|
|
Posted by
,
October 28, 2014 |
|
Email
Aaron Nielsen
Twitter
@ENBSports
|
|
Read this on your iPhone/iPad or Android device
|
|
|
|
|
|
|
Prior to this season I did statistical modeling for all MLS clubs going into the 2014 season, including Toronto FC. The basic model is fairly simple where you take historical data both on each individual player, the club itself, comparable players in the league and the model will try to predict what the player’s statistical totals will be for the upcoming season. Publicly I only show results for minutes played, goals, assists, shots on goal and disciplinary record, but my work includes more advanced stats to come to final projections. With the combination of all players on a team it can give you an idea on how successful the club should be based on its current roster.
What excites me most about my work is with this approach I have also predicted results for leagues that I don't even follow closely. Last year I used a similar model for the Dutch Eredivisie as part of a contest and at the end of the season this model was more successful than any other statistical models. Even more amazing, it out-predicted by a far margin all the "experts" of the league, including television broadcasters, pundits, magazines and reporters who exclusively work on covering the Eredivisie, a league I spend maybe 40 hours a year observing.
The model has also worked very well with regards to analyzing MLS, including this season. The model successfully predicted Seattle, LA Galaxy, DC United, New England as some of the power teams in the league, and I projected all of the playoff teams correctly except for Portland over Vancouver in the West and Toronto over Columbus in the East.
Recently, fellow RNO writer Steve Bottjer published a report card on the 2014 Toronto FC season. For those who follow RNO, they are aware that Steve writes the majority of the match reports for Toronto FC games, also acts as a beat reporter being with Toronto FC on many training days and has been doing so since 2009. In my role in looking for information about the club, I have a great deal of respect for his work because I feel his views are honest and not trying to create a story or pander to TFC.
With that said, I thought it would be interesting to look at his grades and compare them to my preseason expected stats for Toronto FC going into the season this year.
I did have Toronto as a contender and in first place in a tight Eastern Conference through my evaluations and I'm very disappointed how the season played out. I also agree on most of Steve’s assessments regarding the players, although I feel a discussion around expectations versus reality is very interesting.
In this 2014 final grades, Steve rated Justin Morrow as his best player and Jackson as his worst. I agree Morrow had a good season, is a solid quality MLS player and finished the 2014 season playing in 31 games with three assists. Morrow did almost exactly what he was expected to do based on his history.
Jackson played in 26 games, scored four goals, one assist and 19 shots on target. I had him projected at three goals, six assists and 16 shots in 27 games. Although his assists were down from expected, most of Jackson stats are very similar to what he was expected to do and if you look at his more advance stats of key passes, passing percentage, defensive involvement and shooting percentage, his statistics are almost identical of what he achieved last season in Dallas. I wrote in my preseason evaluation that Jackson had a tendency of being selfish and making mistakes, especially regarding discipline.
In analytics terms, both Morrow and Jackson performed basically how they rated prior to the season. Long term, I'm not sure if they either helped increase or cost TFC that many points, and I question the salary both players are on in terms of managing the cap.
The second group of players who stood out in terms of grading were Mark Bloom, Nick Hagglund and Daniel Lovitz. Again, I agree with Steven that these players had strong seasons and I commend TFC management this year for their drafting of two players who I rated, and for signing Mark Bloom last season from a lower league in North America.
The main story with regards to them, in terms of the expected statistics, is the analysis projected them to play 11 games and 725 minutes, while they actually were asked to play for a combined 69 games and 4960 minutes. Despite their strong seasons, if the model knew this many minutes would be expected out of these players the projection for Toronto going into the season would of been worse, especially their defensive record.
This leads us to some of the disappointments from 2014. Bradley Orr, the underused Ashtone Morgan, Jeremy Hall, Andrew Wiedeman, the departed Gabe Agbossoumonde and Alvaro Rey, were all players who had higher expectations going into the season, based on what they achieved in previous years. I think this issue has more to do with management than the individual players. Agbossoumonde, Rey, Morgan, Hall and Wiedeman had some value to the club, yet it was never reflected in the minutes they were given. If you add inn the fact that all these players salaries count against the cap, it meant Toronto was willing to waste close to 1/6th of their salary cap room.
