Super Bowl LVII is over and it was one for the history books. An offensive shoot-out between the Philadelphia Eagles and Kansas City Chiefs made for a dramatic football game that culminated in a narrow victory for the Chiefs — though one that was a source of controversy thanks to a hotly debated holding call. Considering that the game’s pre-show panelists almost unanimously picked the Eagles to win, it’s safe to say the result was a little unpredictable.

And no one predicted it worse than Madden NFL 23.

Every year, EA simulates the Super Bowl match-up about a week before it happens. Sometimes its results can be close to getting it right. Other times, like last year’s entirely erroneous prediction which had the Bengals beating the Rams, it’s entirely removed from reality. The latter went down with Super Bowl LVII.

This year, Madden NFL 23 predicted that the Eagles would win 31-17. To the game’s credit, it did seem like that was a possibility for a moment. Heading into halftime, the score was 24-14 with the Eagles leading (the Chiefs would have had 17 had they not missed a field goal attempt in the first half). It was realistic to think that the defense on both sides would tighten up in the second half, holding the Eagles to one more touchdown and getting them to 31.

Madden 23 | Official Super Bowl LVII Prediction (feat. Chad Johnson)

That wasn’t the case. The Chiefs’ offense exploded in the second half, running up the score as the Eagles’ Jalen Hurts struggled to find the end zone. With two minutes left in the game, the score was locked at 35-35. A bit of sneaky metagaming from the Chiefs resulted in a last-second field goal, ending it at 38-35. Madden once again picked the wrong winner, but it also blew its prediction by simulating a one-sided blowout for Philadelphia.

Part of the problem is that the game managed to both under- and overestimate Chiefs quarterback Patrick Mahomes. While EA only released a few stats and plays in its simulation video, it did predict that Mahomes would throw an interception to C.J. Gardner-Johnson. In reality, the Chiefs didn’t turn over the ball the entire night (the Eagles did, though). Otherwise, it predicted Mahomes would only have one touchdown during the game; he threw three in the real game.

There is one thing the simulation got vaguely right when it came to Mahomes. It predicted a below-average game for the quarterback, guessing he’d have 249 passing yards — well below his season average. Mahomes did indeed underperform in that regard, but even more drastically. He walked away from the game with 182 passing yards with 21 completions out of 27 (Madden had him at 29 for 39).

A Chiefs and Eagles player run down the field in Madden NFL 23. — Image used with permission by copyright holder

It was a little more accurate when it came to Jalen Hurts, predicting the Eagles star would earn 282 passing yards. He actually ended up with 304. It also nailed that Hurts would have a historic game when it came to rushing yards, though it was a little too overenthusiastic in that regard. The simulation had the quarterback with a whopping 88 rushing yards, which would have shattered a Super Bowl record. While Hurts didn’t reach those heights, he did in fact break the record for Super Bowl rushing yards with 66. Though perhaps no simulation could have predicted he’d get three rushing touchdowns on his own (Madden guessed he’d get one).

At the very least, it did get one player’s stats pretty close. The simulation predicted a good game for Eagles receiver A.J. Brown, with 114 yards and a touchdown to his name. Brown did indeed score one touchdown and hauled in 96 yards. The game had him catching eight passes, which wasn’t far off from the real number: six.

Overall, this year’s game was another loss for Madden’s annual simulation. Despite a stat or two lining up, you’d have to squint to see a link between the video game and reality. Perhaps the 2024 edition of the game will need to buff the Chiefs — or at least make the referee AI a bit worse.

Madden NFL 23’s Super Bowl LVII prediction couldn’t have been more wrong

Editors' Recommendations