1. Who are you?
2. Why are you posting about COVID-19?
- It’s first and foremost a way of organising my own thoughts.
- Sharing thoughts with friends has been a great way to solicit input and thereby improve my mental models
- More generally, I try to cut through the noise by identifying and sharing the best data sources, analysis, articles, academic papers, and frameworks.
- I’m biased towards quantitative approaches and clear frameworks for thinking about COVID-19.
3. What are the origins of the project?
- This project started on 28 February when a friend and I started trading updates and observations about COVID-19.
- I noticed that a number of professional epidemiologists were saying they had created models, but no one would share the output. One possible explanation was that the models were predicting a severe epidemic and they didn’t want to cause panic.
- So I built my own (extremely simplistic, exponential least-squares) model to forecast confirmed cases outside of mainland China, starting on 3 March, and shared it with a few friends. The model performed surprisingly well. It continues to have the right shape, but typically under-forecasts cumulative daily cases by 15-20% (likely because as testing ramps the degree of underreporting may be decreasing).
- More and more friends requested the daily update to the model.
- When the list got above 25 people I moved to MailChimp to simplify distribution and improve formatting.
- I’ve largely stopped maintaining the model for three reasons. It was taking too much time to update, particularly given nearly daily data quality issues (not to mention Excel collapsing under the weight); the model would soon have exhausted its useful life as it was only ever designed to forecast the early exponential stages of the epidemic; and professional epidemiologists starting sharing their forecasts and methodologies broadly.
- In late March I migrated to this blog (with associated newsletter) as I realised that many of my posts only made sense with reference to earlier posts.
4. You’re not a medical professional, epidemiologist, public health expert, doctor, nor professional investor. Are you qualified to give advice?
- Definitely not. No one should take action based on my comments here.
5. Why “Subjective Probability?”
- I aspire to be a good Bayesian in prediction and decision-making. “Subjective probability” has a technical sense in this domain.
- While I’m a frequentist about some things, most of the decision-making that interests me is under conditions of significant uncertainty.
- I also like the fact that, in everyday use, the title emphasises (a) that the blog expresses just one person’s point of view and should be approached with healthy scepticism; and (b) that at most I make claims for what I believe is probable, not certain.
5. How can I contact you?
Please use the contact form on the left of this page.
Subjective Probability 