19 Apr 2023 5 min read

Diversification isn’t a free lunch, but we think it’s worth the price

By John Roe

I've lost count of the number of times I've said "diversification is a free lunch". But the longer I’ve been investing clients’ assets the more I’ve realised it does have a cost: regret.


To minimise regret, strong investment beliefs should underpin any investment strategy. The dual decline of equities and bonds in 2022 pushed many to question their belief in diversification. But as we move through 2023 and towards a possible recession, we remain stalwart believers, despite the costs.

The argument that diversification is the only free lunch in investing is at least 70 years old[1] and stems from the idea that by owning lots of different assets, an investor can achieve a better outcome for a given level of risk than with fewer, more concentrated positions. Intuitively that makes a lot of sense. But that perspective is very mathematical. Unfortunately, as humans we’re a complicated bunch and not entirely rational.

When an investor is diversified across a wide range of assets and regions, the great news is they’ll never have very large exposure to those that do worst. But the problem is they won’t have concentrated exposure to those that do best, either.

Dicing with diversification

Using dice as a simple analogy that allows us all to think a bit more like robots, a diversified strategy might be to take the average roll of six dice. A concentrated strategy would be to just roll the first dice and take that outcome. The former strategy is clearly less variable, but both strategies have the same average outcome. So, if an individual were to receive £1,000 multiplied by their dice roll, most people would take the less variable approach, all else equal.

In the case of dice, we know the outcomes are random. So, if someone chose to take the average score of six dice, and then a single dice roll was higher, they could largely shrug it off as a game of chance and it’s unlikely to change their beliefs of the future merits of their preferred approach.

Unfortunately, despite the great uncertainty around investment outcomes, we see people suffer much greater regret risk when they miss out on an outsized return in one asset class.

We can see an example of this in US equities over the five years to the end of 2021. Firstly, we show our estimates at the end of 2016 for the five-year expected risk and average expected return of lots of assets. The general point being that most investments incur some risk and that over the long term we’d expect to be rewarded for taking that risk.

We pick out US equities and emerging market (EM) local debt as two examples of investments. EM local debt has historically been a bit less risky, and so had somewhat lower expected returns than US equities, but the relationship between risk and reward is clear.


The actual returns over the following years inevitably turned out to be very different from their estimated average outcomes, even over one year; there’s about a one-in-three chance equity returns are 15% above or below their long-term average.

Depending on how events turn out, including random shocks like COVID and policy decisions like interest rates, some will do better or worse. In the period we chose, US equities were a big winner and EM debt struggled, as we show below.


With investments, once the outcome is known, it can be very easy to see the outcome as knowable in advance or inevitable; in this regard the perception is very different from with dice. This is formally known as hindsight bias and is a well-documented flaw in human thinking. This can help convince people that they can do better than to remain diversified.

It’s inevitable that in any period there will be some individual assets that do a lot better than a diversified portfolio. That could lead to a perpetual cycle of regret risk and negative feelings as well as the temptation to shift towards whatever has recently done well.

Yet to dismiss diversification on this basis would be to overlook the fact that given that diversification has the potential to reduce risk without lowering return expectations, logic suggests it could be possible to use diversification to increase expected returns while keeping risk constant. Also, with the dice example, investors should bear in mind that investment horizons aren’t typically fixed, so you might have fewer throws of the dice than you think, making a consistent, diversified approach more appealing.

For example, diversification can also potentially reduce the changes of negative outcomes for investors early in their careers – negative outcomes which might otherwise encourage investors to give up on investing altogether, a decision that could prove a costly over time.

So, as believers in diversification, how can we learn to cope with regret risk?

Three steps to reduce regret risk

Well, one can try to be more like a robot. In practice, that means taking some steps to prepare oneself for what’s coming and a framework for how to react when it inevitably happens:

  • Write down very clear investment beliefs, like that diversification is a free lunch for robots; they can then be referred to as a reminder of what you believe and why
  • Always take a look at what performed worse and imagine how you’d feel if that had been a much larger exposure in your portfolio
  • Accept that investment returns are hard to predict in advance, so a large element of the outcome isn’t known and is, in effect, a large random component. This can help you think more like the dice analogy, rather than having too much hindsight bias

If we do all of the above, we can reduce regret risk and the chance of following the herd into investments that have recently done well, making that diversification lunch taste a little sweeter.


[1] First attributed to Nobel laureate winner Henry Markowitz in 1952 but of course may really be much older. After all, Columbus didn’t discover America. 

John Roe

Head of Multi-Asset Funds

With failed football dreams behind him, John applies the same level of enthusiasm to investing and how to improve outcomes by battling behavioural biases. He leads on oil research, but also gets involved in a wide range of macro topics. That love of variety also explains his craft beer fascination. Hard to shut up, he’s a regular guest on Bloomberg, a conference speaker and an LGIM Director. His analytical thinking benefits from being an Actuary with an economics degree and having previously worked as a strategist at RBS.

John Roe