Using a Modified RPI Formula
to Predict the Playoffs
1/6/00
Email Mark
|
|
Despite all the pissing and moaning about the evils of parity, I think
it was a fabulous year for the NFL. The mighty fell, the meek inherited,
coaches found their inner, bawling child, and the lucky sometimes
prevailed over the good. So what? You can't tell me it wasn't exciting
trying to figure team and player performance from week to week. To
me, the game is much more interesting when the outcome isn't a near-given.
And for all the discussion of teams being mediocre, this is the first
year that four teams finished with 13 wins or more. For one of them,
13 wins wasn't even enough to win the division---that's never happened
before, either. OK, so a couple of 8-8 teams made the dance. Is this
really a big deal? Every year there are 9-7 teams that make it. One
game closer to .500, and those playoff-worthy 9-7's become lame 8-8's?
I doubt it.
But such confusion makes a tough job of distinguishing who's really
good from those just pretending to be. The Jaguars and Rams feasted
on the sick puppies in their division, yet performed poorly in tougher
contests. The Dolphins, Seahawks and Cowboys look like they should
be replaced with the Chiefs, Ravens and Panthers. Or should they?
Well, that's the point of the following exercise. After 16 games,
we have a full slate of data from which to compute power rankings.
If you follow NCAA basketball, you may be familiar with the RPI
rankings they use. The NCAA isn't officially talking, but it's generally
believed that teams are ranked on winning percentage (WPCT), that
of their opponents (OPCT), and that of their opponent's opponents
(OOPCT). Each of these figures are weighted and then summed. In the
RPI rankings, win percentage is weighted at 50%, with each of the
other two at 25%.
While this is a solid model, it needs some tweaking. The difference
of one victory can have a notable effect on the power rating when
you're only talking about 16 games. Had the Rams anything to play
for in Week 17, they probably would have won, giving them an .875
WPCT instead of .813. That would have elevated their power ranking
by 1/4 of a point, which as you'll see in the table, would be significant.
Still, victories are the most important indicator of power, so it
needs to be the most heavily weighted individual factor. I chose to
apply a weight (35.5%) that was just larger than the best statistical
measure of the group----OOPCT. Since those latter figures are based
on a sample of over 4000 games, fluctuations in those numbers are
the most meaningful. I weighted them at 34.5%, which leaves 30% for
OPCT. Each of the three percentages are multiplied by their weighting
factor, then added together. The result then has the mean power ranking
(essentially .500) subtracted from it, to show a relation to average
for each team. If the ranking is positive, the team is better
than average. If it's negative, they're worse than average. The
greater the distance from zero, the further away from average they
are.
Even with these adjustments, the vagaries of NFL play make any system
less than foolproof. It doesn't take into account home vs road records,
or margin of victory, or wins against tough teams/losses against bad
ones. And even with all of that, you can bet that the numbers would
have told you Jacksonville didn't have a chance on the road against
Denver back in 1996. So as they say in the wagering industry, the
following information is for entertainment purposes only. On the other
hand, I feel confident enough about what I see in the results to make
some playoff predictions from them, following the table.
By the way, perhaps you're wondering why I didn't use a BCS-style
ranking system. For one thing, a major part of that system-polls-doesn't
exist in the NFL. For another, assigning single-interval ranks (1,2,3,
etc) to categories, then averaging and summing the ranks, distorts
the differences between teams. The distance between 1 and 2 is not
necessarily the same as the distance between 2 and 3. I thought about
ranking the percentages and adding them that way, but the effect was
too distorting. So let's take a look at the table, and then talk a
bit about what we find…
| N F L P O W E R R
A T I N G S |
| TEAM |
WIN Pct. |
OPP W Pct. |
OPP OPP W Pct. |
POWER |
| Colts |
0.8125 |
0.4883 |
0.5310 |
1.1853 |
| Jaguars |
0.8750 |
0.3906 |
0.5100 |
1.0418 |
| Titans |
0.8125 |
0.4297 |
0.4993 |
0.9000 |
| Bills |
0.6875 |
0.5273 |
0.5212 |
0.8250 |
| Rams |
0.8125 |
0.3633 |
0.5042 |
0.7176 |
| Buccaneers |
