Tuesday, December 21, 2010

"BigWin%" Looks At the Coming Playoffs

by Jim Glass

First, a bit of explanation of why anyone might care about the charts below.

A little while back I looked at the playoff performance of the 100 teams over the last 15 years with season records of 11-5 or better. The nine teams with the best records in close games (decided by 10 or fewer points) during the regular season, winning nine or more, in the playoffs went only 8-9. That may not be surprising to those who believe close games are determined mostly by luck. But the seven teams with the worst losing records in close games dominated in the playoffs, going 14-4 and winning three Super Bowls. That was a surprise.

Yet on reflection it shouldn't have been -- nothing else seems possible, given a key fact reported on this site: half of all NFL games are decided by luck.

Say a 10-6 record is needed to make the playoffs and two teams do it. In each's 16-game schedule eight games are determined by luck and eight by merit. Team A is a bit lucky and two of its "luck" games switch to wins, so it goes 6-2 in them (instead of 4-4). In its "merit" games it need go only 4-4, 50%, for its 10-6. But Team B is a bit unlucky and two of its "luck" games switch to losses, so it goes 2-6 in them. It has to be powerful enough to go 8-0 in its "merit" games, 100%, to get to 10-6.

Head-to-head, Team B should score 75% against Team A in spite of their equal 10-6 records, because on the merits it is a 100-0 team playing a 50-50 team. (In four games versus A it should go 2-0 on merit and 1-1 on luck.)

And this points to something else: As Team B, much the better team, did worse in close games but better in one-sided ones, it is success not in close games but in one-sided games (those determined "on merit") that predicts success in the playoffs. Among teams with equal records, those that win more one-sided games and lose more close games will do better in the post season.

The 15-year playoff record confirms this as very true. Within every "win cohort" (11-5, 12-4, 13-3, etc.) teams that won the most one-sided games and lost the most close games did best. And overall...

* The top 37 teams by regular season W-L pct had records of 13-3 or better (total: 497-95, 84%), in the playoffs they produced a record of 43-31, a 58% winning percentage (with six Super Bowl winners, 16% of the 37 teams).

* The top 30 teams by Pythagorean expectation won the same 43 playoff games and lost 23, 65% (with seven Super Bowl winners, 23% of the 30).

* Only the top 24 teams by "BigWin%" (counting close games as ties) had the same 43 playoff wins against only 15 losses, a 74% winning percentage (with nine Super Bowl winners, 38% of the 24).

* It took the top 45 teams by "close wins" to win 43 playoff games, losing 42, only a 50.6% winning percentage (with three Super Bowl winners, 7% of the 4).

For more data see the prior post. But that's all looking backward. We're interested in the future, and in picking this year's playoff game winners.

Let's see who the "big winners" are behind the regular W-L records of this season, as the 2010 playoffs approach.

The 2010 BigWin% Index

Below is a table showing Big Win Percentage, or BigW%, for all teams this season after 14 games played. It is equal to the number of 10+ point wins, minus 10+ point losses, with close games counted as ties at 0.5, all divided by total games. BigW% for each team is calculated using a strength-of-schedule adjustment based on opponents' BigW%.

A few observations:

* New England is #1 by BigW%, and also #2 by net close wins with 6 (#1 by close win percentage, 6-0, 100%), and has done it all against the fourth-toughest schedule in the league. An impressive combination! (It doesn't hurt to be both very good and lucky too.)

* The "Team B" of 2010 is San Diego. If the Chargers make the playoffs and you see they are a big underdog because of their poorish W-L record, take the points. "If ..." (This isn't the only system to rank them this high.) And the NFC's "Team B" is Green Bay, the top team in the conference by BigW% but also probably not going to make the playoffs.

* Atlanta, the team everyone talks about here, is #12 by BigW% in spite of being tied for the best W-L record in the league. The Falcons are a nice +3 net on Big Wins, pretty good, but owe their impressive 12-2 to having the most net close wins, +7 (8-1 in close games), and they've also played a weaker than average schedule. However with the NFC being much the weaker conference this year, and the Packers likely missing the playoffs, they could still be a legit favorite on the NFC side of the playoff bracket.

