LON:ABDP | IPO Analysis: Part 1 - Assassins And Rabbits; Multibaggers And Dogs

19th Jun '19 by dscollard

20 comments 2761 reads

Update July 2019: Part 2 of this Series is now available covering the addtional 133 IPO placings

Topline

408 IPOs from 2011 to 2019 were analysed quantitatively: the median return was -17% with a 60:40 split of losers to winners over the time period. This fits with published academic studies of IPO performance of -8 to -23% . A deeper analysis of the numbers gives more insight on opportunities and the role of stop losses in converting trades into very profitable investments some which yielded several thousand percent returns. A furhter 133 placings were not excluded as they were nto amenable to the datamingin processes used. These are dealt with separately in Part 2 of this series which should be read in sequence.

Introduction

To IPO or not to IPO? that is the question. Whether it is more profitable to risk the slings and arrows of a placing or to wait and see how the market receives it? These days I tend to ignore the barrages of broker emails and don’t get carried away by FOMO: I prefer to give a placing some time in the market before taking a position. Intuitively this feels right (I've had a few knocks in the past) but I decided to go beyond heuristics and take a deeper look at the data to see if being more systematic was of merit. Does it make sense to get in early at the placing price? How do IPOs tend to behave post the placing and in their formative years? The initial results surprised me so I did a bit more digging before drawing any firm conclusions. This article is the first in a series looking to unpack some of the insights and devise actionable trading and investing strategies. The full version of this article with an interactive report of the data allowing DYOR analysis is published on runprofits.com. ( Stockopedia data is only included in this summary article and not elsewhere )

Background to the Study

530 companies in the period 2011 to 2019 abstracted from a London Stock Exchange list off over 3000 IPOs and placings
avoided the GFC and the extreme volatility in its aftershock so ruled out listings before end 2010.
further whittled down by another 130 or so which didn’t appear in the datamining:…

Unlock with Email

Unlock with Google

Already have an account?
Login here

Howard Marx

19th Jun '19

I'd be keen to know the performance of the 133 companies that 'de-listed'

This could radically change or reverse your conclusions, especially where the reason for delisting was a takeover.

JohnEustace

I have only bought two IPO's at launch, Alpha FX and Manolete, both of which make your good performers list. My reason in both cases was that they were listing to raise capital to expand their businesses to meet existing demand.
In Alpha's case they needed more regulatory capital to be able to take on customers that they were having to turn away and in Manolete's case they needed to open new offices. Those seemed "good" reasons for an IPO to me. There are plenty of "bad" reasons like the private equity cashing out element of floats such as Aston Martin.

I'm reading your conclusions as being "Run the winners, cut the losers"? But that there are more losers than winners.

I'm not sure there's a substitute for doing the fundamental research, reading the prospectus, etc. And if time doesn't allow that I would just stay away based on the win/lose ratio you found.

jonesj

Thank you for producing this outstanding piece of analysis.

I won't say much more before studying it comprehensively.

dscollard

Indeed but it will be a qual assessment : the list includes Patisserie Valerie, Naibu, MYSQUAR Cloud Tag ... with a concentration of companies domiciled in BVI and the Caymans.

Academic research puts the under performance at -8% to -28% so pretty much in keeping with the above

albeit the win:loss in Year 1 is outstanding and the risk is kept to 25% so the expected value of the multibaggers is the real prize

Gromley

This looks like a great piece of analysis and deserving of a much more detailed read and consideration (bring on the weekend!)

In the meantime though a couple of questions (and one piece of pedantry) sprang to mind on an initial skim.

The questions :

What was the step that got you from over 3000 to 530 stocks - that wasn't immediately apparent and could possibly create an unconscious bias.
What (post IPO) period do your results cover? I'm guessing that it might be from date of IPO to now. If that's the case then isn't there a risk that the more recent IPOs are skewing the data, especially if the first year does indeed tend to be more positive?

And my pedantry : Well I think from reading your text I think there were actually 23 multi-baggers and 28 baggers. (In fairness, you did define multibaggers as "any share that has more than doubled" so in your terms you are correct).

Anyway that's just from a very quick skim - I'll try to have a much more considered look in due course.

Thanks for sharing this.

InvestorJohn

Well look at that...

A stop loss of 25% would have kept 143 of the 166 winners in play and only stopped-out 7 of the 51 multibaggers while removing 215 of the 230 losers..

