Google To $2000/Share? Somebody Muzzle Blodget

Henry Blodget made his name by predicting outlandish price increases for Internet stocks in the late nineties. A lot of people lost a lot of money (or, all their money) by listening to his recommendations. The government charged him with securities fraud in 2003 and he was subsequently banned from the securities industry for life.

But Blodget is a bit of a one trick pony, and he likes to stay in the headlines. So he continues build cases for big valuations of Internet companies. The only difference is he publishes these thoughts on his blogs. And people still listen to what he has to say.

He isn’t always bullish (he’s recently trashed Yahoo and eBay). But he can’t seem to contain his regular predictive outbursts that such-and-such stock is worth massively more than it is now.

When he’s talking about Facebook being worth $6-$20 billion that’s ok, because it isn’t a public stock and no one is going to go out and throw away their life savings. But when he builds a case for Google’s stock to go to $2,000/share, he’s crossing a line.

Remember a couple years back when some analyst floated the idea that Google could eventually be worth $2,000 a share–and was ridiculed from coast to coast? Well, first it’s worth noting that Google is now almost a third of the way there. Second, it’s worth noting that $2,000 a share would mean a market cap of about $750 billion, which–given a reasonable time horizon–just isn’t that far-fetched.

The problem is that Blodget, like all analysts, build authoritative sounding but essentially bullshit predictive models to back up whatever prediction they’ve just pulled out of their ass. When Blodget predicted a massive Amazon price increase in 1998, for example, he used three models: price to revenue multiples, revenue growth assumption, and an earnings multiple growth model. When you read it, it sounds like he really knows what he’s talking about. But he’s really just predicting future growth based on past growth and backing it up with a lot of smoke and mirrors. If the data doesn’t fit or doesn’t exist, a common trick is to use a competitor’s or analogous company’s data instead. One way or another, a model can be built around that headline grabbing prediction.

Blodget builds his Google $2,000 prediction on similar models – in this case he talks about a multiple on free cash flow.

But market conditions change and these models just aren’t capable of taking that into consideration. Anything could derail Google’s current growth rates – a credit crunch, a housing collapse or a recession could all have a big impact on consumer spending and the advertising market, and impact Google massively. The market, over the long run, is fairly efficient at predicting the value of companies. If Google really was going to go to $2,000 per share, it would be priced there already, minus only a discount based on the time value of money. It isn’t, and so if you’re betting that Google is going to $2,000 that means you are betting against the market and all its participants. And all you have to go on are Blodget’s bullshit predictive models.

We are often criticized for being overly optimistic about young startups. That’s worth arguing over, but if we get it wrong at least we’re not moving the market. Venture capitalists do their own due diligence and don’t last long if they place too many incorrect bets. But when journalists start writing about public companies, stock prices can (and do) move, and people can lose a lot of money.

Blodget wants to stay in the headlines, but he has little concern for those that follow his advice. “If it doesn’t happen, don’t come whining to us,” he says in the Google post. That’s a disclaimer of sorts, but it also shows that he’s not all that interested in the fallout that may occur from his words. And if his past predictions are any indicator of Blodget’s ability to pick stocks, a fallout is almost certainly coming.

Update: Well, I sure was wrong. Blodget nails it in a new post valuing TechCrunch at a cool $100 million. My thoughts at Crunchnotes.