11 September 2007

Retrospective: Why does it go right (or wrong)?

Continuing some earlier thoughts on the season-to-date, we've already broken down the season into some distinct phases, but I want to take a slightly closer look. First, let's take a look at the most recent annotated pace graph (click to enlarge):

Now, we speculated on some reasons why the season might have behaved this way, but let's look just a bit closer, shall we? The graph below is DC United's goals for and goals against in league games. Don't worry if it looks like a mess, because it is a mess and we're going to move beyond it shortly. I just wanted you to see it for the data purists among you (Again, click if you'd like to enlarge):

So that's a bloody mess, but I think we can refine this to something that's useful. The problem with this graph is that there's too much variation from point to point to discern any real trends. A lot of noise, if you will. However, let's borrow a tool from our wall-street friends and apply a moving average to this data. If we start at the third game of the season, and take the average of the first three games, we'll plot that point at game 3. Then average games 2-4 for game 4, average games 3-5 for 5, and so-on. In so doing, we smooth out a lot of the noise, and I think a story becomes apparent.

What I like about this graph is that it kind of syncs up to many of the turning points that I and several commenters were referring to. First, we see how bad DC was in the beginning of the season. United has allowing 2 or more goals a game, and averaging one in response. When you don't score. Starting in Game 5, we see the goals pick up to 1.5 to 2.5 goals per game and defense stabilize at around 1 goal game.

Now, the defensive move can be explained by the switch to the 4-4-2, but why would goals per game suddenly pick-up? Part of it is the Emilio slump, but I think a large factor is simply that United's talent level wasn't a 1 goal per game team. This team's forwards and midfielders should generate 1.5 to 2.5 goals per game, regardless of whether you're using Moreno, Emilio, Gomez and Fred, or you can use Addlery, Kpene, and Olsen. There's enough talent there, really. The 1 goal per game was a team underperforming, and the ship was righted enough to generate some good results. Our talent level is 1.5 to 2.5. Keep that in mind, it'll be important in a bit.

Now, when scoring 1.5 to 2.5 goals per game, and surrendering about 1 goal per game, we were a decent team, but not setting the league on fire. Then we took a slide. Goals fell to the 1-2 goal per game range, and we were allowing a 1 to 1.5 goals per game. This is the period when we lost to New York and Houston. The big change was a defensive one: at similar scoring rates we had done acceptable previously in the season, but the goals allowed cost us points. Yes, there was a dip in scoring (attribute it to Moreno and Bolivian National Duty if you'd like) but had the defense played the way it had previously, I doubt you would have seen a real effect. The problem was defense. Were there any changes in the defense around this time? I can think of one - the Erpen for Vanney deal.

Erpen for Vanney happened one game before the slide starts to develop, and to me it is clear evidence that while Erpen and Boswell may have given you heart attacks, they meshed better than Boswell and Vanney did to start Greg Vanney's tenure in DC. The two simply didn't play well together, with Boswell certainly looking the worse in the exchange. It also further unsettled a backline that was making due with Josh Gros at left back. Things weren't jiving right.

But the story changes yet again with the introduct of Marc Burch in Game 18, and Clyde Simms wins the starting job a few games later. Suddenly the shaky United defense settles down to averaging less than a goal per game, their best performances of the year. Scoring is consistent with the rest of the year, but with Burch at left back things change. The six game win streak was fueled by defense. Why is Burch the key? A few thoughts on that:

Burch frees Josh Gros to take on midfield duties, which he's more suited too from years in Nowak's system. Burch, perhaps more importantly, makes Olsen and Fred more comfortable on the wing. Suddenly freed from having to run all over the field and worrying about what's behind him, Olsen settles in on the wing and really shines. Clyde Simms improves the midfield defensively, and Namoff returns at right back feels more comfortable. The benching of Boswell, it should be noted, happens after the defense is showing results, but perhaps it doesn't hurt, as either Vanney or Boswell seem decent in the center, but both seems problematic.

