Why I’m Against Optimisation

Whenever I’m browsing investing and personal finance content I see a word that makes me sick:


We are told we need to optimise everything.

This approach is wrong.

By “optimising”, you increase vulnerability and reduce robustness. This happens for two reasons.

First, you can’t increase the probability of a better result in one dimension without sacrificing another dimension. It’s a trade-off. There is no free lunch. If you’re an Economics first-year student who for some sick, twisted reason wants 100% in “Introduction to Statistics”, your “UK Economic Policy” mark may suffer as a result.

Second, optimisation usually involves some type of maximisation, which involves pushing to the limits and trying to get everything perfect. Mistakes are both more likely and more costly. Consider the likelihood and severity of a crash of skiing at 30mph vs. 60mph. I’ve tried both – the former is vastly more pleasant (but not as fun).

Optimising is particularly ill-advised in the realm of personal finance and investing. Because this world is complex. Shit happens. Things change. In complex domains, where outcomes and probabilities of said outcomes are unknown, it’s better to be robust than optimised. You don’t know what’s going to happen, so you must be able to cope with multiple different scenarios.

4 Bad Ideas

Building the perfect portfolio

The most frequent manifestation of the optimisation mistake is in trying to construct a “perfect” portfolio.

Academics, and even some practitioners, will have you believe that there is an optimal way to build a portfolio (Modern Portfolio Theory comes to mind). These people claim that using fancy mathematical models and (naive) statistics will maximise your expected return whilst minimising the “risk” associated with your portfolio. This is a lie. Wrong for a variety of reasons, but the biggest errors with these types of approaches are the result of some type of failed statistical inference.

The second flavour of optimisation in this area comes in a simpler form: the psychotic devotion to one asset: “Just buy the S&P 500”; “Buy BTC”; “Cash is king”; etc.

Don’t don’t do that. Don’t buy that. No it isn’t.

By trying to choose the best asset to invest in, these people are completely reliant on the asset they choose to be the best. They are vulnerable to this choice.

Specifics aside, the problem with both of these methods is the attempt to maximise returns. Your eyes will get gouged out trying to do this. So don’t. A better approach is to try and reduce the probability of a severely negative outcome.

Because expectation is not reality. Expected returns are good n’ all but you only actually experience O.N.E rate of return. This is why we are paranoid about bad outcomes. Focus on the downside, let the upside take care of itself.

Archimedes told me to

A by-product of this fundamental error is the use of leverage. A model might imply that you could generate more return at an acceptable risk level by using leverage. So you use leverage.

MPT tells use to use leverage. Source: Flirting with Models.

Some of those who blindly invest in one asset – some, not all – know how to comprehend numbers and look at pretty pictures known to us as charts. They see these numbers and see that they are rising over some period of time. They see that line go up in picture. They use these observations to make the genius prediction that numbers and line will continue to go up in future. Then comes the real mind-bending idea. “What if”, they think, “I take on debt”, they usually don’t know the word ‘leverage’, “and use money from debt to buy more…”. Eureka.

Hopefully you have realised the problem with this approach. A leveraged position leaves you extremely susceptible to decreasing prices. If the chosen asset isn’t the best thing since sliced bread, your leveraged position will be toast.

“Good” debt

The use of debt is also popular outside of portfolios. A key tenant of personal finance is the existence of “good” and “bad” debt. The former being of the low-rate, long-term, part-of-an-optimised-personal-finance-structure variety, the latter being of the BNPL, credit-card, loan-shark type.

What this classification of debt as “good” misses is that life is not purely about what the best expected payoff is. If I wanted to increase my returns I could murder my grandmother, blame it on my dodgy cousin, invest the substantial inheritance (I am her favourite) and reap the rewards 20 years from now.

Even if I were one of those losers who make most life decisions based on maximising the size of their net worth, it’s not clear that utilising debt to increase expected payoff is a good strategy. Because this payoff is expected. What if something goes wrong? What if you lose your job? What if your grandmother needs private medical attention after a murder attempt by your dodgy cousin (it was him, trust me)? You have to pay that debt. No matter what.

Building the perfect personal finance structure

Debt is just one part of constructing the “optimal” personal finance structure. Everything is perfect. Everything pays off just in time. You have just enough moving to various accounts to keep the machine ticking over. Just.

Because this maximises how much you save and invest. But, as we have discussed above, if anything goes wrong you can be left scrambling. The time required to set this up is also usually not worth the extra money generated from it. Time is an investment cost, too.

Taking a Step Back

These manifestations are just applications of the same fundamental idea. I’ll repeat what I said at the start of this piece: in complex domains, it’s better to be robust than optimised. Because it’s impossible to get everything perfect when we are dealing with uncertain outcomes. This means that by sprinting, by committing to one particular worldview, a small error can knock you severely off course. And by using resources to maximise something, you have fewer resources to commit to other areas, to make yourself robust to other possible scenarios.

This approach is also in violation of 2 of our principles, giving us – not that we needed it – more reasons to be sceptical:

  1. Start with risk. Optimisers try and maximise expected return, fuck the risk. Risk is secondary. This leads to disaster, eventually.
  2. Focus on the big picture. These people necessarily think in terms of detail. They try to get every little thing just exactly right. In doing so they inevitably neglect the bigger picture.

Principles aside, the real reason optimisation drives me insane is the arrogance required to pursue it. You must believe that you know everything, that you know what’s going to happen,  that you are right and others are wrong. For a fact.

The truth is we know faaaaar less than we think we do. Ignoring this will end in tragedy.

What do you think?

This site uses Akismet to reduce spam. Learn how your comment data is processed.