0.6875 |
0.4531 |
0.5188 |
0.5940 |
| Dolphins |
0.5625 |
0.5625 |
0.5129 |
0.4851 |
| Vikings |
0.6250 |
0.4727 |
0.5122 |
0.4079 |
| Redskins |
0.6250 |
0.4648 |
0.5112 |
0.3811 |
| Chiefs |
0.5625 |
0.5156 |
0.5107 |
0.3099 |
| Jets |
0.5000 |
0.5859 |
0.5071 |
0.2863 |
| Seahawks |
0.5625 |
0.4844 |
0.5261 |
0.2692 |
| Raiders |
0.5000 |
0.5469 |
0.5103 |
0.1801 |
| Patriots |
0.5000 |
0.5195 |
0.5291 |
0.1629 |
| Lions |
0.5000 |
0.5430 |
0.4810 |
0.0673 |
| Chargers |
0.5000 |
0.4844 |
0.5271 |
0.0507 |
| Packers |
0.5000 |
0.4961 |
0.4978 |
-0.0152 |
| Ravens |
0.5000 |
0.4961 |
0.4878 |
-0.0497 |
| Cowboys |
0.5000 |
0.4688 |
0.5090 |
-0.0585 |
| Giants |
0.4375 |
0.5273 |
0.4873 |
-0.1795 |
| Broncos |
0.3750 |
0.5625 |
0.5054 |
-0.2336 |
| Panthers |
0.5000 |
0.4180 |
0.4883 |
-0.2824 |
| Bears |
0.3750 |
0.5430 |
0.4893 |
-0.3478 |
| Cardinals |
0.3750 |
0.4727 |
0.5154 |
-0.4686 |
| Eagles |
0.3125 |
0.5352 |
0.4822 |
-0.6175 |
| Falcons |
0.3125 |
0.5156 |
0.4539 |
-0.7738 |
| Bengals |
0.2500 |
0.5508 |
0.4683 |
-0.8406 |
| Forty-Niners |
0.2500 |
0.5039 |
0.4612 |
-1.0056 |
| Saints |
0.1875 |
0.5039 |
0.4668 |
-1.2081 |
| Browns |
0.1250 |
0.5586 |
0.4695 |
-1.2567 |
|
The rankings most definitely reflect the general sense that has prevailed
all season-the AFC East is the toughest division, while the teams
with the best records in other divisions have benefitted from weaker
schedules. Note that five of the top 11 are from the AFC East, while
3 teams from each of the NFC West and AFC Central sit among the bottom
seven. This combination is what elevates the Colts to the top
spot-playing half their games against teams with .500 records or better
mitigated the effects of playing a 5th place 1999 schedule. And of
course, they won 13 games.
Despite the second worst OPCT in the league the Jaguars still
come out 2nd, albeit distantly so. Their OOPCT was good enough to
help them, as was their NFL-best WPCT. By the way, don't discount
the importance of OOPCT. The main benefit is that it normalizes what
I would call the "intradivisional reciprocal effect." If the Jaguars
get to play the Bengals and Browns twice, and beat them, they hurt
their own OPCT because they've added four losses into that column.
OOPCT gets some of that disadvantage back, because the Bengals and
Browns have their own OPCTs raised by having to play the Jags twice.
In clearer English, the Bengals and Browns are lousy in part because
they had to play the really good teams. The OOPCT factor accounts
for this effect.
Below the Colts and Jags there is a definite dropoff in quality. You
might expect to see the Rams in this 2nd tier, but once again divisional
play reorders the standings. The Titans are hurt by the bottom
feeders in their division, but get a big boost from two games (and
two victories) against the 14-2 Jags. They fall out of the top two
by virtue of an OOPCT under .500. Still, their performance in big
games, and an 8-0 record at home, makes them nearly indistinguishable
from the big boys.
One of the bigger surprises would be the Buffalo Bills. While
finishing two games behind the Titans in the standings, their exceedingly
tough schedule (11th-highest OPCT; 5th highest OOPCT) vaulted them
over the Rams in total power. If they were playing a team other than
the Titans, and doing so in Rich Stadium, I'd be tempted to pick them
to have a shot at Atlanta. (But they are playing the ex-Oilers, and
it's in Nashville, and they're handing the QB reins to a guy who hasn't
gotten enough reps to be comfortable yet, so I'd be surprised if they
made it past this weekend).
The Rams follow up behind the Bills, but it's almost entirely
on the strength of their 13-3 record. Had they beaten the Eagles,
they would have likely moved into 3rd, but if there's a team that's
primed for disappointment, it's the Rams. At least, that's what the
numbers say. What they don't tell you is that aggressive defense,
a playmaking QB, and a great running back are the key ingredients
to a Super Bowl visit, and the Rams have those in spades. While untested,
they are the healthiest team, and the most balanced overall (especially
if you include the great special teams ability of Az Hakim). Near
as I can tell, the only combination that prevents St. Louis from getting
to Atlanta is a strong rush defense and potent air offense. The Vikings
and Redskins have the latter but not the former, while the Bucs and
the Cowboys have the former but not the latter. The Lions are essentially
average at both, and suffer from absolutely NO rush offense.
