Only 1 underdog won across the entirety of the NFL last week - the Giants beat the Colts, which we didn't pick - and what did we learn? That week 17 may well be an extension of the final week of the regular season, and that there may be no regression toward the mean in 2024. All season I have picked with that in mind, but the last two weeks have seen us go winless, the first time that has ever happened.

Some context:

This is... unusual. Our model is based on the assumption that around 1-in-3 games end in an underdog victory. This season it's close to 1-in-4, and more importantly, the distribution of those wins has been dramatically different. Rather than seeing underdog wins spread out across the season, with few 'down' weeks, we've seen more weeks with 3 or fewer underdog wins than we have seen since I started Pickwatch.

The key mistake made this year has been to assume that regression would happen, and picking on the basis that a glut of underdogs are due. Instead, as the season has drawn on, we have if anything, seen an unusual lack of upsets, culminating in the lone underdog winner in week 17. As you can see from the final column, the number of underdog winners between weeks 13-17 is also the lowest it has ever been, and compared to last season, has yielded 10 fewer winners.

I always own our losses, and the theory behind this column has never been perfection, rather percentages. We know the bar to profitability with underdogs is very low - 39% - which means if 33-36% of underdogs win, that is a very achievable goal. That said, as I've discussed in Discord, I will likely change my approach to start 2025's season, and focus very much on getting profit every week from a smaller number of picks, than picking the 'right' amount of underdogs.

Is something changing in the NFL? My instinct is no, but there has not been a season like this over the last decade, and we have to keep in mind that historic patterns are only a guide. It's unusual that we're going to finish the season with a losing record - it's the first since we started 'properly' in 2017 - but I am confident that underdogs are not going to continue along this path. If there's one thing our data does show us, it's that this will correct itself, and being in the right place for when it does, is the right way of thinking.

In the meantime, I'm totally up front that this season hasn't been the best at all. Rest assured I'm not just blaming the lack of upsets, and we're already working on ways to improve the quality of data based on what we've seen this season, as I suspect most people would prefer fewer high value picks than many lower value 'leans'.

Top picks

Our first ever double-goose-egg, as none of our top picks won.

All underdog picks

See above!