.

  2. I can’t say much about R, but in python with the pandas package it should be easy to replace revenue/revenues and the like.

    But I really liked your idea, and just a few weeks ago I was almost into buying trakm8 too, because they just invest more into future in a business where size matters much. But remember there are much bigger competitors like tomtom. And one of the reasons for different earnings quality is the fact that insurance telematics seems tutu be structurally less attractive than fleet, where customer relations are very sticky. So the focus on fleet might be the true reason.

    1. Bruce Packard

      Hi Tobi, thanks. I think it should be easy to combine “revenue” and “revenues” in R too. I think it is my own ineptitude to be honest. And agree on fleet v insurance. I suppose one area of research (which the computer wouldn’t be able to do) would be to find fleet customers of both companies, and ask for their experience.

  3. Yes, this could be an interesting question, also the question how much the related consulting business at trakm8 is able to generate cross selling and synergies…

    By the way, did you compare also too other listed companies? I decided to more into south Africa , where you can find a (in my opinion much more attractive) interesting competitor with far more scale 😉

Leave a Reply