This was very interesting indeed! I cut losers at between 20 and 25% max of 30% in really exceptional circumstances so nice to see that it works in this scenario

Thanks Gromley, as ever insightful remarks

The 3000 refer to IPOs and placings dating back to the 1990s: the universe was selected from 2011 to 2019 to avoid the GFC and its aftermath and to concentrate on our new ultra low interest rate world. The selection criteria was purely date based and had no bias as they were fully representative of that total period

TBH my bias was upended by the research which is always good.

Given comments on win:loss ratios it strikes me that i should add something on expected values: that is where all the good stuff is. I intend to write some more from it as it is a nice microcosm on trading versus investing and the role of execution and risk management

Yes there is a skew from recency in the data: it's in the notes on the full version but I've appended bellow it doesn't materially change the outcome except to worsen the median return from the total universe. It underpins that IPOs have a maiden year or two when they tend to be profitable before diverging into winners and losers, multibaggers and dogs

I also include a plot of the total IPOs as a function of time.

3. Relative performances have been derived using time from IPO date: this will be skewed to some extent by more recent data: 30% of the total sample were placed in the last 3 years (removing these drops the median return to -24% for the total universe

underscored

Chinese AIMs and sector analysis (resource stocks)?

Appended to the full article - see below

Appendix: A note on Sectors, Countries
The total universe of 408 was analysed by sector, country and market as were the winners and losers to examine sector and country over/under-performance. Junior oilers and miners are fairly notorious for overpromising and under-delivering with sometimes spectacular losses. Many investors will shun companies from some countries owing to track records of frauds and limitations on prosecution in the event of fraud.

Figure 8 shows that more of the winners came from Financial Services, Support Services with an even more pronounced concentration of Support Services and Software companies in the multibaggers. Mining and oil and gas producers did concentrate in the losers with the ratio of mining doubling from 6% in the total universe to 12% in the dogs;

Figure 9 shows a similar segmentation by country, I suspect that this would change further if the 130 delisted companies were included since it contains several from Israel and China which have proven to be frauds. The plot shows a doubling (albeit from a low base ) of Israeli and BVI companies in the dogs : the percentage of UK companies increased from 81% in the total universe to 87% in the winners and decreased to 73% in the dogs. This does seem to confirm the bias against companies from these countries.

jwebster

20th Jun '19

Thank you for this useful and thought provoking article

"further whittled down by another 130 or so which didn’t appear in the datamining: the bulk of which seem to have delisted (some may have changed name)"

As mentioned above this would create some bias. Delisted implies take over or bankruptcy which are quite binary outcomes for investors' profits. Appreciate it could take too long to manually find and plug in the data though.

Looking at the chart of performance by time, it looks like (visually) a few jump up 50% - 100% within the first year, then not much happens for the next three years then from year five onwards the big gainers start grinding up with big wins over a few years.

As you say this implies a dual strategy - trading year one 'pops' and investing in growth companies four or five years post IPO when their serious growth phase seems to kick in.

This is a shape which mirrors the big tech firms in the USA as well, I recall Amazon, Facebook having the same sort of pattern. Year one exuberance for the new listing, then a few years of negative cash flow reality as the firm struggles to gain scale, burns money and holds successive funding rounds to stay alive. Then the (successful ones at least) hit critical mass on customers and earnings and zoom higher.

Thanks again for the article

Thanks, I intend to do a deeper dive on the big winners and characterise them in more detail

I am not sure bias is the right phrase for those that delisted - skew maybe but bias suggests a cognitive effect. Given the sample size and duration that amount is probably "about right" statistically given going broke or getting bought out are also outcomes of an IPO : it happens all the time and just is without bias.
.
As you say given the binary nature of that outcome then quantifying does become more manual but not impossible : at least on a first pass of "broke or bought": broke is easy as a zero, there is then a degree of bought wrt the IPO issue price. What it does add is the risk of losing everything which is of use.

timarr

The missing 130 IPOs represents nearly 25% of the total set - so it's clearly significant. To take a silly extreme position if the entire 130 went bust or got taken over at a significant premium then it would significantly skew the results.

Technically this is a bias, but not a cognitive one: it's survivorship bias, a form of selection bias, and the outcome is that the results usually look better than they really are. It's a well known problem in financial stock analysis and its always something to look for in backtesting: any back tests than only use current data are inherently flawed (personally I think all back tests are flawed, but I'm probably in a minority there).

