Weekly Commentary 29/12/21: Performance and distributions

Bruce assess his performance for 2021, and notes that he would have been better off doing nothing. He also looks at the skewed return distribution of returns of AIM shares, plus 2 companies that reported positive news last week.

The FTSE 100 reached 7,400 on Christmas Eve, up +14% this year, versus the AIM All Share +3.6%. Nasdaq was up +3.3% last week and +26% this year and S&P500 was up +2.2% and +26% for the year. The US 10y Government bond yield was 1.46% (v 0.95% a year ago, the range peaked at 1.75% in late March and fell to 1.2% in August.)

The FTSE China 50 Index is now down -20% this year. China seemed to have a good start to the Covid epidemic, with commentators suggesting a totalitarian regime is better able to enforce strict lockdowns. But it now looks as if Omicron will spread there too, with some doubt about how effective the Chinese vaccine is against it, plus the familiar worries about Chinese governance and the property sector.

The Turkish Lira was volatile last week falling to 16.5 to the USD before rallying sharply, as Erdogan announced he will backstop any losses incurred for holding high interest rate lira deposits if the currency declines by more than the bank interest rate. This looks like a form of currency peg, but the FT reported that the Turkish Central Bank’s foreign reserves fell by over $6bn in a week and were now in negative territory. I’m not sure this policy is going to have the desired effect. Volex, which bought a Turkish business DE-KA last year for $60m has been weak, with the shares down by over a quarter in the last 3 months.

Performance update: I finished the year up +22% (not including dividends.) My 3 best performing stocks were Impax which doubled, Quarto +90% (but still well below the price I paid) and Solid State +79%. Richard wrote about Quarto in May, and the uncertainties that he highlights explains why I neither cut my losses nor increased my position as fundamentals improved.

The three worst performers in my porfolio were all stocks I bought during the second half of the year: Andrews Sykes, Arcontech and Argentex. Add to that Impax AM is currently above the price I sold some of my shares, means that I would have been better off moving to a remote island, and coming back in 2022.

The second half of the year has been more difficult for me, which I suspect is true of most people. I was up +28% from January to the start of June. You may be wondering why I didn’t take more risk off the table, particularly as I warned about a strawberry pickers’ market in April and a year of two halves at the end of June.

The answer is that I try to limit my trading. Selling well is even more difficult than buying well. My trading used to be restricted of necessity, because when you work as an equity research analyst at a bank you can’t trade unless your Head of Research and the Legal and Compliance Department both sign a form. Originally I found this hugely inconvenient, but slowly realised that the constraint of making less decisions improved my portfolio performance. If you hold shares in good quality companies, most of the time, the best thing to do is “nothing”.

AIM performance The AIM All Share index was up +3.6% last year (chart on the next page). That is disappointing relative to the US markets, and even the FTSE 100. It’s also below the rate of inflation, so real returns were negative in 2021. I mentioned last week that Barclays Equity Gilt survey showed that negative real returns occur surprisingly frequently: 39% of the years in the Barclays 120-year sample (or twice in five years).

AIM has grown at +5.6% nominal 10 year CAGR, but there were plenty of stocks that had done far better than that. All the discussion on index returns misses that if you can pick a few winners and hold on, then your returns will be much better than the index.

I’ve used SharePad to put together a table below showing that concept.* So for instance, there were 5 shares that increased by 25x or more, 4 that were up 20x (but less than 25x) and another for that were up 15x (but less than 20x). The name for this is kurtosis (or fat tails), indicating that the number of high returning shares doesn’t drop away smoothly as you would expect with a “normal” Gaussian distribution.

Similarly a third of the stocks with a 10-year history (or 252 in absolute numbers) are down versus 10 years ago. Of this, most (200) are down by more than 50%. So the majority of the 455 stocks with a 10-year history lose money, but the average return for the index has been positive. The +5.6% 10Y CAGR has been driven by the less than 3% (13 companies) of the 460 that delivered greater than 15x returns.

