Review: Statistics Series, Part 1: Analyzing the 2012 Season and Predicting 2013

Discussion in 'D.C. United' started by Shammypantaloons, Feb 12, 2013.

  1. Shammypantaloons

    Apr 20, 2012
    Club:
    DC United
    [​IMG]

    I am going to make a few threads relating to specific topics that facilitate statistical analyses of various degrees. A few potential subjects include:
    1. Attendance and related issues.
    2. The Castrol Index and other measures of 'effort' and 'skill.'
    In this first series I was hoping to discuss different measures or alternative ways to analyze and rank the various strengths of last year's team and our closest rivals as well as their weaknesses. I think that analyzing the facts, we can likely predict another playoff birth, but it will likely have to come a very different way.

    First, a table:

    [​IMG]

    What do we see from the table:

    1. The race for the playoffs ended with the DC United getting in by 6 points.
    2. The teams DC beat had the lowest average points earned among all Eastern Conference competitors. The teams DC beat had the lowest average position at season’s end.
    3. The teams DC lost to had the highest average point total of all Eastern Conference competitors. The teams DC lost to had the highest average position at season’s end.
    4. DC United earned an astounding 53% of it’s points from teams ranked in the bottom 6 at season’s end, higher than any other team in the playoffs in 2012.
    5. DC United earned only 27.5% of it’s points from teams in the top 10, lower than any playoff contender in 2012.

    There are of course some limitations here. I think it might be a better idea to use the opponent's strength at the time the teams play rather than "end of season record." For example, LA might be rank 3 early on, but end the season at rank 12. As team's play opponents, you should probably count whether they were a winning club at the time of the game and add them up at season's end to determine who was the strongest team.

    I am curious what some people think of these statistics. Some questions:

    1.What other data is relevant to the discussion here? What would you like to see as part of the conversation?
    1a. How important are statistics like goals scored? What about when goals are scores? Is it relevant to look at goals per 10 game blocks?

    2. When a team gets a very high number of points from teams low on the table and few points from higher ranked teams, what strategy is best taking on the next season?
    2a. Given collectible data, is the "look at the last 5 matches" approach to ranking teams or predicting success even worthwhile?

    3. How do you think teams did to adjust to specific limitations perhaps apparent in certain stats?

    4. What do you predict, the playoff race, that is, the top 6 teams in the league will finish as for 2013? I want to hold off my own views for a little while until people provide some input of their own.
     
    nobletea repped this.
  2. fatbastard

    fatbastard Member+

    Aug 1, 2003
    Lincoln (ish), Va
    Club:
    DC United
    Nat'l Team:
    United States
    I think you and Knave should get a room - that's a lot of (probably mostly meaningless) numbers.

    I think "strength at time of actual game" is much more meaningful than "strength at season's end", but still fairly subjective - were they on a winning or losing streak, etc.......
     
    Boloni86, Barbieri and nobletea repped this.
  3. nobletea

    nobletea Moderator
    Staff Member

    Sep 29, 2004
    HarCo
    Club:
    DC United
    Nat'l Team:
    United States
    Look I love data, I love to collect it and analyze it, and I use it more than most people. I d/l thousands of debit card transactions a year into a spreadsheet and analyze the hell out of it, just to see what I can learn from it.

    But sport, and especially soccer, has way too many variables and not enough importance in my life to care about this much data. Even in championship manager, a game that almost perfectly tracks all the data you could want and has the tools to analyze it, the only things I ever really paid any attention to were the PTS column and the Form column for players. I could give two craps if one of my defenders hasn't won a header in 10 games if his form is good and the team is winning.

    Alas, I'm not here to completely poo on your efforts.

    I think for a team that clearly, much more clearly than almost any team in the league, won points from the bottom and lost them from the top (we were talking about this last summer at mid-season), the strategy for the next season would be to just make the personnel better. It tells me the coach is probably doing the best he can with what he has and the players you have are doing the best they can with who they are, so if you're going to get better, you need new players. (allowing for player improvement year to year, of course)

    Goals scored doesn't really seem useful to me, particularly in any small chunk, as with most soccer stats. Most of the stats from soccer, in my opinion, are only useful over long periods of time with a lot of data to mark clear patterns, and even then, mostly only on the fringes. For example, you can say the team who scored the least goals in the league needs to make changes to the offense, and the one who scored the most, probably doesn't need to change much (even then you have to consider that offense and defense in soccer are not unrelated facets), but you can't really say a lot about goals scored if you want compare offenses between probably half or more of the teams in the middle of that table.

    It's just not baseball, to me at least.
     
  4. fatbastard

    fatbastard Member+

    Aug 1, 2003
    Lincoln (ish), Va
    Club:
    DC United
    Nat'l Team:
    United States
    bigger balls
     
  5. Barbieri

    Barbieri Member+

    Jul 8, 2004
    Decatur, GA
    Club:
    DC United
    Nat'l Team:
    United States
    What I want to know is, based on ALL data, how many points do we need for the playoffs and how many goals per game to get the maximum points.

