DVOA and Postseason odds tracking

[Bearcat]

Originally Posted by TwistedChief

So let me get this straight: if Sneed hadn't punched the ball out from Flowers at the goal line and we lost, then DVOA would be a success? But because we won the game, it's a failure?

These are predictive statistics that are a single metric by which to judge like anything else. I think some of you don't understand that if a team has a 60% chance to win, that team still loses 40% of the time. And in a sample size of 1, anything can happen, and it does nothing to necessarily invalidate the model.

DVOA attempts to level the playing field. Would it be better to just simply look at wins and losses? Of course not. And that's why we care about strength of schedule. But are wins and losses combined with strength of schedule the best predictor? Not necessarily, because you may have played a team at the beginning of the season and beaten them when they were playing poorly versus later when they're on a 6-game winning streak. DVOA accounts for that and goes deeper based on how both sides of the ball are playing.

That Ravens team last year was phenomenal. Just because the buzzsaw of Mahomes and Spags won out doesn't mean these don't add something important to the discourse.

Yep, and I misstated "not a crazy strong predictor"... it's probably a very strong predictor in terms of being in the top 5 of DVOA and making it far in the playoffs, versus being ranked 20th and making the SB.

However, the difference between being 1st and 2nd in DVOA isn't some huge difference and the numbers are saying exactly that... literally a 2% difference in the chance to make or win these games.

And people can talk other variables... injuries and so forth, but given the predictions are literally just looking at past results, all of that's already baked in.

The Ravens losing doesn't make it a failure, it just ever so slightly moves the needle for future predictions.