Modelling returns Below is a binomial distribution model (which is a complicated way of saying there is a probabilistic way of calculating expected success/failure). Like compounding, I think that if investors can understand simple probability distributions, this can help our process. There are 3 inputs:

P = probability of owning a top 10 performing AIM All Share stock is (10/761) or 1.31%

N = number of trials (number of stocks you buy and hold in your portfolio.) I have assumed 20 holdings.

K = success (that is, the chance of owning zero, 1,2,3 or 4 stocks in the top 10).

The point of the model is not to tell you what to buy, but instead give you a different way to think about the trade-offs of portfolio selection. As you own more shares, you’re more likely to own one (or more) top 10 performers, and the model simply quantifies that. You’re also more likely to own more poorly performing stocks, so if you increase N, your returns will be diluted.

The most likely outcome, more than ¾ of the time, is not owning any top 10 performers. Then there’s a 20.4% or roughly 1 in 5 chance of owning one stock in the top 10, dropping to 2.6% chance of owning 2 stocks, and a 0.21% or 1 in 500 chance of having 3 top 10 stocks in your portfolio.

My aim is to increase “P” (probability of owning multi baggers). One way is to relax my criteria of success. Rather than a top 10 stock, a ten bagger would be a +26% CAGR stock, which would put it in the top 25 of AIM Allshare 10 year performers. But I think the most interesting way of increasing your P is to use a filter, to remove stocks that obviously have little chance of being high performers. So, for instance, filtering out companies that have no revenue, or EV/revenue of greater than 20x. Or that have never made a profit. Or that have a Piotroski F score lower than 5. It’s possible that you are excluding some potential multibaggers, but probabilistically you are likely to be increasing the odds in your favour over 10 years.

The table below shows that if your filters are good, and you reduce the number of investable stocks by half (to 370) or to a quarter (185), without reducing the number of likely top performers, you can increase your chance of owning 3 out of the top 10 stocks from just 0.21% to 6.6%. Or in frequency terms you go from an almost 1 in 500 event to a 1 in 15 event.

This seems to me a good reason to be using SharePad. I know that this is a theoretical exercise, but it is similar to how I behave in real life. My portfolio is 24 stocks, more than half of which I’ve owned for >5 years and 4 of these stocks have been ten baggers or more. That equates to a 1 in 6 frequency of ten baggers, some of that is chance but some of that is filtering out companies that have a low probability of going anywhere.**

This week it has been rather quiet for RNS announcements, but I look at two enterprise software companies Sopheon, which announced a FY Dec trading update, and D4T4 which was up +10% when they announced contract wins.

Sopheon FY Dec trading update

This software company that helps clients manage innovation and New Product Development announced a trading update and a small acquisition ROI Blueprint, for $3m in cash, half of which will be paid as an earn-out. ROI Blueprint has revenues of $0.1m in 2021F, but Sopheon think that there’s a lot of potential to up-sell.

Current trading Sopheon also updated on current trading saying that they expect revenues and EBITDA for FY 31 Dec 2021F to be “comfortably in line” with market expectations. That sounds like they are slightly ahead, but not enough to formally raise expectations. There will be another update at the end of January.

Broker forecasts Andrew Darley at FinnCap has tweaked his numbers higher due to the acquisition, he’s forecasting +10% revenue growth FY 2021F to $33m and +12% revenue growth to $37m FY 2022F. Although the trading statement talks about EBITDA, their broker is forecasting PBT to be £0.5m FY 2021F and just £0.7m FY 2022, which is not very impressive. SharePad shows that RoCE has fallen from over 30% FY Dec 2018 to low single digits this year and last. To some extent this is caused by a shift to a SaaS model and increasing recurring revenue. But recurring revenue is not good in itself, if the company can’t generate a decent Return on Capital Employed.

Valuation The low level of profitability means the shares are trading on a PER of > 350x 2022F. If they can improve margins to 10% of revenue then that implies 30-40x PER multiple in a couple of years time, which still appears expensive. At the end of November Sopheon had $24m (£18m) in cash which is 18% of the market cap.

