Ferg's Finds

This is a short weekly email that covers a few things I’ve found interesting during the week.

A bit of housekeeping: I've paused paid subscriptions on my Substack while I restructure a few things in the background. It should only be for two weeks, but for now, no one new can subscribe to paid tiers. For paying subscribers, your current subscription will be extended by the period I've paused payments for (~14 days).

Article(s)

Adam Robinson on Understanding

In 1974, Paul Slovic — a world-class psychologist, and a peer of Nobel laureate Daniel Kahneman — decided to evaluate the effect of information on decision-making. This study should be taught at every business school in the country. Slovic gathered eight professional horse handicappers and announced, “I want to see how well you predict the winners of horse races.” Now, these handicappers were all seasoned professionals who made their livings solely on their gambling skills.

Slovic told them the test would consist of predicting 40 horse races in four consecutive rounds. In the first round, each gambler would be given the five pieces of information he wanted on each horse, which would vary from handicapper to handicapper. One handicapper might want the years of experience the jockey had as one of his top five variables, while another might not care about that at all but want the fastest speed any given horse had achieved in the past year, or whatever.

Finally, in addition to asking the handicappers to predict the winner of each race, he asked each one also to state how confident he was in his prediction. Now, as it turns out, there were an average of ten horses in each race, so we would expect by blind chance — random guessing — each handicapper would be right 10 percent of the time, and that their confidence with a blind guess to be 10 percent.

So in round one, with just five pieces of information, the handicappers were 17 percent accurate, which is pretty good, 70 percent better than the 10 percent chance they started with when given zero pieces of information. And interestingly, their confidence was 19 percent — almost exactly as confident as they should have been. They were 17 percent accurate and 19 percent confident in their predictions.

In round two, they were given ten pieces of information. In round three, 20 pieces of information. And in the fourth and final round, 40 pieces of information. That’s a whole lot more than the five pieces of information they started with. Surprisingly, their accuracy had flatlined at 17 percent; they were no more accurate with the additional 35 pieces of information. Unfortunately, their confidence nearly doubled — to 34 percent! So the additional information made them no more accurate but a whole lot more confident. Which would have led them to increase the size of their bets and lose money as a result.

Podcast/Video(s)

Review of: Mastering the Capital Cycle | Capital Returns by Edward Chancellor

It’s one of those insights I come back to over and over again.

Quote(s)

“The highest form of wealth is the ability to wake up every morning and say, ‘I can do whatever I want today.’ People want to become wealthier to make them happier. Happiness is a complicated subject because everyone’s different. But if there’s a common denominator in happiness—a universal fuel of joy—it’s that people want to control their lives. The ability to do what you want, when you want, with who you want, for as long as you want, is priceless. It is the highest dividend money pays.”

- Morgan Housel

Tweet/notes

This was an interesting paper from Apple.

Apple tested today's "reasoning" AIs like Claude + DeepSeek which look smart—but when complexity rises, they collapse. Not fail gracefully. Collapse completely.

They found LLMs don't scale reasoning like humans do. They think MORE up to a point… Then they GIVE UP early, even when they have plenty of compute left.

Charts

Tin is next in line to run, with yet another drawdown in stocks this week.

Something I'm Pondering

I’m pondering how these two decisions have a large impact on investing performance.

One is position sizing, which I've always thought of as a balancing act between risk minimisation and regret minimisation.

Risk minimisation

Next is keeping the position size small so that no single position can harm the portfolio, i.e., 1% per position.

For the maximum position size, it pays to remember math makes it easy for large positions to damage portfolios.

A 30% loss requires a 43% gain to break even.

A 50% loss requires a 100% gain to break even.

An 80% loss requires a 400% gain to break even.

A 90% loss requires a 900% gain to break even.

The larger the position, the more exposed you are to this.

Regret Minimisation

The other side of this coin is the opportunity cost of being too safe (which rarely gets discussed).

Regret minimisation is ensuring positions are big enough to make a difference to the portfolio.

Maximising gains of the winner is just as important as minimising losses.

1% position becomes a 10x = 10% gain.

5% position becomes a 10x = 50% gain.

10% position becomes a 10x = 100% gain.

Sizing is achieved by scaling up positions over time rather than adding large positions all at once.

Positions should only represent a substantial proportion of a portfolio as a result of appreciation.

Take the time to read the 'Art of Execution’ by Lee Freeman-Shor as he really rams the below point home.

“The most successful investors, those who made the most money,

All had one thing in common: the presence of a couple of big winners in their portfolios. Any approach that does not embrace the possibility of winning big is doomed.”

For me, the sweet spot is position size starting at 3%, with the aim to increase to 5% over time, ideally taking advantage of volatility.

The second decision is how much cash to hold.

This is a battle between opportunity cost (holding lots of cash, which could be earning a return) and optionality, which holding cash gives you to take advantage of volatility in the market.

Currently, I sit around 5% in cash, as it's enough to take advantage of opportunities should they arise, but not so much that I feel at risk of great setups getting away from me.

I loved this insight from Ian Cassel on how he approaches holding cash.

I don't believe in holding a large cash position.

In up markets everyone tries to become long-term investors. In down markets everyone turns into macro strategists, geopolitical experts, and market timers.

The larger your cash position becomes the more you become a market timer instead of a stock picker. You are trying to time a drawdown.

Peter Lynch famously stated, "Far more money has been lost by investors in preparing for corrections, or anticipating corrections, than has been lost in the corrections themselves".

When you are sitting on a large cash position it's easy to become too bearish. I witnessed this during the Covid drawdown of 2020.

Too large a cash position clouds your judgement. You wake up hoping the markets crash, and your negativity paralyzes your common sense. Instead of timing a bottom you should be slowly adding to the businesses you love.

The amount of cash you hold is a function of your temperament and strategy. I like to only hold enough cash to buy half of a new position and then be forced to sell an existing position to buy the other half. It forces me to sell my weakest convictions to add new convictions.

The ability to deploy even a small amount of capital during a market drawdown can be a life saver. The ability to at least feel like you took advantage of lower prices even in a small way can’t be overstated. I hold enough cash to safeguard my emotions.

When you have a little cash, you have confidence. When you don’t have cash, you are just like everybody else.

I hope you’ve all been having a great week.

Cheers,

Ferg

P.S. We have been having a great time in Tuscany for a friend's wedding.

In case you missed the first one…

A bit of housekeeping: I've paused paid subscriptions on my Substack while I restructure a few things in the background. It should only be for two weeks, but for now, no one new can subscribe to paid tiers. For paying subscribers, your current subscription will be extended by the period I've paused payments for (~14 days).