Tagged: Predictability

MLS Playoffs – Predictions with Purpose (Updated)

To the chase…  my PWP Predictability Index leveraging my Possession with Purpose Analysis.  Click here for my latest revision or click here to read the initial revision.

NOTE:  Updates for the Red Bulls v DC United and Sounders v Dallas match are at the end of the article.

The Predictability Index itself is the CPWP Index data minus Goals Scored / Goals Against and is split into two diagrams – Home Predictability versus Away Predictability.

Here’s the CPWP Strategic Predictability Index for teams at Home:

CPWP PREDICTABILITY INDEX END OF SEASON 2014 HOME

Here’s the CPWP Strategic Predictability Index for teams Away from Home.

CPWP PREDICTABILITY INDEX END OF SEASON 2014 AWAY

Note the significant differences in how the teams are predicted to perform at home versus on the road; four teams really sucked at home this year, while four teams were expected to perform quite well on the road.  

Here’s how it works; I will compare the two digit number of the home team with the two digit number of the away team.

Whichever number is higher it’s that team which is predicted to win… again… based upon their history of team performance in overall attacking and defending, exclusive of goals scored; this year.

And now the PWP Predictions:

FC Dallas versus Vancouver Whitecaps matchup.  FC Dallas at Home (0.00) while Vancouver on the Road (-.11)  FC Dallas wins.

FC Dallas key indicators are ceding possession and creating quick counter-attacking scenarios that use time and space created by Vancouver being too aggressive in attack.

Vancouver key indicators are maintaining patience in possession and not losing position in defending – they are one of the top defending teams in MLS; they will need to be at their best to beat Dallas.

Next up; New York Red Bulls versus Sporting Kansas City.  New York at Home (0.10) while Sporting Kansas City on the Road (0.05) New York wins.

New York key indicators are their attack from a number of different angles.  They are simply one of the top attacking teams in all of MLS – they need to attack, attack, attack, and hope, with all their hope, that they can keep Sporting KC from scoring a goal.

Sporting KC key indicators are their ability to defend; they are still one of the best defending teams in MLS.  If they can control the wide open attack, I’d expect from New York, and their propensity for fouling in their own defending final third, I can see some individual talent from Zusi or some set-pieces giving them the edge to win.

Columbus Crew versus New England Revolution.  Columbus Crew at Home (0.06)  while New England on the Road (-0.08).  Columbus wins game 1.  Columbus Crew on the Road (0.06) while New England at Home (0.23) -> New England wins game 2.  I offer Columbus advances over New England on away goal difference.

Columbus key indicators include being one of the most consistent teams in overall attacking and defending team performance in MLS – with this being a two game set I’d imagine consistency in attacking and penetration as well as consistency in defending the danger spaces will see them through.

New England key indicators are slightly changed with Jones on the pitch – his leadership may give the edge to a Revolution team who are, in my opinion, outgunned in almost every other category.  They are a big under-dog in my opinion but not everybody rates Columbus as strongly as I do…

Real Salt Lake versus LA Galaxy.   Salt Lake at Home (0.33)  while LA Galaxy on the Road (0.12).  RSL wins game 1.   LA Galaxy at Home (0.19)  while Salt Lake on the Road (-0.01).  LA Galaxy wins game 2.  I offer LA Galaxy advance over Real Salt Lake on away goals difference.  

Salt Lake key indicators include, as noted, a stingy defense at home and a propensity to win in Rio Tinto.  They also have pedigree not unlike LA Galaxy, and perhaps an even more veteran line-up when it comes to big games.  Lest we forget Salt Lake could have done much better last year and didn’t – they will have added energy that might surpass the emotions LA bring with them in pushing to help Donovan raise the Cup once more.

LA Galaxy key indicators are pace, possession, penetration and all around purpose that operated at peak performance for almost the entire year.  It should be noted that they didn’t collect the silverware last week and in all likelihood they could stumble here as well as they may look past Real and consider the Cup is theirs…  So arrogance is an enemy as is the continued lack of mental awareness by Gonzalez…

More to follow after the games midweek after seeing who qualifies to play Seattle and DC United…

As for my own personal predictions I can see New York advancing as well as FC Dallas but the Vancouver defense is very good as is the Sporting KC defense.

