Friday, October 20, 2006

The Myth of the Low Scoring Playoffs

Last season in MLS, we were treated to playoffs that saw numerous upsets culminating in the worst team in the running, the Los Angeles Galaxy, taking the championship. Unpredictable, yes. Too many teams? An unfair system? Possibly. You can make plenty of arguments that the current system is not optimal.

However, the one thing I will not even listen to is the argument that the playoffs are low scoring. After two scoreless draws in the opening weekend (COL-DAL, CHI-DC), people started complaining about the lack of goals, and blamed the system. 'It encourages teams to bunker in the first game,' they said.

Yes, last season's playoffs were low scoring, that's a fact. And yes, over past three years the aggregate quarterfinal series have been lower scoring than the single elimination games (2.29 vs 3.22). However, last year it was the exact opposite as the knockout games were devoid of goals, and it's important to note that there's been only 9 knockout games vs 24 aggregate games. The current system has only been around for three years. That's not a long time, and I think overall it's a little too early to make any judgments there.

Especially since the first two years saw their fair share of goals. It's short-sighted to consider the playoff system as broken, because last year was the exception to the rule. Last season's playoffs had a goals per game average of 2.09, way lower than the regular season average of 2.87. However, that decrease of 27% was the largest in MLS history by 10%.

Let's look at why the MLS playoffs are not low scoring:

1) It is normal for scoring to drop in the playoffs in MLS. In eight of ten seasons, the amount of scoring in the playoffs has been less than in the regular season. The exceptions were 1999 and 2003. Overall, the average playoff game has 7.4% fewer goals than the average regular season game. In 2006, MLS has averaged 2.62 goals per game. If that pattern holds, this year's postseason should average 2.43. Keep that in mind before you start complaining.

A history of MLS regular season and playoff scoring:

MLS Reg PO Diff PCT Diff
1996 3.37 3.12 -0.25 -7.5%
1997 3.26 2.77 -0.49 -15.1%
1998 3.57 3.14 -0.42 -11.9%
1999 2.86 3.06 0.20 7.1%
2000 3.19 2.94 -0.25 -7.8%
2001 3.28 2.72 -0.56 -17.1%
2002 3.01 3.00 -0.01 -0.2%
2003 2.89 3.00 0.11 3.9%
2004 2.61 2.55 -0.07 -2.6%
2005 2.87 2.09 -0.78 -27.1%





OVR 3.10 2.87 -0.23 -7.4%

As you can see here, two of the "best" four years came in 2003 and 2004, the first two years of the new system. So unless there's another big decrease this year, I would say last year was the exception. It seems to balance out overall. Take a look at this table showing how the three MLS playoff formats compare:

MLS Reg PO Diff PCT Diff
1996-9 3.26 3.03 -0.23 -7.0%
2000-2 3.17 2.88 -0.28 -9.0%
2003-5 2.80 2.55 -0.25 -9.0%

Basically the exact same thing on the field. Of course, the current system makes it much easier off the field with no Wednesday games, no "if necessary" games, and knowing the dates ahead of time.


2) It is normal for scoring to drop in the playoffs in American sports. It's pretty much the industry standard. I went and collected the scoring data for the major sports leagues in the USA over the last ten seasons, to compare to what's happened in MLS. I also included the World Cup's difference between the group stage and knockout round (1998-2006). Check it out:


Reg PO Diff PCT Diff
NFL 41.5 43.0 1.48 3.6%
WNBA 137.4 134.7 -2.66 -1.9%
NBA 191.2 185.3 -5.98 -3.1%
MLS 3.10 2.87 -0.23 -7.4%
NHL 5.45 4.97 -0.48 -8.9%
MLB 9.67 8.57 -1.10 -11.4%
WC 2.59 2.21 -0.38 -14.7%


Other than the NFL, all show a decrease. That makes sense because bad teams don't make the playoffs (usually). MLB had a decrease in 8/10 years, WNBA in 8/10, NBA in 9/10, NHL in 9/9. Again, this is a normal thing in sports. MLS is not alone.

