Diversification is sometimes described as “the only free lunch” in investing. But is it? Not in the kind of fundamental value investing that I do.

Increased diversification comes with two potential costs:

At a certain point, new investments are likely to yield increasingly lower returns.

The time required to underwrite new investments reduces the quality of the underwriting of existing investments.

I will examine these costs in more detail below, but first let’s consider just how concentrated is too concentrated.

If additional investments might lower a portfolio’s expected return, why not invest everything in the security we expect to generate the highest rate of return?

The answer is simple: We can be wrong. Investing, no matter how skillfully implemented, rarely deals with certainties: The best we can accomplish is to heavily skew the odds in our favor. Therefore, building in a margin of safety at the portfolio level, as well as in underwriting each investment, is critical.

So if we can be wrong on any one judgment, then at what point will our portfolio lose this desired margin of safety?

In my view, it is when getting any one decision wrong can permanently damage the value of the portfolio to the degree that the rest of the investments are unlikely to recoup the losses in a year’s time.

Notice that I said decision not investment.

Why? Because the long-term business success of several investments may share the same driver such that being wrong on that one driver could permanently impair the value of multiple investments.

To put some specific estimates behind these ideas, I try to avoid any decision that could permanently damage the portfolio by more than 10%. The rationale is that if that 10% loss, however unlikely, occurs, the other 90% of the portfolio, at a reasonable rate of return, could be expected to offset the damage.

Related: What are Mutual Funds? 

Because of that consideration, I normally have a minimum of 10 investments in the portfolio and limit portfolio at risk (PaR) — defined as position size multiplied by the downside to the worst case intrinsic value estimate — on any one investment to 5% at cost and 10% at market. I also assess the potential correlation of long-term business outcomes among the different investments in order to manage the portfolio’s correlated assets to be consistent with the 10% criteria.

The Costs of Diversification

As for diversification’s downsides, let’s consider two very different market environments: one in which many potential investments offer identically high prospective internal rates of return (IRRs), and a second wherein each successive potential investment promises much lower IRRs than the last.

Portfolio Concentration

In the first scenario, the cost of diversification is low based on how much it would reduce expected returns, and so a diversified portfolio makes sense.