I post a lot of numbers. A number is only worth repeating if you can trace it back to where it came from, so this page is the reference library behind my posts. Every finding below links to the original paper or publication. Where a figure I had seen quoted second-hand turned out to say something slightly different in the original, I have used the original.
1. Most stocks never beat cash
In 2018 Hendrik Bessembinder, a finance professor at Arizona State University, studied every US common stock in the Center for Research in Security Prices database since 1926. His finding was blunt: the majority of those stocks delivered lifetime buy-and-hold returns below one-month Treasury bills.
The part that gets quoted most is the second finding. Measured by lifetime dollar wealth creation, the best-performing 4 per cent of listed companies explain the entire net gain of the US stock market since 1926. The other 96 per cent, collectively, matched Treasury bills.
What I take from it. Share market returns are extremely skewed. A small number of very large winners do almost all of the work, which is why a thinly diversified portfolio so often trails the market: the odds are that it simply does not own the few names that matter. It is also why selling a genuine winner too early can be the most expensive mistake in a portfolio.
Journal article · Abstract (RePEc) · Morningstar interview with the author
2. Simple equal weighting is hard to beat
In 2009 Victor DeMiguel, Lorenzo Garlappi and Raman Uppal took 14 portfolio construction models from the academic literature, each designed to find the “optimal” mix of assets, and tested them against the most naive approach possible: split the money equally across every asset, known as 1/N.
Out of sample, across seven different datasets, none of the 14 models consistently beat equal weighting. The models looked excellent on the data they were built on. The problem is that they rely on estimates of future returns, and the errors in those estimates swamp the benefit of optimising.
What I take from it. Conviction in selection, humility in sizing. Plenty of investors can show skill in choosing what to own. Far fewer can show skill in deciding how much of it to own. If you cannot demonstrate that skill, sizing every idea to how confident you feel about it is a risk, not an edge. That does not mean equal weighting everything. In a concentrated market like the ASX it has trade-offs of its own. It means the sizing decision deserves as much scrutiny as the stock decision.
London Business School research page · RePEc listing
3. Bonds have not always protected shares
Most investors who built portfolios in the 2000s and 2010s learned to expect bonds to rise when shares fell. That expectation sits underneath the classic balanced portfolio.
Antti Ilmanen’s 2003 study of the long-run relationship, published in the Journal of Portfolio Management, found that the correlation between stock and bond returns was positive through most of the 1900s. The negative stretches, in the early 1930s, the late 1950s and the period that began in the late 1990s, were the exceptions.
What I take from it. The period most of today’s portfolio models were calibrated on may have been the unusual one. If shares and bonds can fall together, a portfolio that relies on bonds as its only shock absorber is less diversified than it looks, and it is worth asking what else in the portfolio genuinely behaves differently.
AQR research page
4. The best managers are wrong about half the time
Stock Market Maestros, by Lee Freeman-Shor and Clare Flynn Levy, was published by Harriman House this year. The authors screened 73 award-winning fund managers using their actual holdings data, then profiled the 12 best as “maestros”. Even those 12 were right on only 49 per cent of their ideas in aggregate, and none had a win rate above 57 per cent.
The book measures two things. Hit rate is the share of positions that beat the benchmark. Payoff ratio is how much the average winner adds compared with how much the average loser costs. The lower your hit rate, the bigger your winners need to be just to break even. The formula is simple: the break-even payoff ratio is (1 − hit rate) ÷ hit rate.
| Group | Hit rate | Payoff ratio | Needed to break even |
|---|---|---|---|
| Median manager (Essentia benchmark pool) | 37% | ~100% | 170% |
| The 73 award-winning managers | 45% | 137% | 122% |
| The 12 maestros | 49% | 182% | 104% |
Hit rates and payoff ratios as published in Stock Market Maestros and presented by co-author Clare Flynn Levy of Essentia Analytics. Break-even column is my own calculation using (1 − h) ÷ h.
What I take from it. What separates the best is not being right more often. It is making a lot more when they are right than they lose when they are wrong. The book suggests four questions worth putting to any manager, and to yourself:
- What is your hit rate, and what is your payoff ratio?
- Show me a position you got badly wrong. What made you realise, and how long did it take you to act?
- What is your ongoing process for recognising when you are wrong and doing something about it?
- What have you changed about your process in the last three years, and why?
Publisher page · CFA Institute review
5. The money is usually lost after the buy
Essentia Analytics, the behavioural analytics firm co-founded by Clare Flynn Levy, studied 43 professional portfolios over 14 years, covering around 10,000 positions from the first purchase to the final sale. The findings were reported by Institutional Investor in November 2019.
Managers did generate alpha, well in excess of the fees they charged. But that alpha started strong and faded sharply as a position aged. The average manager held an idea for 257 business days. The top-performing quarter of managers held theirs for 133. Managers who avoided becoming over-attached to their positions could have added about 1.2 per cent a year at the portfolio level, against an average fee of 0.75 per cent.
What I take from it. Picking is the easy part. The expensive mistakes happen later: holding on too long, adding to positions without a reason, and letting ownership itself cloud judgement. That is why I write down, before investing, what has to go right, what would tell me I am wrong, and when I will review it, and then let the calendar rather than the mood of the market decide when that question gets asked.
Institutional Investor article
6. Geopolitical risk can be measured
“The world feels more dangerous” is a mood. Dario Caldara and Matteo Iacoviello, economists at the US Federal Reserve Board, turned it into data. Their Geopolitical Risk Index counts how often ten major newspapers report on adverse geopolitical events, scaled so that the 1985 to 2019 average sits at 100. The research was published in the American Economic Review in 2022, and the monthly data is freely available.
Working from the raw monthly series, 30.4 per cent of months since January 2022 have printed above 150, against 3.3 per cent between 1985 and 1999 and 1.2 per cent between 2015 and 2021. I set out the full case, and what it means for holding cash, in Volatility is the regime.
What I take from it. It is the frequency of extreme months, not the average, that has changed. A portfolio built for occasional shocks needs a plan, written in advance, for frequent ones.
Journal article · Index data and paper
What these studies are, and what they are not
Most of this research is about US markets and professional fund managers, and all of it describes the past. None of it tells you what a particular share, fund or market will do next, and none of it is a recommendation. What it does is change the questions worth asking: how a portfolio is sized, what it is really exposed to, and what would make you change your mind before the market forces the issue.
I will keep adding to this page as I work through new research. If you see a number in one of my posts, it should trace back to here or to the article it came from.