Skip to content

This one quick trick will allow you to become a star forecaster

Jonathan Falk points us to this wonderful post by Dario Perkins. It’s all worth a read, but, following Falk, I want to emphasize this beautiful piece of advice, which is #5 on their list of 10 items:

How to get attention: If you want to get famous for making big non-consensus calls, without the danger of looking like a muppet, you should adopt ‘the 40% rule’. Basically you can forecast whatever you want with a probability of 40%. Greece to quit the euro? Maybe! Trump to fire Powell and hire his daughter as the new Fed chair? Never say never! 40% means the odds will be greater than anyone else is saying, which is why your clients need to listen to your warning, but also that they shouldn’t be too surprised if, you know, the extreme event doesn’t actually happen.

I like it.

P.S. Justin Lahert gets priority here, as he wrote about “the 40% rule” back in 2010:

The 40% rule is a time-tried tactic employed by pundits making forecasts. If you say there is a 40% chance of something improbable happening and it does, you look great. And if it doesn’t, you never said the odds favored it.


    • David J. Littleboy says:

      Even worse, your statement of it is way clearer.

      But, man, you make Machiavelli look like a nice guy.

      Also, Falk wrote “Trump to fire Powell and hire his daughter as the new Fed chair?”

      Uh, this one’s unlikely. She doesn’t have the track record of having been wrong about everything she’s ever said that the right-wing (it’s not a bubble, inflation’s around the corner, etc.) economist types Trump surrounds himself with have. Mere incompetence isn’t enough in TrumpWorld.

  1. Chris Wilson says:

    hahaha, a nice little pundit arbitrage :) In all seriousness though, this makes me think about how most people seem to engage in meta-probability, by re-weighting most forecasts as though they had extra unaccounted for uncertainty. The net effect of this is to move forecasts in the middle ranges closer to 50%, and only to take seriously forecasts that are much closer to either end. Peters et al. (2020) show how this mechanism of re-weighting can readily account for some of the “biases” that behavioral econ likes to harp on about for instance:

    • jim says:

      “most people seem to engage in meta-probability”

      I’m not exactly sure what that is but IMO forecasters engage mostly in

      a)social acceptance probability forecasts: the forecast is measured to the likely social acceptability level shaded to the sign of the forecaster’s personal bent.

      b)totally nut forecasts: the forecaster is outside the mainstream, isn’t constrained by mainstream sensibilities and thus is free to forecast any crackpot number or event, like Trump’s 6% growth forecast, or the “goldbugs,” constantly forecasting that “fiat money” will soon collapse, trigger the inevitable return to the gold standard and of course the rise of gold to $xBillion/oz.

      These forecasts are typically amplified by the media by the same method as (a) – i.e., the press prints what is believable. (b) are all ignored until Weird Shit happens. In the light of disillusion and confusion caused among the press by Weird Shit happening, anything suddenly seems possible, so the press amplify (b) forecasts until Weird Shit stops happening, at which time social acceptance forecasting returns to media prominence.

      (a)(forecaster) and (a)(media) act as a self-limiting mechanism in that each provides feedback to the other so that nothing unusual is given a reasonable chance of happening and nothing too unusual is even mentioned, unless it’s far enough away that no one will be held responsible for saying it or printing it.

      • Chris Wilson says:

        jim, you are not wrong. I was thinking more about less polarized/controversial use of probability forecasts, e.g. how people interpret weather forecasting or something like that. Or, even, how scientists evaluate their own and other’s forecasts. For example, recently Andrew was talking about betting on Biden, referred to his model (at the Economist) and then downweighted Biden’s odds slightly to account for extra uncertainty that the model didn’t capture before discussing a hypothetical betting strategy…

        • jim says:

          Yeah, OK. I get you.

          Also the meaning of “pundit” is a bit different in different contexts: there are the full on “public pundits” (formerly known as “intellectuals”) who are visible to everyone; then there are the “sub-pundits”, visible within their profession and to the public pundits – sort of the pundit’s pundits.

          I guess when you think about it all of society can be described as a hierarchy of punditry: it’s pundits all the way down.

  2. Jonathan says:

    A version of take 100 balls, say 30 are red, some are green, which color do you want, the 30 or the ‘some’? And what is your bet which color I’ll pull from the bag?

  3. Not Trampis says:

    Unfortunately all too true.

    What I do find amazing at least here down under is how the media than forget about such absurd calls and only focus on the ‘accurate’ ones

  4. Alex C. says:

    I’m reminded of the crop circle in the movie Waiting for Guffman (1996). The UFO expert in the movie claims, “Once you go into that [crop] circle, the weather never changes. It is always 67 degrees with a 40% chance of rain.”

  5. Gur Huebrman says:

    If everybody adopts the superior 40% probability prediction, it will be the consensus.

  6. Roger H says:

    A couple of people have already mentioned weather forecasts. The traditional Terminal Aerodrome Forecast (TAF) format only uses probabilities of 30% and 40%:

Leave a Reply