* This is a big AFC year. Fully seven of the top eight teams by BigW% are in the AFC. The AFC East is +10 net against the rest of the NFL and has the top BigW% (.582) by a lot, while playing the toughest schedule .558). Miami has played the toughest schedule of any team in the league (.571). The AFC North is #2 by BigW% (.562) – these two divisions played each other this year.

* The weakest division of course is the AFC West, with a BigW% of only .302(!) and playing the weakest schedule, .419. Of its two teams tied for first place and the division's playoff spot, St Louis is ranked 27th by BigW% and has played the easiest schedule in the league (.400), while Seattle is 30th by BigW%. New Orleans in particular has feasted on playing this division, getting its 10-4 regular record against opposition with only a .402 BigW%, the second easiest schedule in the league, making its own BigW% barely over .500, 18th overall.

* The teams that have suffered the most from close game losses are Dallas, Detroit, Cleveland and Cincinnati, all at net -5. If BigW% means anything they are all better than their regular W-L records look.

Below is a breakdown of "big" and "close" W-L for each team, listed in order of BigW%. Given is the regular W-L record for each team, followed by its number of "big wins", "close games" and "big losses", while the last column gives the team's net in close games. (For instance Atlanta's close game column is 9, and its net is 7, indicating it went 8-1.)


James said...

I think you meant the weakest division is the NFC West.

Unknown said...

I guess Atlanta is 8-6 instead of 8.5-5.5 because of the strength of schedule adjustment?

I don't think the NFC is the "much weaker conference" this year. I think a 1 seed facing a field of the Eagles, Bears, Saints, and Giants or Packers (NFC West team excluded because they'll be gone) has at least as tough a road as one facing the injured Steelers, injured Colts or Jaguars, Chiefs, Jets, and Ravens. You may think differently, but it's certainly not as obvious as you make it out to be one way or another.

By building strength of schedule into your calculations based on big win % you are basically double counting the advantage your system gives to AFC teams relative to just about any other evaluations of the teams.

Unknown said...

So Atlanta should be tied for fourth in your rankings if you don't factor in strength of schedule? Is that why you put in the strength of schedule adjustment?

James said...

You factor in strength of schedule because it's easy to win big against bad teams and hard against good teams. That seems straightforward to me.

Jim Glass said...

To clarify the strength of schedule adjustment, it puts all teams against the same equivalent league-average schedule of opposition.

That is, teams actually play very uneven schedules, so their BigW-L ratios in some cases aren't comparable. The SoS adjustment equalizes the opposition level so NE's BigW% is figured against the same strength schedule as Seattle's.

This is no different than any other rating system that makes a SoS adjustment to rate all teams on a uniform scale.

If you want to see the actual raw BigW-L count it is in the second table. E.g. NE's raw BW score is 9-5. But as NE has played one of the toughest schedules in the league, against a somewhat easier league-average schedule that projects to 9.7-4.3. Seattle's raw BW score is 5-9. But as they have played near the weakest schedules in the league, against a somewhat stronger league-average schedule that projects to a lower 3.9-10.1

I don't think the NFC is the "much weaker conference" this year

Our host's just-posted Week 15 efficiency ratings have 9 of the top 12 teams in the AFC, 3 in the NFC.

So Atlanta should be tied for fourth in your rankings if you don't factor in strength of schedule?

Atlanta is 4-1, +3 net, in in raw (non-SoS adjusted) Big Wins. The 4 is = 6th thru 12th, the +3 is = 4th thru 9th.

I think you meant the weakest division is the NFC West.

Yup! :-)

Bruce D. said...

Excellent James!

It follows very closely(obviously), overall points scored vs prevented(margin of victory), which some say is an accurate measure of team ability.

Subjectively, one would think no team wants to face the top teams in your list in the play-offs.

Its also another detailed rating system that points out that there's something weird going on with Atlanta.

Totally agree with your San Diego comment. Hopefully they'll sneak in somehow.

Anonymous said...

Well although based on won loss records one might expect that the Patriots will beat the Falcons in the Superbowl, this research seems to indicate that the Patriots will beat the Packers. The idea presented here seems sound. the thing is with luck such a major factor suprises can sprout too.

Post a Comment

Note: Only a member of this blog may post a comment.