If you'd bought $10,000 worth of Netflix at the IPO in 2002, you'd have about $8.2m today. Unless you'd sold some along the way. And the second criterion is a much more demanding one—in the last year, Netflix's cumulative volume was just under 2x its current shares outstanding, or, put another way, the average holding period for a given share is six months. That is, of course, a very skewed distribution, with plenty of market-makers holding for very brief periods, a population of shorter-term traders, and some long-term buy-and-hold ones. But even the buyers-and-holders will typically flip a typical position within a decade or so.

This gets at what is sometimes a mystery in asset management: what are long-term concentrated investors really being paid to do? There are some funds out there that have a very short list of positions, often fewer than a dozen, and where their 13Fs mostly reflect changes in assets under management rather than changes in the structure of the portfolio. The mystery is: what are they getting paid to do?

The advertising business has an answer for this: there's a story that a cosmetics CEO, probably Helena Rubinstein, complained to her ad agency that she was paying them an enormous amount of money to keep running the same ads. The agency's CEO—if the story is indeed about Rubinstein, would be David Ogilvy—gathered the entire team that worked on the account to meet with her, and informed her that she was paying a premium to the agency that would keep running the ads that worked instead of insisting on doing something different.

There are offsetting cognitive biases at work. On the one hand, it's very hard to reconsider an idea if it's worked well, even if circumstances have changed. On the other hand, it's sometimes even harder to stick with something that worked, especially when loss aversion kicks in: in 2011, a 3,400%+ cumulative gain in Netflix turned into an 800% cumulative gain in just a few months when the stock sold off during an early attempt to split the DVD business from streaming. If a stock gets more than chopped in half in a few months, and it's a business that's making big fixed investments that it amortizes over what is hopefully a growing base of happy customers, and it suddenly has a shrinking base of furious customers, it isn't too hard to pencil in some numbers that take the stock to zero.

What long-term concentrated investors are paid to do, sometimes quite well, is to have a level of fundamental conviction that's compatible with a portfolio of five high-beta stocks.That's not easy to do: in general there will be one story told by the numbers, and a more complex story told by qualitative factors. So the job of the long-term investor is to ensure that the narrative that the company's accounting shows is compatible with reality, basically a continuous cross-examination of the management team.

That can be uniquely hard to pull off, because CEOs are very happy to have their shares owned by this kind of company: it's a shareholder base that isn't likely to be activist-friendly (they'd rather sell), that doesn't need the business model explained, that is probably going to be around for a while, and that will probably not waste much time at conferences asking about the nuances of short-term guidance. They do go away eventually, and that actually comes with risk for the management team: it doesn't mean anything if Balyasny blows out of a position, but if a concentrated, low-turnover fund does, it's a stronger signal. But the more the fund is aligned with management's views, the more that kind of decision is one that 1) will be telegraphed in advance, and 2) is more the fault of management.

Over long holding periods, companies evolve, especially the best ones. Someone who's held Netflix since the IPO has owned a direct-mail DVD distributor, a money-losing distributor of third-party streamed content, and a streaming service attached to a movie studio with productions that span genres and languages.

Which means that one of the knocks against this kind of investing doesn't hold up: there's a straightforward reason to think that if there are two strategies that produce equivalent returns (and the investor in those strategies is not taxable), the best one is the one that makes the most trades—it has the highest sample size for good investments, and is thus the fund that gives allocators the most statistical confidence that the fund knows what it's doing. It's still directionally true that a lower-turnover fund has a smaller sample size, but given enough time its portfolio achieves diversification purely because the underlying companies evolve so much, and it turns out to have been making the repeated decision to hold in addition to the one-time choice to buy.

In The Diff, we’ve covered long-term performance from standout companies many times:

• Google’s growth rate has been surprisingly durable over long periods ($).

• Sometimes, companies can’t predict ROI until they’ve made the investment.

• The best way to think about growth is to work backwards ($).

• In theory, indefinite growth leads to infinite value ($). In practice, growth is not indefinite.

• This piece on mental math is mostly not about investment research, but does touch on one approach: “Looking at a growth company is basically like the process of canonizing a saint: you have some basic criteria to check off, but you spend most of your time on assessing the truth value of a handful of extraordinary claims.”

Join the discussion!