    Concurrently, in the playoffs, based on ALL data - how many goals per match to win at each stage - and a overall average of what we need to score per game. (Knave does this)

    Tell me where I need to be performing to make the playoffs, and what I need to bag to win those game.

    So if I need 60 points, at 2.1 goals per game this works. If I need 1.4 goals in the playoffs, this works. If I need 2.6 goals in the MLS Cup this works.
     
  6. Shammypantaloons

    Apr 20, 2012
    Club:
    DC United
    Well one thing I noticed for instance is that Houston in the last few years has been fairly strong despite not getting the top seed in the league. While "getting a result" is important, they often have the best scores in a number of criteria, like "average position/points of team beaten." I think that while there are limitations to looking at end of season records, there is something to be said for the overall season-long course a team has taken.

    I was hoping to put together a running 'moon poll' this year, based upon "beating teams with winning records," streaks, goals and other factors, but I am still figuring out what might be most relevant.

    Also, Barbieri, I feel fairly comfortable saying that 54 points will net a team a playoff birth nowadays. Given that milestone, I think you can really manipulate how you approach certain games with the objective of reaching that goal. Last year we had two very different trajectories heading into the playoffs. One team had 50% of their points from top 10 teams and the other had 50% of their points from bottom 6 teams. It seems like the conversations we had last year, where we "gave up" on the road versus better opponents makes sense, given a possible strategy to win at home or to win against weaker opponents. Is that what we did at certain junctures? Maybe. I mean, if you're already in the playoffs, does it make sense to push it in the last game of the playoffs (as another example)? RSL basically gave up against Seattle last year because there was no reason to kill themselves when they had already made the mark.
     
    Barbieri repped this.
  7. greatscott

    greatscott Member+

    Dec 21, 2002
    Richmond
    Club:
    DC United
    Nat'l Team:
    United States
    All you need is one pissed off person who wants to win and all these numbers go out the window.
     
    Boloni86 repped this.
  8. nobletea

    nobletea Moderator
    Staff Member

    Sep 29, 2004
    HarCo
    Club:
    DC United
    Nat'l Team:
    United States
    See that last bit is the problem with trying to use stats to figure it all out, I think. Because you might look at that say, well, RSL lost to Seattle, so what does that mean? In that case it doesn't mean anything about how good either team is because of the context. It's like trying to say something qualitative about Manchester City or QPR because they tied on the weekend with a CL semi coming on Wednesday. And I think there are context examples like this all throughout the season that make the data hard to depend on.

    That's just me.
     
  9. Barbieri

    Barbieri Member+

    Jul 8, 2004
    Decatur, GA
    Club:
    DC United
    Nat'l Team:
    United States
    OK, fair - I see what you are driving at, as much to say you can "select" where you can pick up points. To whom and to where I think is subjective (with exception Chivas or Toronto) - but I see what you are trying to tease here.

    What about goal need to beat weak opponents v. stronger opponents - I think there is some math as what we need to get to typically beat a caliber of team, to what we can score based off what chances a player makes a game.

    My example is, against SKC we need 2.7 goals to win... which means if the team wants to net that goal based on their current GPG to SPG/SOTPG, they need % more chances to net that goal, in essence how much harder does the team need to work to generate those chances to get that goal.
     
  10. Shammypantaloons

    Apr 20, 2012
    Club:
    DC United
    Nobletea, Right well once teams cross the threshold into the playoffs I would certainly take that into account and outcomes might receive less weight. However, it is possible that a team might purposely not care about their position in the league, as a strategy, which I think we did to some extent, and that might be purposeful and actually a good idea at times.

    Greatscott, Also, it's true that a player or two can start playing out of their mind and the predictions might now come true. However, what we're talking about are likelihoods and not guarantees. You could say with varying degrees of confidence that X team can beat Y team given ABC factors. Injuries, weather and other things can always interfere with these kind of schemes, but I think at the very least you can make a more accurate 'rankings' table than are generated using either the standard league table (who has more points) or how teams are performing in their last 5 games.
     
  11. Heist

    Heist Member+

    Jun 15, 2001
    Virginia
    Club:
    DC United
    Nat'l Team:
    United States
    The main problem with this portion of the analysis is that every team plays different teams and DCU didn't get to choose its schedule and will have a different one this year. You can expect its overall opposition to be tougher this year and last year turned out to be easier than most others. Remember how easy our last 5-7 games were this year. Part of why the schedule looked easier in retrospect is that we did well against the teams we played 3 times so they ended up slightly lower, but it was an overall easier schedule and we punished the bad teams. I think that if we want to do well this year, we need to bag more road points against better teams and we'll have more chances this year too.
    A better thing to look at would be how far we outperformed our expected outcome based on weighted records of each team. Part of it was us beating up on the bad teams in the East, and part of it was that we mostly played bad teams 3 times and avoided some good teams 3 times.
     
  12. Shammypantaloons

    Apr 20, 2012
    Club:
    DC United
    That's not a problem with my analysis, it's kind of a main issue for next season :p
     
  13. Heist

    Heist Member+

    Jun 15, 2001
    Virginia
    Club:
    DC United
    Nat'l Team:
    United States
    Sorry...I meant it was a problem with the conclusions you drew and numbered 1-5 in terms of their value in describing last year and in terms of planning for this year.
     
