30 March 2006

Setting the Pace


In a concept that I am completely stealing from Nats Blog, sometimes it's too easy to get trapped into the current standings and situation. Say DC United has a 3-1-2 record after six games, good enough for 11 points. What does that really mean? Is that a good enough clip to sustain to get them into the playoffs? What about the Supporter's Shield, which would guarantee Champions play? Really, those are the only two questions the regular season needs to answer, and to help answer them I've gone ahead and stolen the idea of the "Pace Team."

Basically, the pace team is a mythical team that earns enough points that it will arrive a predetermined state. For MLS, I have created two pace teams. The first is a team that will win the Supporter's Shield, and the second is a team that will finish at least fourth in the East. I looked at results for every year since 2001 (when MLS went to two conferences, though not with the existing playoff selection system.) Ultimately, that was a waste of time, since the key data points were all from 2005.

The key data was to look at points achieved by the team that finished second overall in a single table, and the team that just missed the playoffs in the East. Then find the highest points-per-game yield of those teams. Next, find the lowest points per game yield of the teams that did win the supporter's shield or made the playoffs that's higher than the maximums just found. Those are the pace numbers. Now, with future expansion, there may be an upward pressure on those numbers, but this will work well enough for now.

In 2005, the New England Revolution earned 59 points at a pace of 1.84 points per game. That's a great pace, and had they done it any other year than 2005, they would have won the supporter's shield. Instead, San Jose was averaging 2.00 points per game, and took the shield. That makes 2 points a game the lowest possible safe number for the Supporter's shield.

Similarly, in 2005 the Wizards took an average of 1.41 points per game. Good enough to make the playoffs in any other year, but not good enough to match the Metros 1.47 2005 yield. Therefore, 1.47 points per game is the lowest safe number for making the playoffs.

I'll be keeping track of how close DC is to both playoff and the Supporter's Shield pace throughout the year, and hopefully it will give a sense, regardless of the current standings, of how much ground your team needs to make up in order to get the right regular season results.

For instance, in the example proposed above, United's 11 points put them just off the Suppoter's Shield pace team's 12 points after 6 games. A win in Match #7, and things are nicely on schedule. They're also 2 points ahead of the Playoff Pace team's 9 points after six games (actually 8.82, so we'll round it up.) A loss next week is okay, since the playoff pace will only improve to ten points (well, 10.29, but you get it). After that, there isn't much slack.


At 30 March, 2006 13:53, Anonymous john said...

This is a really interesting idea and I'd love to see how this plays out over the course of the season. The scientist in me wants to make plots out of this already.

At 30 March, 2006 16:46, Blogger scaryice said...

Interesting, but it's really pointless. 1.38 is basically a .500 record (12-12-8), so what you're saying is if they're better than .500, they'll make the playoffs.

Which is not news, because KC last year was the only team in league history to miss the playoffs with a winning record. And that's only because two teams were absolutely awful last year. This year, things will be more even and a team with a losing record will almost certainly make the playoffs.

At 30 March, 2006 17:37, Anonymous Anonymous said...

It's not pointless, but I think you should adjust your pace numbers slightly down because Chivas and RSL aren't likely to give everyone else quite so many points this year.

At 30 March, 2006 18:01, Blogger D said...

Ice and Anonymous: Good points, but what I'm hoping to do is actually use at least something resebling a statistical basis for determining these things. So while the number are out of whack, and primarily due to 2005, there aren't enough data points to really consider them outliers yet. Maybe after another five years I can try and estimate a reasonable mean, throwing out the more extreme cases. But right now it is what I've got.

In all liklihood, 1.85 should get you the shield, but since at least one team has hit 1.85 and not gotten it, I need to leave it in until there is more data.


Post a Comment

<< Return to The DCenters Main Page (HOME)