Its IPO season and many companies have come for the IPO in the past few
IPO’s or primary market issues, investing is popularly followed by many
investors, majority get burnt badly while some make money over long periods
of time. Let’s see what the data speaks.
What We Did?
We gathered all book building IPO’s data from past 17 years,
approximately from 2003 to 2020. We calculated their prices at three dates,
listing date, one year and two years. Then we calculated their Logarithmic
returns from the offer price for all three horizons. (This assumes
that you would get the allocation of IPO – which is in itself a big
assumption – however this won’t matter later as we progress on)
If I give you normal stats, it is quite easy to say that more than 50% of the IPO’s give around 0 returns in from offer date to listing date, same is the date for 1 year and same is it for 2 year. However, this gives us the bayes information that is the role of luck (50:50). But this is not true lets go deep.
Since we are considering the horizon of 2 years, In markets it is few extreme outliers both positive and negative which really determine your results (Few events determine majority of information, think about taking average income of USA, if you include Bill Gates in it, the average is something else) and some of the extraordinary successful or failed IPO’s will have most of the information. If we study just the average cases we will most likely noise not true factors which really determine success or failure of 2 years.
Now before we need to define what is extraordinary:
First we arrange all datasets ( listing returns, 1 year returns and 2 year returns) on the basis of returns starting from least.
1) Then we select companies in the first 25% of the data (Meaning least returns) and top 25% of the data( meaning highest returns) from the listing returns dataset.
2) Then we again repeat the above exercise in 1 year data set and then in 2 year dataset.
3) Finally we pick companies common in all 3 datasets.
So our negative outlier would be the one giving us the least returns in all 3 time frames, while positive ones would be the maximum ones.
Identifying Factors: (Real Job Begins)
Now most important part would be to find 3-5 to factors which really determine very failed IPO’s and very succeeding ones. So we need to go back in time and ask what are the commonalities in losers and winners? Could that have been identified beforehand.
Suppose its the market size of the IPO, once a factor is decided it needs to be tested on the dataset and if it does not work we need to go back again and find a new one. So we will clearly follow a method of elimination and tinkering (Hit and trial).
Like if you think its market size: Study the market size announcements of the IPO prospectus, see if you could have identified early and it differentiates really between a very successful, normal and failed IPO’s. Chances are its very tough to find such factors, most of the time you would need to go backward and forward.
Like one thing is for clear the market environment of next 2 years really determines the IPO returns of two years. As we can see most extreme successes and failures are clustered around at same times. Failures in bear markets and successes in bull ones. Though the key is can you determine market environment for next 2 years? Chances are its very tough.
Though one can hit and try and find something. Even if you find one factor you if you add another one you need to remember predictive power of 2 factors should be better than one. Your system should help you predict at least more than 50% of successful outliers, avoid large variety of bad ones. It is quite likely extreme failures have a different factor, however avoiding extremes losses is the key and helps build a robust approach.
Remember 3-5 factors max, else one gets lost in factors.
Even if you find something, and test it, you may use it in real markets, but start with less money as mistakes initially shouldn’t kick us out of the game. Remember this does not tell us the path, as stop losses may not work, you may have extremely high outliers, however the journey may be very volatile for two years.
Conclusion: So back to the starting question, are IPO’s for 2 years really worth it? Our answer as of now if you could predict market environment for next 2 years then may be. For us the answer is overwhelmingly NO. However, may tinker a little bit and may change this in future. But we won’t allocate anything meaningful to IPO’s as of now, and we won’t recommend anybody. As even the bayes experience tells us, IPO’s are a losing game and we don’t have the confidence to go with huge sums in the same.