Still, the most recent games show some warning signs. United is once again allowing more than a goal a game (in fact, closer to two goals a game now) and is getting results because of an atypical surge in offense. I don't expect our strike rate to remain the way it is, and that means once it returns to its seasonal level we'll need the defense to re-stabilize. Part of it maybe the quality of opposition: United is #1 ranked for goals scored, but they just faced #2 and #3. That would be the "it's a momentary blip" explanation, and it is the explanation I want to be true (and, to be fair, I think it is probably true). Or it's a warning sign that things have become unsettled (and perhaps I just don't want to see that.) Again, the strike rate is ridiculously high at the moment, but pretty good at where it's typically been since June.

The lesson for me? Offense is what allows United to compete, but defense is what will decide whether United will dominate or not.

Labels: , ,


At 11 September, 2007 00:26, Anonymous Anonymous said...

This comment has been removed by a blog administrator.

At 11 September, 2007 01:57, Anonymous jeff said...

...and now back to our regularly scheduled program!

At 11 September, 2007 07:49, Blogger D said...

Jeff: What an odd bit of comment spam that was... I've deleted it, mainly because it was, well, odd, and possibly anti-semetic, and certainly not soccer related.

At 11 September, 2007 09:20, Anonymous Anonymous said...

Good analysis. I think the combination of Brucj and Vanney really opened things up Clearly Burch's offesive threat plus his defensive capabilities has really opened up Fred's game (just look at his numbers). this in turn has relieved pressure on Gomez, which has relieved pressure on Jaime. Namoff's speed has freed Olsens run, though Olsen is just having a banner year no matter what.

Simms has also helped by creating a defensive triangle with vanney and McTavish that completely disrupts play and drives it to the wing.

Against RSL, I would like to see Soehn take a risk and bench Vanney to give him a rest. I know it's a risk, but truth be told, it's time vanney got to take a breather for the stretch run.

At 11 September, 2007 10:08, Blogger Oscar M. said...

ACK the data purist in me has to point out that the Goals For/Against graph should be a bar graph, not a line graph.

At 11 September, 2007 10:49, Anonymous Skippy said...

There was no mention of McTavish in the back in your analysis. He's had a couple of really good games as I recall, although there has been criticism as well. Still I felt that he did generate some good defense. Or maybe I just like him because "McTavish" is a great name to scream from the stands whenever he touches the ball.

At 11 September, 2007 11:56, Blogger The Bird said...

Great analysis, as usual. Anonymous makes some good points about how one piece falling into place (Burch) can allow others to stop compensating, and concentrate on their primary tasks.

The idea that Vanney and Boswell should be kept apart is interesting, too. Any way you look at it, it's nice to have options.

Speaking of options, it'll be interesting to see how Gros' availability in the midfield will affect things. He's both a reliable 75' sub and a manageable starter if Fred or Olsen needs a game off.

At 11 September, 2007 12:29, Blogger elopingcamel said...

Fantastic analysis! I've been gone on my honeymoon (in a little town in Mexico--no access to MLS) and it's been so refreshing to get back and see DC at the top.
I'm right there with you on the subject of Vanney/Boswell. That duo isn't nearly as strong as the McTavish/Vanney or McT/Boswell duo.

At 11 September, 2007 19:04, Anonymous Anonymous said...

Too much math to get to "offense wins games; defense wins championships."

At 12 September, 2007 08:44, Blogger Bob said...

I think that the third graph should also include the full season moving averages, so you can quickly see how the past few games have shown above average scoring. You will also see a danger point if the scoring average drops below that trend line.

At 12 September, 2007 12:05, Anonymous matt w said...

I'd like to know what was said at that players-only meeting before the Revs game.

At 12 September, 2007 16:19, Blogger Sean said...

Hey, D? I think -- you're a geek.

Just so you know.


Post a Comment

<< Return to The DCenters Main Page (HOME)