TWO HOURS WITH DAVE DREDGE

I have just spent two hours with Dave Dredge and Niels Kaastrup-Larsen, and I am still thinking about it.

Here is the odd thing Niels noticed, and it is what the whole conversation grew out of. Dave and I have been on his show separately for years. He comes from options, convexity and tail protection. I come from trend following and quantitative research. Different tools, different vocabulary, different everything. And yet, listening back, Niels said the two of us sound like echoes of each other.

He went looking for why. What he found was that we both see markets as complex adaptive systems rather than as tidy statistical distributions to be sampled and estimated.

Almost everything else follows from that single commitment.

If you have not come across Dave before, you should know what you are in for. He has a rare ability to take an idea spanning physics, mathematics, behavioural finance and portfolio construction and compress it into a single sentence that somehow makes everything clearer than it was a moment ago. His memos have become quietly famous for these. People call them Dredgeisms. They are memorable because they are true.

What actually moves a price

Nothing moves a price except a transaction.

Which makes a market an ecosystem. Some participants amplify moves. Some fade them. Some are following a rule and never look at price at all.

I built a market populated only by random traders, and out came the textbook world. Bell curve, no memory, no volatility clustering. Exactly as advertised. Then I added participants who amplify, and once they got past roughly a quarter of the population, every signature of a real market arrived at once. Fat tails. Memory. Volatility clustering. Not gradually, the way you would expect. All together, like a switch being thrown.

The bell curve is not a rough approximation of markets. It is an accurate description of a market where nobody agrees with anybody else.

Dave’s world, in Dave’s words

Half the conversation was really an excuse to get the Dredgeisms on tape.

“Positioning is the only thing that matters.”

Everyone wants a reason why the market moved. Dave wants to know who owns what, and with how much borrowed money.

“Infinite paths inside a finite space, chasing an unattainable strange attractor.”

Ed Lorenz, applied to a portfolio. You cannot forecast the shape of the race. You can build a car that survives it.

“Over a long enough time horizon, all levered ETFs approach zero.”

Not an opinion. Arithmetic.

“When you’re positively convex, more risk is less risk.”

Read that twice. It reverses everything you were taught, and it is correct.

“Most people do a good job of managing the risk they measure. They do a very bad job of managing the risk they assume.”

Which is usually correlation. And correlation is exactly what stops behaving at the worst possible moment.

“Risk is everything that isn’t in your back test.”

“Risk is about accountability.”

Dave closes every risk conference presentation with that one. No model, no regulation and no formula substitutes for knowing that if you crash, you are out of the race.

And then my favourite, because it explains the whole of market history in a single image.

The forest fire.

Everyone blames the lightning. But lightning strikes constantly. What decides whether you get a small burn or a catastrophe is how much dry brush was allowed to build up between the trees while nothing much was happening. And here is the twist that took me a while to fully appreciate. Dave says he doesn’t need to know which trees you own. He only needs to know where the dry brush is. It turns out that is wherever the insurance is cheapest, because the models pricing that insurance are looking backwards at a period in which nothing burned.

Why the calm is the dangerous part

This is where Dave and I arrive at the same place from opposite directions, and it was my favourite part of the conversation.

Calm is not the absence of energy. It is a rubber band being stretched.

Volatility targeting divides a risk target by recent volatility, so when volatility falls, position sizes rise automatically. Nobody decides to take more risk. The arithmetic does it for them. Then behaviour piles on top, because in a quiet market people reach for leverage to make a living.

Japanese government bonds in 2018 had trailing volatility of 0.9 percent, which authorised sixteen times leverage. The bond then moved six tenths of one percent and took nine percent of the capital.

The market barely moved. The silence did the damage.

Do not confuse calm with safety.

The number that cannot see what matters

October 2008 was the single best month in forty years of trend following history I looked at. It contributed 3.4 percent of every dollar the portfolio ever made.

Its effect on the Sharpe ratio was to move it from 0.914 to 0.931.

Two cents on the dollar for the best month in four decades.

Dave has a beautiful way of showing this. He draws a football pitch, lays a bell curve over it, and points out that the average tells you where the ball spends most of its time. But football matches are not decided in the middle of the pitch. They are decided in the penalty boxes.

So his team built a model and ran it through the World Cup to see whether the story held up. It did.

Two workshops, one shape

Here is where people expect Dave and me to disagree, and we do not.

A long position with a stop is a synthetic call. A short position with a stop is a synthetic put. Trend followers get the shape of an option without paying an explicit premium for it.

But we do pay a premium, and I want to be honest about that, because our industry often is not. Ours does not appear as a line item. It arrives as whipsaws, as stopped-out trades, as growth we never got. Dave’s premium is visible on the statement every month. Ours is invisible, which makes it easier to carry and harder to govern.

Same payoff shape in the end. Two entirely different workshops.

The one real difference is what each of us survives. Dave buys a guarantee, so a gap costs him nothing. We need the price to travel through the stop, so discontinuity is our failure mode, and we cover it with small bets and wide diversification instead.

He is protected against discontinuity. We are protected across duration.

Which is why the honest conclusion is not that one of us is right. It is that these things belong together.

What to do about it

Dave’s answer is to change the incentive structure, then add convexity. You cannot optimise for an unknown future. You can put brakes on the car and then go and learn how to drive it fast.

Mine is four things, none of which require you to forecast anything.

Cap leverage in absolute terms, which works not by measuring risk better but by severing the automatic link between a quiet market and a large position.

Size every market to the same risk rather than tilting toward whichever edge currently looks strongest, because the differences between edges are smaller than the noise involved in measuring them.

Grow by widening the book rather than scaling the bets. Another cable on the bridge, not a heavier load on the one you already have.

And enter small, then let the market build the position out of open profit. Let the market do the lifting rather than trying to impose your will on it.

The bit that stayed with me

Richard Feynman once said imagine how hard physics would be if electrons had feelings.

That is exactly what a market is. Every participant is watching, adapting and responding to a system that is simultaneously responding to them. It is not a distribution waiting to be estimated. It is a process being written as we speak.

The future is not hidden. It has not been written yet.

Which means the job was never to predict it. The job is to build something that is still standing whatever gets written next.