Why I Stopped Predicting Anything

I used to end each quarterly note with a view about what was likely next. I stopped, for two reasons that are both mundane.

Reason one: the record

Nine right, eight wrong, six too vague to score, across five years.

That is chance, on the scorable half. Worse, the entries written with high confidence were wrong four times out of five, which means my sense of certainty is not merely useless but inversely informative.

Once you have that written down in your own handwriting, continuing to publish forecasts requires an explanation you cannot give.

Reason two: nothing depends on them

This is the more interesting one.

My contributions are scheduled. My rebalancing is mechanical. My exits are governed by written conditions about the asset rather than about the market.

Not one of those processes takes a forecast as an input. I was producing predictions as a by-product of thinking, and then treating them as though they mattered, when the actual strategy was indifferent to them.

Removing them changed nothing except the notes.

What I do instead

Describe conditions rather than outcomes. Funding rates are elevated. Long-held supply is growing. Depth has thinned. All observable, all stated without a consequence attached.

Conditions are useful because they are checkable and because they do not require me to be right about anything. The commentary that says “positioning is crowded” is doing work. The commentary that adds “therefore a decline is likely” is adding an unsupported claim to a supported one.

Note what would change my mind. More useful than noting what I expect, because it commits to a test rather than an outcome.

The objection

Someone will say that refusing to forecast is itself a forecast, or that allocating capital implies a view.

That is true in a weak sense. Holding an asset implies I expect it to be worth something. It does not imply a view about the next quarter, and the strategy is built specifically so that quarterly views are irrelevant.

What this cost

The notes are less interesting to read. A quarterly summary that describes conditions and declines to conclude is duller than one with a prediction in it.

That dullness is the point. The predictions were the entertaining part and they were also the part with a documented accuracy indistinguishable from chance.

What replaced the impulse

When I feel the pull to state a view, I write down the condition I am actually observing and what would have to change for it to matter.

Most of the time that exercise reveals I have an impression rather than an observation, which is information about me rather than about the market, and the standing order at the venue I buy through executes either way.