After the Rams there's another dropoff to the Bucs. The NFC
Central was competitive all year, and might have been even more so
had Jim Miller been allowed to complete the season for the Bears.
The combination of 11 wins and a strong OOPCT make them the best candidate
to upset the Rams. If Warren Sapp and the rest of the Tampa pass rush
can put NFL MVP Kurt Warner into hot-read situations, they have a
chance, but Shaun King would need to have the game of his life to
keep up offensively. They match up well with the other teams on their
side of the bracket, however, and I expect the Bucs to make it to
the NFC final.
At least the Seahawks won the games they needed to against Kansas
City-the Dolphins truly appear to have backed into the playoffs,
and currently wear the mantle of Team Least Likely To. Yet, what's
this? They're SEVENTH in the power rankings, ahead of five other playoff
teams, including this week's opponent, the Hawks? Believe it. Only
the Jets had a tougher schedule (the Broncos have the same OPCT),
and the Fins' OOPCT is similarly tough. That's enough to overcome
their 7 losses, and put them ahead of teams like the Vikings and Redskins,
who both won 10 games (and are nearly identical from a power ranking
standpoint). Does this mean they have a shot on the road against Seattle?
Very much so, especially if Ricky Watters can't play. Despite the
crowd noise of the Kingdome, it could be Jon Kitna who gets rattled
in his first playoff game. Every year one visiting team wins a wildcard
game, and I predict the Dolphins will be that team this year.
To cover the vastly similar Skins and Vikings together: they
both have one home game, they both have dominant offenses and pathetically
weak defenses, they both play teams way down in the power rankings,
they'll both win this weekend, and they'll both be victimized on defense
the following week (Skins on the ground, Vikes in the air).
So who should replace the Seahawks, Lions and Cowboys in the playoffs?
It looks like the Jets, Chiefs, and Raiders. The Chieftains actually
finished ahead of the J-E-T-S in the rankings, but entirely on the
strength of one more victory. The Jets managed an 8-8 record despite
the toughest schedule in all the NFL, and did it without key players
along the way. If Bill Parcells is smart, he'll take $6 million to
"rescue" Gang Green for one more year, take advantage of a 5th place
2000 schedule, and ride the recovered Vinny T for a run at the next
Super Bowl. Of all the teams shut out of the playoffs, the Jets are
clearly the most deserving. What's impressive is the way they finished,
when it was clear that they had no shot.
The Chiefs and Raiders also performed well, faltering only in the
key moments when they just had to win. Both of these teams would have
benefitted from 16 games at home, but the NFL tends to frown on that
kind of favoritism (and although they probably would consider it for
longtime good guy Lamar Hunt, they'd have to do it for Al "AntiHunt"
Davis, a prospect less appealing than daily colonoscopies).
Along with the Dolphins, the Lions are probably getting the
most press as undeserving, and while it's certainly true that they've
fallen and can't appear to get up, you have to take the express downtown
for quite a few stops before you get to the Cowboys. Dallas
faced the 7th-easiest schedule, and still only finished at 8-8. Losing
to teams like Philly and the Saints is particularly galling. And yet,
because of Emmitt Smith and Troy Aikman, they have a shot to beat
the Vikings. It should be close, but I still like the will of Jeff
George, Randy Moss and Cris Carter to prevail.
So that leaves some unfinished business. We have St. Louis representing
the NFC; who's in from the American Conference? Assuming the Dolphins
and Titans win, that sends Miami to Jacksonville, and Tennessee to
Indy. I don't think the Fins have a realistic shot, to be honest,
even without Boselli and a hobbled Brunell. I think Taylor gets it
done. I have to favor the Colts at home, based on the dome advantage
and the brutal preparation of the regular season in the AFC East.
Unless Eddie George can really eat the clock, the Titans won't be
able to keep up. Which gives you the expected conference final of
Indy at Jax. I'm going to assume that Brunell is ready for this game,
and he will be successful. But teams with solid rushing attacks have
been the most trouble for the Jags, and that's an edge to Edge. When
the linebackers and safeties can't tee off, holes open up. It should
be close, and fairly high scoring, but I like Indianapolis, despite
being on the road and playoff virgins. Hey, the power rankings say
so! And in a Super Bowl matchup against the Rams, the numbers say
take the Colts. However, the loss of Cornelius Bennett makes an ordinary
defense even more ordinary, and the Rams offense can match Peyton
and Co. score for score. Throw in a dose of revenge factor for Marshall
Faulk, and things look sunny in St. Louis this February!
Best of luck in your playoff pools, and
enjoy the fruits of a wild 1999 NFL campaign!
:: comments to mark
bunster
|
|
