Thursday, November 12, 2015

The Example Larry and Sergey Should Follow (It's not Buffett)

Originally published at

Coverage of Google’s restructuring has focused on the idea of Alphabet as a new kind of internet-era conglomerate. Optimists compare the new entity to Warren Buffett’s Berkshire Hathaway; skeptics point out that the conglomerate model seldom works.

But there’s a better comparison than Buffett: it’s John Malone, the “mastermind” who built the cable TV powerhouse Liberty Media. Malone is a brilliant financial engineer, who creates separate capital structures — each with a unique stock — for his different lines of business. Liberty Media, Malone’s holding company, owns a portion of the stock in each business. This approach allows Malone to attract equity and debt investors whose preferences regarding risk and payoff horizon match those of the business in question.

Having separate stocks also allows Malone to place bets when he believes that one line of business is not correctly valued by investors. If he thinks one of Liberty’s companies has a rich valuation, he can sell more stock to outside investors at a high price. Or, if investors are enthusiastic about a new business that Liberty is incubating inside one of its existing companies, he can spin out that business as a “newco” with its own stock. Conversely, if Malone thinks that the market has undervalued one of the companies, Liberty can buy back its shares.

As The New York Times put it in 1997, when describing the spinoff of TCI Ventures, which provided innovative new broadband internet services:

For Mr. Malone, the spinoff is a way to get more value from assets that he believes have been neglected by investors and overshadowed by Tele-Communications’ prime business, the ownership of cable systems with 14 million customers.

In this case, Malone traded investors some of his shares in the core cable business for a bigger equity stake in risky new ventures, which he thought the investors didn’t fully appreciate. Sound familiar?

To be clear, Google has thus far given no indication that it plans to create separate stocks. But its new structure would making doing so much easier. As Todd Zenger wrote here earlier this week, many investors want to buy stock in a safer, maturing search advertising business without buying into Google’s riskier moonshots. Alphabet promises to make the financial performance of its companies more transparent, but the big change would be if some of these companies traded under different stocks.

At that point Alphabet could sell shares of Google to more risk-averse investors; it could raise capital for its self-driving cars from the same sort of people who are betting on Uber or Tesla; it could seek funding for its renewable electricity projects from energy financiers.

A potential downside to this strategy is that it would make it more difficult for the ad business to directly cross-subsidize riskier tech investments. Formal contracts would be required for transactions between the companies, and each side would need to be able to make the case that a transaction was in the best interests of its respective shareholders.

And there’s still the question of why a holding company with full or partial stakes in a diverse set of somewhat unrelated businesses can be expected to create value. We know from years of academic research that on average, diversification into unrelated businesses doesn’t create value for shareholders. Why should Alphabet be different?

But Google’s founders seem more than happy to challenge the conventional wisdom. “Google is not a conventional company,” they wrote in their original founders’ letter. “We do not intend to become one.” Google’s founders have strong views about where technology is going, and where it should go, and they don’t seem that bothered if others disagree. Malone — also a technology visionary — feels similarly; he’ll bet against the market when he thinks that it is wrong. That’s something he has in common with Larry and Sergey.

Saturday, July 4, 2015

Resources for Getting Started With Data Science

Many of my MBA students who pursue jobs in product management, on growth teams, or as founders want to build data analytics and data science skills.

Peter Jamieson, a former student of mine who is now a data scientist at Pixability, a Boston-based video ad analytics startup, shared the following suggestions for getting started with data science:

Here's a list of some of the resources that have been helpful for me as I've gotten up to speed in data science. 

My go-to tool for working with data these days is Python. It can be tough to get everything you need to get started. Fortunately, Anaconda and Enthought both offer free distributions that are nearly plug-and-play. 

IPython is a tool, included in those downloads, that (among other features) lets people manipulate snippets of code in their browser. Becoming comfortable with launching and using IPython notebooks helps you take a number of online courses and share code. 

There is an incredible amount of content out there and it can be hard to sift through it all. I've picked some highlights, some of which are focused on business understanding, some on implementation. 

  • Data Science for Business:  Some math but no programming; a good resource for getting started that provides business use cases.
  • DataSmart:  Implements popular data science models in Excel and contains an intro to R. I'd recommend this for people without a programming background who are just starting out.
  • Here's a list of more advanced machine learning titles from Quora -- very technical, not for the faint of heart.
  • Other (free) books covering everything from coding to managing data science teams.


Online Classes:


Other Lists -- places to look if you can't find what you want above:

Thanks to Peter for sharing this list. Readers: if you have items to add, please use the comments section.

Addendum, Oct. 26, 2015: has compiled a list of free ebooks about data science (thanks, Brendan!).