Variant Forecast (click here for main COVID forecasts)
The following forecasts show what variants will be most prevalent in England and the US in the next month. They are generated using a Multinomial Logistic Regression (MLR) model. If you're wondering why it looks like only a few variants are shown, I wanted to avoid displaying dozens of nearly identical variants, so I grouped them into several variant groups based on significant characteristics (similar growth rates and mutations). You can view more technical details about these groups here.
WARNING: MLR tends to significantly overestimate the growth advantage of new variants due to small sample size. In this case, the model predicts that the 24.4/XEC variant group will become dominant at the start of November. While I am confident that that this variant group will become dominant unless a significantly stronger variant appears, it may not become dominant until later November or even December due to the overestimation effect.