Most of us like guessing what will happen next.. in our life, in our neighbour’s life, in a celebrity’s life, in the world in general.
To learn from these predictions, I used to capture them in a Google Sheet and see if they went right or not.
Then, I learnt about the Brier Score from the book, Superforecasting: The Art and Science of Prediction. This is made around the idea that every forecast has a probability. And a good forecaster does not make yes/no prediction, but more probabilistic predictions.
For example, Haruki Murakami has been in line to win the Nobel Prize for literature. So, just before this year’s literature Nobel (which went to Kazuo Ishiguro, btw), I wanted to predict whether Murakami will win it. A more binary prediction would be – he will win or not win. But a more probabilistic prediction is – chances of winning are 20%.
Making this type of prediction makes you think more deeply about the problem.
Now, since Murakami did not win, I noted my Brier Score as : |0 – 0.2| = 0.2. If he had won, it would have been, |1 – 0.2| = 0.8. This score basically measures how far you are from the reality. And over time, you can learn better. This is an adaption of the Brier score. The actual, statistical definition is explained better here. However, this simple adaptation works well enough for me.
Here is a Google Sheet with the basic format:
You can find the sheet on this link and play around with it. Feel free to adapt it according to your needs.
Will probably share more insights about how to make good predictions, until then, give this one a spin.