Survivorship bias is often used by mutual fund managers to improve their performance figures. They never mention the failed funds that they quietly folded, they only promote the successful ones.

Even so, what would be interesting would be the common characteristics of the winners and losers. In the AIM research in this area the main factor was longevity of the company prior to IPO, but I'd guess levels of debt might be relevant as well.

shipoffrogs

We should assume that a company that has been around a long time pre-IPO is more likely to survive than a younger company, all other things being equal.
There's a 50% chance that the IPO occurs in the middle half of the company's life.

In 1969 J Richard Gott III worked out a method to estimate the expected life of the Berlin Wall (https://www.asc.ohio-state.edu/kagan.1/AS1138/Lectures/GottIII.html), which seems reasonable to apply to company lives.

Ever since reading about it I've paid attention to how long a company has been around when considering an investment.

No doubt the delisted companies do have an impact on the overall results

@timarr, I'll defer to your expertise on bias: my misunderstanding of survival bias was based on the exclusion of failures to improve the overall proportion of success: those delisting will do so both from failure and success so there is not a skew to either in terms of selection.

Clearly it is outstanding and needed to improve the overall interpretation

Thank you!

... my misunderstanding of survival bias was based on the exclusion of failures to improve the overall proportion of success: those delisting will do so both from failure and success so there is not a skew to either in terms of selection.

You may be right, and I don't know one way or the other. However, I don't think we can assure ourselves of that without looking at the actual data. Survivorship bias doesn't automatically give a skewed better result, it might equally give a skewed worse one if a lot of recent IPOs attracted sizeable bids.

If I were a betting man, though, I'd bet against it: IPOs are generally favourable to insiders who know what they're selling so it would be surprising if a lot of companies then got bought out at a significant premium.

The other thing to be wary about is outliers. Generally in any cohort there are a few superstocks that massively outperform and bias the results. If (say) you had two ten baggers in 100 IPOs and investors on average bought 1 in 10 of them you'd have a high probability of getting worse results than the averages would suggest. I'd suggest that the size of the outperformance is less than the proportion of the stocks that outperform.

It's interesting, though. I was long a refusnick on IPOs but I've bought a few recently which have done OK. Still wake up in a cold sweat thinking about Accrol Group (LON:ACRL) though ...

I would imagine the 25% of non-surviving stocks would be a mixture of takeovers and failures, possibly more failures. If the latter, perhaps the overall conclusions would not be too different (median IPO performance still poor).

Anyone planning to amend their investment process in view of this study will either have to make an estimate on the effect of the missing stocks, or do the work to find out how they departed from the market. In my case, since I am unlikely to become a regular participant in IPOs, making estimates is a more likely outcome.

@timarr, I'll defer to your expertise on bias: my misunderstanding of survival bias was based on the exclusion of failures to improve the overall proportion of success: those delisting will do so both from failure and success so there is not a skew to either in terms of selection.

Whilst I would agree that the bolded bit is true, I would hypothesise (from only casual observation) that "negative delistings" would probably be in the majority for this group. Whilst there will certainly be some that are taken out at a premium (to what?) there will also be outright failures, those that are bought out in a "rescue" and those where the board choose to delist out of frustration at poor share price performance (I've actually seen a few of those recently).

It is a source of great frustration to me personally that universally the providers of stock price information expunge any de-listing companies from the public record. Which is why I have been working on capturing and retaining contemporaneous price information for as many companies as possible . My work on this is unfortunately only comparatively recent, but I might still be able to cast light on the fate of some of the 130.

For the others, it might be possible to make some form of qualitative assessment of outcomes. Good old "investegate" (or UK Wire as it always will be to me) do provide a useful archive. (For example if anyone wants to read up on the demise of Woolworths over a decade ago it's all there for you.)

I have a number of data mining tools at my disposa,l so would be happy to have a look at what conclusions could be tentatively drawn from the 130 if you have a list of names and/or epics?

21st Jun '19

Thanks, very kind offer Gromley but I am ok to run them, maybe over the weekend: this has been a bit of a side project that grew organically as I found it of interest. I'll write a Part 2 to update

IPO Analysis: Part 1 - Assassins and Rabbits; Multibaggers and Dogs

Unlock the rest of this article with a 14 day trial

About the Author

Dscollard

20 comments

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

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