Opinion The business has a long term track record of revenue growth, albeit with a couple of years of bumps in the road 2020 and 2019 but also 2014. I’d like to have a feel for what might increase margins and RoCE back to historic levels. The recurring revenue and £18m of cash on the balance sheet should give some comfort, but improving returns seem to be already priced into the valuation. Not for me.


This consumer data analytics company with a product called Celebrus announced a couple of positive contract wins. The first win was actually a contract extension with a US top ten financial services company, and the second a European top ten bank. The financial implications of these multi-year contract wins will be over £3m per annum to the Group’s recurring revenue as well as over £9m to the FY Mar 2022F revenues, according to the company.

Broker forecasts D4T4 has a March FY, and H1 to September was released at the start of this month. Last week FinnCap, their broker, put out a positive one-pager explaining the bull case, that this could be the first of many such deals and shows that the initial trials of the new product are going down well with customers. It’s notoriously difficult to make enterprise sales (as Arcontech’s performance shows) so I think it’s fair enough for the broker to highlight this. However, FinnCap haven’t raised their forecasts for FY Mar 2022F of £24m and FY Mar 2023F £27m. So although the news was positive, it also seem to be already in the financial forecasts.

This shows that the concept of “market expectations” is fluid, because earlier this month SDI and GAW reported “in-line” results, and both fell more than -5% in response. D4T4 however rose +10% in response to last week’s RNS, despite contract wins seemingly already being in the forecasts. Thus it’s not unusual for the marginal buyer/seller to have higher / lower expectations versus the published brokers’ numbers.

Using the “Quarterly” Tab on SharePad we can see that there’s a big H2 weighting with results in most years, as contracts tend to be renewed between October and the March year end. The company says that the new wins “underpin the Group’s confidence in achieving the board’s expectations for the full year.” So that sounds like there could be upgrades ahead, but nothing in the numbers so far.

H1 results At the H1 results earlier this month, revenue was up +49% to £7.6m (albeit still below the £8.8m in their H1 results 2 years ago.) On a statutory PBT basis the business was loss making £0.3m in the red, although that’s partly the result of the H2 weighting. Net cash was £16.1m and despite the accounting loss, cash generated from operations was £3.1m after some positive working capital movements.

I note from the chart that the shares sold off heavily for a couple of weeks before these H1 results came out, falling from 383p at the start of November to 279p following the release of the results. That’s a -27% decline, so presumably investors were fearful that the contract wins might not come through in the second half.

Valuation The shares are trading on almost 50x PER FY Mar 2022F, dropping to 21x PER FY Mar 2024F in 2 years time. That’s for a company where RoCE has been trending down (10.1% FY Mar 2021) and where FinnCap warn that profits are likely to be suppressed in the next couple of years as management spend on marketing. The shares are trading on 4.1x March 2024F revenues, so I think a lot of the successful roll out is already being anticipated.

Opinion This does look like a good quality stock, with management doing the right things to invest and grow the business. I’m a little worried that they could be competing with large, well-funded US tech firms like Palantir. I think you might need specialist knowledge to assess how good their technology is versus the competition. But so far the early contract wins would suggest that they have something that their clients find valuable. I’m put off by the valuation, but I can see that it could appeal to some readers.

Bruce Packard


* One caveat with this data is that there’s a strong survivor bias in the table. I looked on the LSE website and there were over 1,000 companies listed on AIM 10 years ago, versus this list which is 761. I don’t think anyone has kept track of whether the companies that left AIM delisted to go private, filed for bankruptcy or were acquired, either for shares in another listed company or shareholders received cash. The table also shows that there are 306 companies (40% of the total) that have joined AIM in the last 10 years).

** My “N” is higher than 24, because I’ve had two stocks taken over in the last 10 years (Red24 and Wey Education) and sold a few losers (most recently Hostelworld). I don’t think I’ve owned over 36 stocks in total in the last 10 years though, which would be 1 in 9 stocks being a ten bagger, though Red24 and Wey Education were also bought for multiples of what I paid, and if they’d remained listed could well have continued to do very well.

Leave a Reply

Your email address will not be published. Required fields are marked *