Sports predictions can often go awry. It’s been that way ever thus. In any given sports league in any given season, there are always teams that do badly, and teams that we expected to do badly that end up doing well. The same goes for players. Yet, every season we are still greeted by predictions laid out through statistical reasoning, algorithms, and, more recently, AI.
As with other technologies and sciences, you can see that using analytics to predict sports has improved over time. That’s broadly true, and it’s the reason why sports teams will have data departments given just as much precedence as the coaching setup. Yet, it is not – and probably never will be – a perfect science.
If we look at the 2025 regular season, for instance, you can see just how wide off the mark a lot of the data-driven predictions were. The favorites in the Super Bowl LX betting odds in early September (just before the season began) were the Baltimore Ravens, Buffalo Bills, Philadelphia Eagles, Kansas City Chiefs, Detroit Lions and Washington Commanders. Only two of those teams made the Playoffs, and the other two, the Eagles and Bills, were eliminated early in the postseason.
The Super Bowl betting favorite does not always win
FiveThirtyEight, the polling data and analytics firm, published an article on this phenomenon a few years back, noting that the betting favorite at the start of the season rarely goes on to win the Super Bowl. It’s an interesting phenomenon, and it feels almost contradictory, but it remains true nonetheless.
Nonetheless, that doesn’t mean that sports analytics are useless for predictions – far from it. Moreover, we can also argue that the 2025 season was something of an outlier. Rarely has there been a season in history when so many favorites were flattened early on in the campaign, but it does provide an example of why we should take such analytically driven predictions with a pinch of salt.
<iframe width=”560″ height=”315″ src=”https://www.youtube.com/embed/n29etL4LtiA?si=ZGiE6HaC3CbtTbS0″ title=”YouTube video player” frameborder=”0″ allow=”accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share” referrerpolicy=”strict-origin-when-cross-origin” allowfullscreen></iframe>
Predictions should be taken lightly
In the simplest terms, analytics and sports predictions provide a kind of prognosis based on factors without external influences. This is what the data says does not take into consideration, for instance, star players being injured, players falling out with coaches, losses in team morale, and a myriad of other factors that can impact the outcome of a game or an entire season.
And that, as such, is the rub: We can look at other sports predictions, such as the 2025/26 NBA season and see that the data-driven analysis was spot on. It was almost universally recognized that the Oklahoma City Thunder would be the dominant team in the NBA season, and that has proven to be the case so far. It does not mean they will definitely win the NBA Championship in June, but the predictions have proven correct so far.
But the 2025 NFL season has shown that analytics can only tell a small part of the story. Sports are human endeavors, and humans make errors – consistently – and can confound forecasts of failure. It’s one of the reasons why we love sports, because you cannot compress it into numbers alone.
