“Truth has predictive power”
Predictions are as hard as useful; in hindsight they are easy
This past week has been unusual in the markets — stocks, crypto, etc. Awareness of the past may help to predict the future, but more so awareness of the present.
There is no shortage of competing explanations of what happened, and predictions of what will happen. They may seem correct but still fail. They ignore significant aspects of the present. They are not true or not the whole truth, and the missing part has a stronger impact in the future. To explain in more detail:
- Some explanations and predictions are not true; they have never been. They may be misleading or seem true for some faulty reasoning, e.g. fallacies.
- Some are not true at this time, but they have been in the past, and they may be in the future. The prediction: “the sun will rise in 3 hours” is true only 3 hours before the sunrise. It is a matter of situational awareness.
- Some are true, but the opposite is also true, and quantitatively overrides the original. I like the example of two opposite relativity effects, causing time to run both faster and slower for GPS satellites. To know the net result, you have to quantify both, and subtract.
Other relevant aspects of predictions are:
- Refutable, specific: avoid confirmation bias. Quantifiable is best.
- Actionable: correct predictions are good, useful predictions are better.
- Second order predictions: often more useful, and refutable; requiring the first order predictions to be specific enough for the second order.
To improve at your predictive capabilities, as usual: study and practice. Any time is good to study, this is a particularly good time to practice. At this time, study less, practice more.
Needless to say, you do not need to gamble to practice your predictive skills. Just make predictions. If you tend to forget what you predicted or why and how, write all of it down, it is free.
On a side note, for some artificial intelligence (AI) researchers “intelligence” was a synonym for “predictive capabilities”. Unsurprisingly, that stage of AI seems over. Nevertheless, “predicting” is a good exercise to surface mistakes, improve your competence, and decrease your confidence, i.e. fight Dunning-Kruger. Do it.
Cross-posted from the Sigmoid newsletter