I recently wrote about the issue of building models that cause us to focus on precision sometimes at the expense of accuracy. In building models to predict or forecast future outcomes, it’s important to focus on accuracy over precision — particularly when you first start making forecasts.

However, another key area to focus on is how your forecasts — and the models you used to make them — change over time. As you gain a deeper understanding of the area you are forecasting, your models should gain complexity over time. In particular, as you more precisely define ranges for your assumptions, you should be able to add other factors into the precision of the model.

The most important thing is to view forecasting as a continuing process. It’s not a one-time activity that we complete and then move forward from. It’s an ongoing, iterative process where our accuracy and precision both continue to improve over time.

I recently read the book Superforecasting: The Art and Science of Prediction based on a recommendation from Upstart’s Head of Growth, Jungwon Byun and it has really helped me think through this issue and get better about making forecasts — and holding myself accountable for accurately assessing them and continually improving them over time. If your work requires you to make predictions of forecasts, I strongly encourage you to read it as well — one of the best books I’ve read recently!