  14. Shammypantaloons

    Apr 20, 2012
    Club:
    DC United
    I see, I really wanted other people to offer input before I proposed what I thought next season could be like for us. Clearly, I don't think we will get 30+ points from bad teams again. This poses an interesting dilemma for us, especially if the bottom tier clubs improve over last year (that is, compared to last year, the last 10 teams are generally better than this year).

    1-5 was really just about 2012 and any predictions were meant to be inferred or proposed by others :p
     
  15. GumbyG

    GumbyG Member+

    DC United
    Mar 22, 2007
    Chesapeake, VA
    Nat'l Team:
    United States
    For me then, the question is, what makes the bottom six the bottom six, and the top 10 the top 10, aside from a statistical inflection points (?) from last year. Because the number of teams we'd classify as bottom or top tier will almost certainly change if we were doing this analysis at the end of any other year.

    As you can see, I don't brain number things.:coffee:
     
    Barbieri repped this.
  16. BigKris

    BigKris Member

    Jan 17, 2005
    Falls Church, VA
    Certainly, one can over-do it with the stats, especially in a sport like soccer that isn't as easily quantified as say baseball. However, I think it's worth trying to glean what we can from it.

    I'm very interested in this phenomenon where we earned more points from bad teams -- and failed to earn points from the top teams -- more so than our peers. Nobletea, above, says this could be a sign that the coach is doing the best he can but needs better players. I was wondering if it might be the opposite: could it be a sign that United's tactics are more naive, such that they work against poor teams but aren't good enough so to beat more sophsticated opponents?
     
    Winoman, Barbieri and nobletea repped this.
  17. nobletea

    nobletea Moderator
    Staff Member

    Sep 29, 2004
    HarCo
    Club:
    DC United
    Nat'l Team:
    United States
    Great point. That fits well also. Clever tactics, which are really just maximizing the resources you have under stress, could fill the talent gap. If Ben just rolls out the same players or gives the team the same instructions every week, ie, "play your game" that may not be enough to compensate for a talent gap.
     
  18. Shammypantaloons

    Apr 20, 2012
    Club:
    DC United
    Well for this upcoming season, we still play a lot of potentially bad teams. Of the teams that could be bottom 6 clubs next year and how we play them, home (H) or away (A) in the series:

    2013
    Philadelphia (x 3): HAH
    New England (x 3): AHA
    Toronto (x 3): HHA
    Dallas (x 1): A
    Portland (x 1): H
    Chivas (x 1): A
    Montreal (x 2): HA
    Colorado (x 1): A
    7 home, 8 away

    Whereas last year we played the potentially bottom teams like this:

    2012
    Philadelphia (x 3) AHA
    New England (x 3) AHH
    Toronto (x 3) AHA
    Dallas ( x 1) H
    Portland (x 1) A
    Chivas (x 1) H
    Montreal (x 3) HHA
    Colorado (x 1) H
    9 Home, 7 Away
    From these games our H record was: 7-1-1, and A record was (4-2-1)

    So we kind of lose home field advantage against potentially bad teams as well, making it more difficult to win against them.

    Looking specifically at our away record (5-3-9):
    Loss: LA Galaxy, San Jose, Houston, New York, Columbus, Kansas City, Montreal, Real Salt Lake, Houston
    Win: New England, Toronto, Toronto, Philadelphia, Philadelphia
    Draw: Vancouver, Portland, Chicago

    So, we never beat a good team away, we always beat a bad team away, and we drew against teams that will likely not be in the bottom 10, I believe.

    This creates a different set of problems as we look at our away matches this year where our schedule looks like this (in order) with better competition bolded:
    Houston; New York; Kansas City; Columbus; Chicago; New England; Seattle; Colorado; Chicago; Philly; Montreal; New York; Chivas; New England, Toronto; Kansas City

    So in effect, our road schedule is harder in the first 3/4 of the season.

    What do you think this means for the season or for when the best time to acquire talent is for DC United?
     
    nobletea and Atouk repped this.
  19. Barbieri

    Barbieri Member+

    Jul 8, 2004
    Decatur, GA
    Club:
    DC United
    Nat'l Team:
    United States
    I think if we lose a string at home Ben gets panicky - his philosophy is Fortress RFK

    *To elaborate - I don't think away losses make us reach into the purse unless it beings to impact the ~PTS-to-PLAYOFFS total. At this point, the cash might not be available due to injuries. This then becomes the pinch, and expect to grind like we did when DeRo was out.

    We lose DeRo early - we're scrambling. Cashgone.
     
  20. Shammypantaloons

    Apr 20, 2012
    Club:
    DC United
    I think moreso than last year our hopes of the playoffs rest in the first half of the season. If we win even 2-3 games on the road against top competition, we're in a a decent position to finish where we were last year ± 6 which is all we need really.

    I haven't looked closely yet, but I want to also look at our competition and who they face in the first/second half of the season. I suspect that our competitors have an easier path in the second half than they did last year (Columbus/Chicago) but I need to check more closely.
     

Share This Page