Here's the individual data for each league:

(some WNBA figures may be slightly off because for a few seasons I could only find team averages and not actual total points)

MLB Reg PO Diff PCT Diff
1996 10.1 8.4 -1.7 -16.6%
1997 9.5 8.4 -1.1 -11.8%
1998 9.6 7.2 -2.4 -24.6%
1999 10.2 9.0 -1.1 -11.2%
2000 10.3 8.3 -2.0 -19.4%
2001 9.6 6.8 -2.7 -28.5%
2002 9.2 10.3 1.1 11.4%
2003 9.5 8.1 -1.4 -14.9%
2004 9.6 10.4 0.8 7.8%
2005 9.2 8.7 -0.5 -4.9%





OVR 9.7 8.6 -1.10 -11.4%


NBA Reg PO Diff PCT Diff
1997 193.8 188.0 -5.8 -3.0%
1998 191.1 182.5 -8.7 -4.5%
1999 183.2 175.2 -7.9 -4.3%
2000 194.9 183.7 -11.3 -5.8%
2001 189.6 187.2 -2.4 -1.3%
2002 191.0 188.5 -2.5 -1.3%
2003 190.2 179.5 -10.7 -5.6%
2004 186.8 176.1 -10.7 -5.7%
2005 194.4 194.2 -0.2 -0.1%
2006 194.0 195.6 1.6 0.8%





OVR 191.2 185.3 -5.98 -3.1%


NFL Reg PO Diff PCT Diff
1996 40.9 44.3 3.42 8.4%
1997 41.5 37.5 -4.03 -9.7%
1998 42.6 45.6 3.07 7.2%
1999 41.6 42.9 1.28 3.1%
2000 39.9 35.4 -4.56 -11.4%
2001 40.4 43.3 2.85 7.1%
2002 43.3 50.9 7.56 17.4%
2003 41.7 44.7 3.06 7.4%
2004 43.0 47.3 4.30 10.0%
2005 40.1 38.0 -2.10 -5.2%





OVR 41.5 43.0 1.48 3.6%


NHL Reg PO Diff PCT Diff
1997 5.85 5.35 -0.50 -8.5%
1998 5.28 5.22 -0.06 -1.1%
1999 5.27 5.05 -0.22 -4.1%
2000 5.49 4.64 -0.85 -15.6%
2001 5.51 4.80 -0.71 -12.9%
2002 5.24 4.78 -0.46 -8.8%
2003 5.14 4.70 -0.44 -8.6%
2004 5.14 4.61 -0.53 -10.3%
2005 No Season
2006 6.17 5.64 -0.53 -8.6%





OVR 5.45 4.97 -0.48 -8.9%


WNBA Reg PO Diff PCT Diff
1997 137.8 113.3 -24.4 -17.7%
1998 139.8 137.9 -2.0 -1.4%
1999 138.0 130.5 -7.5 -5.4%
2000 137.6 132.3 -5.4 -3.9%
2001 131.4 126.9 -4.5 -3.4%
2002 135.3 141.6 6.4 4.7%
2003 136.2 137.3 1.1 0.8%
2004 134.2 131.0 -3.2 -2.4%
2005 134.6 134.2 -0.3 -0.2%
2006 150.5 144.2 -6.3 -4.2%





OVR 137.4 134.7 -2.66 -1.9%

Comments on "The Myth of the Low Scoring Playoffs"

 

Anonymous Anonymous said ... (6:13 AM, October 20, 2006) : 

One change that *has* occurred is that the playoffs have gotten shorter. Because there are fewer games now, the post-season rate of scoring is more random.

For instance, you've predicted that this year's playoff scoring rate is 2.43 goals per game, but with Poisson variability, a 95% predictive interval spans all the way from 1.55 to 3.36 goals per game.

Unless the Poisson model is badly wrong, it's hard to predict how many goals will be scored. Perhaps scoring will be right where we expect, but it won't be surprising if it's not.

 

Anonymous Mike said ... (8:02 AM, October 20, 2006) : 

bad teams don't make the playoffs

Just eliminating the leagues worst teams would likely cause some drop off in scoring because the remaining teams will be, relative to the regular season, more balanced.

Perhaps this explanation could be tested by recalculating the regular season averages using just games between teams that made the playoffs. Presumably such games would be more balanced and lower scoring than the other regular season games.

 

Blogger ERic said ... (9:13 AM, October 20, 2006) : 

Ah, scaryice. Killing them with stats.

"Bad teams don't make the playoffs" is the most concise and reasonable explanation. Thanks for that!

 

Blogger ERic said ... (9:14 AM, October 20, 2006) : 

Interesting suggestion, Mike.

 

Blogger henryo said ... (11:33 AM, October 30, 2007) : 

Would love to see Mike's suggestion being realized(recalculating the regular season averages using just games between teams that made the playoffs), that would be comparing apple to apple.

Doing that any time soon?

 

post a comment