I will go with Sporting over New York and Vancouver over FC Dallas because I think those team defenses are better – and for me it’s all about defense.

With respect to Columbus – I agree with my PWP Prediction model for that game as well as the game between LA and RSL…  and in this case I also happen to think the defenses for Columbus and LA are better.

More to follow:…

Seattle Sounders at Home (.22) while Dallas on the Road (-.20).  Seattle wins when playing at Home.  FC Dallas at Home (.00) while  Seattle on the Road (-.04).  FC Dallas wins at home.  Seattle advances on away goals difference.

For me, I can see Seattle beating FC Dallas at home and on the road.  Dallas may be a bit tired for game 1 and the Predictability Index hasn’t been built to address ‘tired legs’…

At the end of the day this should be a clean sweep for the Sounders…

DC United at Home (.03) while New York on the Road (-.03).  DC United wins at Home.  New York at Home (.10) while DC United on the Road (-.08).  New York wins at Home.  New York advances on away goals difference.

For me I can see a clean sweep here as well – it may be surprising but I can see New York, riding the wave of Phillips and, most likely, the last season for Thierry Henry, all the way into the Finals.  This is not intended to diss DC United.

They are a very good team but somehow I don’t see the ‘tired legs’ syndrome impacting the Red Bulls as much as Dallas… too much at stake for a team that has invested huge money in their players and program.

Best, Chris

COPYRIGHT, All Rights Reserved.  PWP – Trademark.

 

Advertisement

Gluck – Predicting Team Standings in Professional Soccer

CAN IT BE DONE?

Over the last four years I’ve conducted research on various professional soccer leagues and competitions.  To include Major League Soccer, the English, German, and Spanish Premier Leagues, as well as the UEFA Champions League and the Men’s World Cup of 2014.

Here’s my latest analyses on how the Possession with Purpose Index can be used to predict which teams will make the playoffs, qualify for the UEFA Champions League, or make the semi-finals of the World Cup..

Before beginning here’s a rerun on a few important items of interest about Possession with Purpose:

Intent:  Develop a simplified, strategic set of performance indicators to better understand the outcome of a game based upon primary inputs.

End State:

  • A documented method for measuring team performance from those indicators.
  • An index that ranks teams for their performance based on this method.
  • The index, while excluding points, comes close to matching results in the MLS league table.
  • Bonus – unexpected outcome – a tool to predict teams making the MLS Playoffs.

Key events to date:

  • Objective index developed in 2013
  • Results presented at the World Conference on Science and Soccer 2014
  • Approach published in the book – International Research Science and Soccer II – Routledge, Taylor, and Francis 2016
  • Leagues/Competitions evaluated
  • MLS 2013, 2014, 2015, 2016
    • English Premier League 2014
    • Bundesliga 2014
    • La Liga 2014
    • European Championship League 2014
    • Men’s World Cup 2014

Major League Soccer 2013 – The Maiden Year for PWP:

mls-2013

  • Nine of the top ten teams in the CPWP Index made the MLS Playoffs in 2013
  • Internal outputs from team performances showed that teams who cede possession (have lower than 50% possession) can be ranked within the top ten so the index is not biased towards teams that possess the ball greater than 50%
  • This doesn’t even include all the internal evidence on the various tactical styles of play each coach advocated.
  • Three of the bottom four teams replaced their head coaches as well.
  • It’s the initial results here that provided me compelling information to investigate deeper into what the outputs of the index might offer.
  • Each subsequent index shows a gold and red star – indicating which team finished first and last in the league table.

English Premier League 2014:

epl-2014

  • Winner of the League, Chelsea, finished 2nd in the index.
  • All four of the top four teams in the index advanced to the UEFA Champions League; those teams with green bars.
  • –By week 16, of 38 weeks, the four teams who advanced to 2015 UEFA Champions League were the top four teams in the Index; and they didn’t move out of the top four the rest of the season!
  • Three of the bottom four teams in the index were relegated in 2014; those teams with red bars.

