Stocks A to Z / Stocks B / Berkshire Hathaway (BRK.A)
No. of Recommendations: 1
An interesting study of buybacks and how to predict them.
"...buybacks tend to increase a company's stock price."
"Machine learning was applied at the end of each year to predict the likelihood of a buyback occurring in the following year."
"This demonstrates that share buybacks in the following year can be predicted with a high degree of accuracy in all years, and that past buyback and other financial data can be used to predict future buybacks."
"...when the portfolio [JP stocks] was composed of stocks with a high probability of share buybacks, the cumulative excess return ended up at about 30%..."
http://www.nikkei.co.jp/nikkeiinfo/en/global_servi...Shaun
No. of Recommendations: 4
The problem with sending data through the Machine Learning black box, is that you're never quite sure what data they focused in on. Whatever data it chose, we should be aware that *Berkshire's* repurchase plan is not like those of most companies. It has a valuation limit above which Buffett will chose to keep the cash or find another opportunity, and the size of the repurchases may not rise and fall with company earnings like other repurchasers.
Excess returns of 30% is really why I clicked the link. But that's the cumulative return from 2013 to 2022. And the methodology has just treaded water the last 5 years. From 2013 to 2017, the portfolio gained maybe 32% excess returns, then gave some up, made a bit more, and ended back at the 30% reported number.
No. of Recommendations: 9
"Excess returns of 30% is really why I clicked the link. But that's the cumulative return from 2013 to 2022. And the methodology has just treaded water the last 5 years. From 2013 to 2017, the portfolio gained maybe 32% excess returns, then gave some up, made a bit more, and ended back at the 30% reported number."
I think you hit the nail on the head.
Anyone trying to discern information from the past 1.5 decades (actually more) is ignoring just how unique the recent past has been. Never in history has there been such a loose rate policy (supercharged with QE). Money has never ever been cheaper than it has been in recent history. Take lessons from an anomaly at your own risk.
No. of Recommendations: 26
I have a somewhat more fundamental concern that there might be a bit of the post hoc ergo propter hoc variety.
Do firms doing buybacks do well because of it? Or, for example, do firms making more cash than they can usefully invest do better than average, and those firms tend to have a limited range of option for the excess?
I once did some consulting at the HQ of Bristol Myers Squibb. This was a fantastically profitable business at the time. Incidentally not because they were particularly well managed (it would turn your hair white), but that they were in a business through which flowed oceans of cash.
Anyway, I noticed that their head office had an extraordinary art collection, everywhere you looked. I imagine many such very fat headquarters were like that. So, imagine a study that attempts to show that a great art collection causes great shareholder returns. It would be a great fit to the data, but in fact it's just synchronicity: neither item (art and shareholder returns) caused the other, but rather both were the product of a hidden third cause: tons of money rolling in the door.
Now, maybe buybacks do cause higher prices a bit at the margin. But I don't think that this type of retrospective study can demonstrate it in a meaningful way.
Jim
No. of Recommendations: 0
The problem with sending data through the Machine Learning black box, is that you're never quite sure what data they focused in on.
Well, the age of reason certainly turned out to be a short one! My ancestors did a ton of empirical observations and prescribed lifestyle accordingly without bothering to find cause and effect. Then these arrogant "scientists" came along with their long chains of cause and effect and laws of nature. Never mind that that allowed them to invent and discover and build on previous knowledge. Let's go back to correlations and forget causations. Long Live AI, all hail Renaissance Technologies!
