Better Indicator of Good/Bad Teams - Points Differential

[ghz in pittsburgh]

Here is an article talking about baseball run differential as a tool being a true indicator of a team. http://www.nytimes.com/2015/08/11/sp...-surprise.html Supposedly it works for all sports.

In the past, I never bought into the argument of winning big. Heck, score one more run, one more point, one more goal is always a more exciting win. But coming to think of it, the ability to score big should be an indicator of the overall team ability. All sports have elements of luck, especially coming down to each play. But the collection of all plays is no luck. It is a true indicator of a team. I guess that is the basis of math statistics. In the NFL world, we know a game decided by one score of less can certainly be influenced by a lucky throw or bounce. The Giant's 2012 superbowl victory coming to mind.

Here is a website of NFL points differential for all those years. http://www.sportingcharts.com/nfl/st...-against/2011/

A team with a good QB is usually associated with an average of high 20s points a game. And the Seahawks is sick with 15.9 points against per game, far lower than any other team (my guess is great defense and low turnovers).

If I'm predicting the Bills this season, I have reasons to believe that the Bills should be able to build on their 2014 21.4 points per game offensive output even with their current QB crop. O-Line, RB positions are a sure upgrade; TE, WR in theory as well. However, I don't know if they can get into the projected Miami Dolphin's Tannerhill average 25 points per game territory. I'd guess around 23 points for the Bills.

If Rex and company can truely improve the D from 2014 level, 18.1, to the Chiefs' level of 17, then we are getting close to the 100 points differential which historically is a very reliable indicator of a top 6 team in NFL.

What is your opinion on the likelyhood where the Bills offense can improve its average from 21 to 23 or Rex's defense can approach Seahawks mark in 2014?

[OpIv37]

Re: Better Indicator of Good/Bad Teams - Points Differential

I have a hard time buying into this. When games get out of hand, both the winning and losing team play guys who normally wouldn't see the field/court/ice in a close game. It skews the results.

[ghz in pittsburgh]

Re: Better Indicator of Good/Bad Teams - Points Differential

I understand your point completely. It is the same reaction I had initially. But as statistics/big data people say, look at the big picture, not just numbers in microscope. When a team is ahead by 20 points late in 3rd quarter, they may not press all that hard on offense but defense is sure pinning ears going after them - if they are successful, it is an indicator of a good team.

Likewise in a baseball game, when you scores 8 runs when it still matters is good indicator. There are some skewed number, but the trend is there.

Look at the NFL numbers, I can't argue that points differential over 100 is a better indicator of top 6 teams year after year.

[IlluminatusUIUC]

Re: Better Indicator of Good/Bad Teams - Points Differential

You are touching on the Pythagorean expectation model (https://en.wikipedia.org/wiki/Pythagorean_expectation)

Focusing on points rather then wins does help predict teams who are true outliers, but it's not perfect. Bruce Arians' teams have defied it somehow.

[trapezeus]

Re: Better Indicator of Good/Bad Teams - Points Differential

bills points for was 343, average for group was 361. So we score less than most team. though not an unusual outlier. We scored more than 13 teams. It puts us in the middle pack of teams.

Our defense, on the other hand, had 289 against, where the average was 361. Only 3 teams were better than the bills.

I think you have to look at this and say, "being a middle of the pack offense, is not good enough." we were that last year by this metric..

this year, we don't have to be light years ahead from last year offensively. we just have to be crafty enough that where marrone played for the FG's early in a game, we have a little identity and get the 3 yards and keep a drive alive for a td. Maybe not all the time, but every other game? that alone could be enough to put us in the top 10 of point differential. i think all but 3 playoff teams were top 10 (cards, panthers and bengals were not). Two of those teams were wildcard teams and one was just in a bad division.

One could argue that we had a punt return, a kickoff and maybe a defensive touchdown to help the points for. on an adjusted basis, maybe we fall down a little further. So we will have to pick up the slack, but i don't see a reason to think our ST will fall off or that our defense isn't aggressive enough to keep contributing to the scoreboard.

This kind of analysis gives the coaching staff every reason to pick the guy who is the most stable and who the team seems to trust the most. It makes the playoffs a real possibility, but long term, it may result in a coaching staff settling for a QB to just make the playoffs with no plan to go any further.

[GingerP]

Re: Better Indicator of Good/Bad Teams - Points Differential

In the NFL, Turnover differential is just as important. Point and turnover differential together have a high correlation to winning.

[IlluminatusUIUC]

Re: Better Indicator of Good/Bad Teams - Points Differential

In the NFL, Turnover differential is just as important. Point and turnover differential together have a high correlation to winning.

This is true, but turnover differential fluctuates so much that it's not very predictive. For example, Seattle's had the top scoring D for 3 years but their takeways have been 31, 39, 24. You can coach guys to hold the ball and go for strips/picks as much as you want, but tip drills and fumbles once they are loose are basically random.

[Mike]

Re: Better Indicator of Good/Bad Teams - Points Differential

Problem with using stats like this (ones that can vary wildly) is that a few great games in a 16 game season can really alter the end result. Unlike the NFL, MLB teams play 162 games and anomalies even out.