Germany Premier League 2014:

bl-2014

  • Winner of the League, Bayern Munich, finished 1st in the index.
  • All four of the top four teams in the index advanced to the UEFA Champions League; green bars.
  • –By week 21 the four teams who advanced to 2015 UEFA Champions League were the top four teams in the Index; and they didn’t move out of the top four the rest of the season!
  • Augsburg and FC Schalke, who advanced to Europa League, finished 6th and 8th, respectively, in the index (light green bars).
  • For those teams relegated (red bars), SC Paderborn, finished worst in the league table and index, while Freiburg was 7th worst in the index and Hamburger SV was 3rd worst in the index.

Spanish Premier League 2014:

spl-2014

  • Winner of the League, Barcelona, finished 1st in the index.
  • All four of the top four teams in the index advanced to the UEFA Champions League; green bars.
  • By week 14 the four teams who advanced to 2015 UEFA Champions League were the top four teams in the Index; and they didn’t move out of the top four the rest of the season!
  • Sevilla and Villarreal, the two teams advancing to Europa League finished 5th and 6th, respectively, in the index; light green bars.
  • The three teams relegated in 2014 were Cordoba, Almeria, and Eibar.  They finished 2nd worst, 3rd worst, and 4th worst (respectively) in the index; red bars.
  • Of note; Levante, who finished worst in the 2014 CPWP Index finished last in the 2015 La Liga Standings.

UEFA Champions League 2014:

uefa-cl-2014

  • Winner and top team in the Index – Barcelona
  • Four of the seven top teams in the index advanced to the semi-finals
  • –Barcelona 1st, Real Madrid 3rd, FC Bayern Munich 5th, and Juventus 7th; green bars.
  • By the end of round one the top four teams to make the semi-finals were all in the top 10 for the index; with Barcelona 1st, Bayern Munich 3rd, Real Madrid 4th, and Juventus 9th.
  • Poor performers, APOEL Nicosia and Galatasaray finished 2nd and 4th worst (respectively) in the index; red bars.

Men’s World Cup 2014:

mwc-2014

  • Winner of the World Cup. Germany, finished 1st in the index, with 2nd place finisher, Argentina 5th best in the index.
  • Four of the top seven teams to reach the semi-finals finished 1st, 2nd, 5th, and 7th in the index; green bars.
  • By the end of round one, the four teams to make it so the semi-finals were all in the top six of the CPWP Index; with eventual winners, Germany 1st, Argentina 3rd, Netherlands 5th, and Brazil 6th.
  • With Brazil giving up seven goals to Germany in the semi-finals they dropped from 7th to 18th in the index.
  • France, Colombia, Belgium, and Costa Rica are the teams who made it to the quarter finals; light green bars.
  • All three teams that failed to earn a point in the World Cup finished worst (Australia), 2nd worst (Honduras), and 4th worst (Cameroon); red bars.

Side note about the Men’s World Cup:

  • USA finished 5th worst in the index (blue bar).
  • At that time I called for Jurgen Klinsmann to be sacked.  Why?
  • My two most compelling reasons were:
    • –Omitting Landon Donovan from the squad (huge reduction in squad mentality/leadership without his presence – plus he was simply the best striker/forward in the USA).
    • Replacing Graham Zusi with Omar Gonzalez late on in the game against Portugal – that replacement (a huge tactical error) created a vacancy in the area where Graham Zusi was defending; the exact same area where Ronaldo delivered his killer cross from.
  • Two years later, after numerous tactical and mental leadership errors, Jurgen Klinsmann was finally sacked.
  • I wonder where our team would be (NOW) if Sunil Gulati would have had the backbone to sack Jurgen Klinsmann back then?
  • I’m not afraid to say I told you so Sunil Gulati…

