This weekend, Warren Buffett released his annual letter to Berkshire Hathaway shareholders. As always, it was full of memorable quips and the Oracle’s take on where the US economy currently stands. But what most people seem to have missed is that this year’s Berkshire brk.a letter also contained some of the best investment advice the Oracle of Omaha has ever written down. That’s perhaps because it was in a portion not devoted to investing or the stock market, but insurance. In the section of the letter detailing the results of Berkshire’s General Re subsidiary, there was an interesting tidbit on the four disciplines that must be adhered to in the insurance business. Buffett wrote:

At bottom, a sound insurance operation needs to adhere to four disciplines. It must (1) understand all exposures that might cause a policy to incur losses; (2) conservatively assess the likelihood of any exposure actually causing a loss and the probable cost if it does; (3) set a premium that, on average, will deliver a profit after both prospective loss costs and operating expenses are covered; and (4) be willing to walk away if the appropriate premium can’t be obtained.

This got me thinking about portfolio management (which is pretty much all I ever think about, let’s be honest) and it struck me that the parallels between insurance and investing were pretty perfect. What else, after all, does a portfolio represent but insurance for the expenses and needs of the future?

Let’s take these four disciplines one by one:

Understand all exposures that might cause a policy to incur losses.

First, let’s define “losses” from the standpoint of a long-term investor: It’s the possibility of having a quantity of money permanently go away. This can happen in a bond or an individual stock. In some cases, it can happen in a mutual fund. It can even happen in an index fund if it tracks an index that makes a high it never returns to.

But a diversified portfolio will almost never incur a permanent loss other than due to the behavior of the investor holding it. By selling asset classes at a market bottom or wagering too heavily in an obscure area of the market, investors can absolutely cause themselves permanent losses.

The only way to guard against this outcome that I am aware of is diversification and systematic rebalancing. A diversified portfolio cannot avoid periods of loss, but these periodic declines are more likely to be drawdowns as opposed to permanent losses. A Japanese investor with a 100% domestic stock portfolio invested in the Nikkei could still be in drawdown from the market’s peak 25 years ago. That same investor with a globally allocated, diversified portfolio would have recouped his losses long ago and would be doing much better today.

Conservatively assess the likelihood of any exposure actually causing a loss and the probable cost if it does.

What is your potential downside should adverse events or market activity hit your portfolio? What is the probability of heavy drawdowns? How bad could it get? How much risk are you willing to endure—in the form of volatility—for a given return? Having an idea of the historical probabilities of asset class declines should inform the way a portfolio is structured. It is said that by seeing to your risks, you can let the potential upside take care of itself.

Set a premium that, on average, will deliver a profit after both prospective loss costs and operating expenses are covered.

This cannot be emphasized enough—there are risks that pay and risks that do not pay enough to justify taking them.

Concentrated portfolio risk falls firmly into the latter camp. Betting heavily on a small amount of investments is how fortunes are made, but only in hindsight. It doesn’t work out well for the majority of people, but you rarely hear about them. You will hear a lot, always after the fact, about the supposedly brilliant people who made concentrated bets and hit the lottery.

Different asset classes have different risk and return probabilities. Equities return the most over time, but their holders have to go through more duress than investment grade fixed income, for example. Balancing your tolerance for potential losses against your need for future gains is the key to the whole endeavor. In this analogy, your “operating expenses” would be the future liabilities you’ll have in retirement or even sooner if you’re saving for your children’s’ education. Will these operating expenses be covered under a variety of potential market outcomes? Talk to a financial planner if you can’t answer this question on your own.

Be willing to walk away if the appropriate premium can’t be obtained

Some investments offer a lot more risk than what you’d deserve to earn in “premium,” i.e. returns, for your trouble. A recent example would be the MLPs that many investors had been holding in place of where they’d traditionally be buying bonds. For a 7% pass-through dividend yield, investors piled into MLPs. But as oil prices have dropped, the MLPs have on average dropped 40% in the past year, a lost equivalent to six years worth of the dividend income they were chasing.

Investors increasing their current yield by taking credit risk in junk bonds have recently learned a similar lesson. At some point, investors who are conflating high-yielding consumer staples stocks with bonds or who are taking interest rate risk in long-dated Treasurys will see drawdowns as well. So will investors taking counter-party risk via swaps and collateralized debt securities issued by a financial institution.

Bottom line: There’s no free lunch on Wall Street and just about every asset class carries its own form of idiosyncratic risk. Still, there’s nothing wrong with taking risk today in order to earn future rewards. Just make sure you’re being paid for taking it in the form of an adequate potential return. Warren Buffett clearly does. And his returns haven’t been all that bad.