Major League Soccer 2014:

mls-2014

  • Four of the top ten teams, after week 1 CPWP Index, made the playoffs; with SSFC, eventual Supporter Shield winners in third.  After week 13 Seattle never fell further than 3rd in the Index.
  • Eventual Cup winners, LA Galaxy, were 11th after week one.  By week 8 they were 1st in the Index and did not fall out of the top two after week nine.
  •  Slow starter award goes to DC United, who were bottom of the Index until the end of week 5; when they finally breached the top ten.
  • It was here, along with seeing FC Dallas, at the top of the Index, that reinforced the Index was not overly influenced by teams who have high amounts of possession.
  • In other words, the Index would, and does, rank teams in the top ten even when they cede possession and play more direct/counter attacking football.
  • Although the first four weeks of the Index didn’t predict more than four of the top ten teams making the playoffs by week eight the Index showed nine of the top ten teams making the playoffs.
  • The level of accuracy, from week eight, going forwards never dropped below 70% and reached (and sustained 90% accuracy) by week 25 for the remainder of the year.
  • Accuracy in predicting the top ten teams making the playoffs was no worse than 40% (the first four weeks) and no less than 70% throughout the remainder of the year with 90% accuracy first attained by week eight – and sustained by week 25.

Major League Soccer 2015:

mls-2015

  • Seven of the top ten teams, after week 1 CPWP Index, made the playoffs; with NYRB, eventual Supporter Shield winners in ninth.
  • Eventual Cup winners, Portland, were 8th after week one.
  • Slow starter award goes to New England, who started at bottom after week one, but had breached the top ten by week seven.
  • At no time did the CPWP Index have less than seven eventual playoff teams in the top ten.  And by week seven nine of the top ten teams in the Index were bound for the playoffs.
  • Accuracy in predicting the top ten teams making the playoffs was no worse than 70% at any given time – and as high as 90% accurate by week seven.

Major League Soccer 2016:

mls-2016

  • Seven of the top ten teams, after week 1 CPWP Index, made the playoffs; with FCD, eventual Supporter Shield winners in first.
  • For those who were surprised by the Colorado Rapids this year – you shouldn’t be.  By week four, the CPWP Index had Colorado Rapids as third best in MLS; and they didn’t move out of the top four, in the Index, the rest of the year.
  • Slow starter award goes to New York Red Bulls; it wasn’t until week 12 that the Red Bulls breached the top four, but by week 14 they found their place at the top of the Index.
  • At no time did the CPWP Index have fewer than six of the eventual playoff teams out of the top ten.  And by week 25 nine of the top ten teams in the Index were bound for the playoffs.
  • Accuracy in predicting the top ten teams making the playoffs was no worse than 60% at any given time – and as high as 90% accurate by week 25.

Closing Thoughts:

  • The CPWP Index, and the sub-indices for team attacking and defending, show great value in looking to understand where failure/success may be occurring relative to team results.
  • It’s evidence – one piece of evidence – that shareholders should pay attention to when looking to make changes – it is not a substitute for what the eye sees or the gut feels.
  • I know more can be offered in drilling down into individual statistics relative to these team statistics.

Best, Chris

You can follow me on twitter @Chrisgluckpwp.

COPYRIGHT – All Rights Reserved.  PWP – Trademark

Real Madrid and Barcelona – Two Horse Race

For me, it’s not the top two that peak my interest this week, it’s the prime movers from mid-table – downwards while looking at the League Table from Week 12 to Week 19.

Here’s how they stand comparing Week 19 to Week 12:

La Liga League Table Through Week 19The teams highlighted in Green and Red I”ll get to in a bit, for now note the positional changes have been significant for many teams in and out of the lower half.

For those interested the CPWP Family of Indices continue to have strong correlation to the League Table without using Points Earned in the calculations.  Here’s how the Indices show things from a team performance standpoint through Week 19:

CPWP Strategic Index Week 19

APWP Strategic Index Week 19

DPWP Strategic Index Week 19Overall the CPWP Strategic Index has an R2 of .89; while the APWP sits at .89 and DPWP sits at -.81.

For those new to the Indices here’s an explanation on how they are created.  No other publicly created set of Indices comes any closer to the League Table – not even Expected Goals – a popular Predictability statistic.

I should point out that these Indices are not Predictability Indices – they are not built to predict the future based upon past data – but……..  this Index, developed from the PWP Process is a Predictability Index:

CPWP Predictability Index Week 19The caution I offer in using it as a forecasting tool is this – when developing a forecasting model you need at “x” amount of samples to reach 95% Confidence Level in your data and its ability to represent trends for the future.

The “x” amount of data needed for this Index is at least 15 games — since games is the primary sample point.  The twist is that since teams behave, for the most part, somewhat differently at home versus on the road you need 15 games of data at home and 15 games of data away from home.

Since this is only Week 19 that threshold has not been reached to substantiate that this predictability portion of this Index hits the 95% Confidence Level limit…

But, you say, the R2 is .77 – agreed – so yes, I would venture that those who like to gamble might want to rely on this tool to help them pick a winner – I did a test run in Major League Soccer, where the home and away statistics are notoriously different and my test run varied in success – straight CPWP PI # of one team compared to another.

That success ran as high as 75% to as low as 30% week to week for about 8 weeks – your choice…  By the way – the Predictability Index created from PWP is simply my Index outputs minus (missing goals scored for or against)…

Back to the movers in La Liga these last seven weeks…

Recall the teams Espanyol (+6), Real Sociedad (+7) (Nice one Moyes!!!), Cordoba (+6), Levante (-6), and Granada (-6)…

In reviewing the APWP Index for each team, from Weeks 1-12 and Weeks 13-19, only one team has seen their Attacking Index increase, Cordoba – all the other teams have seen their overall attacking performance drop slightly during those two time-frames.

Why has Cordoba shown an increase?

It’s down to improved accuracy in Scoring Goals based upon Shots on Goal – all others have experienced slight decreases in quality; either with respect to percentages of Shots on Goal, Shots Taken per Penetration, or Goals Scored from Shots on Goal.

In reviewing the DPWP Index for each team, from Weeks 1-12 and Weeks 13-19, two teams have seen their Defending Index decrease, Levante and Granada – all other teams have seen their Defending Index improve , with Cordoba seeing the most improvement by as much as 11%.

Cordoba’s improvement in Defending comes from Opponents having less quality in putting Shots on Goal from Shots Taken and Goals Scored from Shots on Goal.

Clearly Cordoba has improved on both sides of the pitch, while with the others it’s slightly more difficult to pin down a specific area…

A few interesting notes here are:

  1. Cordoba were bottom of the table, and even after having to play Barcelona, Villarreal and Eibar during this stretch they still gained 6 places, and
  2. The CPWP Index had Cordoba rated 12 best after Week 12, and that Index rating has not changed through Week 19 – meaning it is likely the CPWP Index really did a great job of accurately representing the true team performance of Cordoba compared to other teams in La Liga…
  3. Finally, the CPWP Predictability Index (PI) had Cordoba rated 12th best, after week 12 as well… (perhaps??) an independent data point to substantiate that the predictability nature of  the CPWP PI has value???

In Closing? 

Cordoba showed improved performance on both sides of the pitch while the others didn’t…  (perhaps???) this means that some of the new positions, for these teams, are as much a function of how others have gotten better, or worse, as it is a function of how those teams have, themselves, gotten better or worse…

Meaning position in the League Table, even when seeing changes by as much as six or seven places, may not mean that individual team is playing better – it may mean that other teams, with less noticeable drops in position are playing worse…

Reinforcing again that predictability is not solely associated with goal scoring – it’s also a function of not scoring because some teams are doing better, however slightly, with improved defending but not improved attacking…

If you are a writer for any team in the Bundesliga, La Liga, Barcley’s Premier League, or Major League Soccer and you’d like to use outputs from my Possession with Purpose Family of Indices in your articles please let me know…

I can provide a broad range of support that may help you better tell the story, (explain) to your readers, what or how well your team is doing compared to others… or even itself given certain time-frames (before and after a coach gets sacked, player gets injured, etc…)

If you’d like an example of the type of support I can provide please read this latest article by @7amkickoff.

Best, Chris

COPYRIGHT, All Rights Reserved.  PWP – Trademark

You can follow me on twitter @chrisgluckpwp.