For me, Bradley Orr was probably the biggest disappointment from a player who played at levels perceived to be higher than MLS. Considering his relationship with the coach, you expected the player to be more of an integral part of the club. The rest of defense mostly did as expected, although the injury to Caldwell did effect the club, while not being fully recovered and a lack of confidence in Doneil Henry, meant the team struggled especially during the last five games of the season.
Overall, I think my grades would have been more critical and the grading that I disagree the most with Steve is in goalkeeping and Joe Bendik. It is difficult to look at the team’s goal difference as well as position in the table, both close results and blowouts, and not question if this area can be upgraded. Bendik’s statistics show a 1.67 goals against average this season and .648 save percentage. While the argument might be the defense was weaker this season, other than an injury to Caldwell there weren’t many players lost in those positions for an extended period of time. Even last season, Bendik’s stats were not great as he had a 1.39 goals against average and a below average .662 save percentage. Meanwhile, the keeper he replaced was the everyday starter for the best club in the league and Seattle managed to have him cost about the same. I would give Bendik a chance next season, but I would bring a number of keepers to challenge him for the position and wouldn't make him an automatic on the team sheet.
I also have my concerns for Jonathan Osorio with the basic question of how should he be used? I will admit he is a good technical player and statistics show he can pass the ball, but to finish the complete season with only six shots on target, from someone who many perceive as an attacking midfielder, brings up some questions. He did score 3 goals and add 4 assists, almost exactly as projected, so I don't think he cost the team points this season. However, I think you would want more return for a player in attacking midfield or wing position and more defensive return than 20 tackles and five headers won from a holding midfielder.
Based on action per minute, Kyle Bekker had a much stronger season than Osorio, with a greater ability to find room and shoot, more key passes and was much more active on the defensive end.
Finally we arrive at the three DP's. Most of the season when Defoe played he was on pace as expected this season, and his eleven goals in his first thirteen games made his return almost greater than projected. Defoe did fail to score in his last six games and only had three shots on target. Overall, I would say Defoe played as expected: he is an exceptional goalscorer with an ability to shoot both inside and outside the box, and is a quality passer. Although unlike other DP's, such as Keane or Henry, Defoe doesn't get involved in the play as much and if he does return I think you are dealing with diminishing value, and overall a player who is more valuable in a team where other players can be expected to hold up their end.
On the positive side, Gilberto had attained some positive statistics in Brazil’s Serie A. This is a league regarded as a higher quality than MLS, and because of this the statistics projected 11 goals and two assists, making him and Defoe one of the top partnerships in the league. However, there was also warning signs from the start that questioned how each player will play together as both relied on others to provide them opportunities. Gilberto had one of the worst rations of converting shots on target to goals, as he led the team with 73 shots but only 22 were on target. Many of these missed shots were taken in games TFC lost, and in some cases Gilberto took the shot instead of Defoe. If Gilberto’s conversion rate was closer to Defoe’s, or the other DP strikers in the league, he would have scored as many goals as expected at the start of the season. What Gilberto can be commended on was his work rate, which helped lead him to a team high five assists, and if he is reined in a bit, he could be a quality player in the MLS.
This was also the main issue regarding Michael Bradley, where he was asked do too much and it had a negative effect on the team. Based on how the team was set up and Bradley’s reputation, it’s not surprising that his offensive expectations were high despite playing more of a holding midfielder role. Based on some of his more offensive seasons in Europe, he was expected to score five goals and add seven assists, where his actual return was just two goals and four assists. Bradley had one of the worst cross, corner and free kick conversion rates in the league and he also took 52 shots, 44 from distance and only hit the target eight times. This doesn't mean Bradley isn't a good player, as his passing is one of the best in the league, and arguably his teammates could have converted on some of his key passes. Michael Bradley compares very well to important players in the league like Seattle's Osvaldo Alonso and LA Galaxy's Juninho, although a reason why these players are so successful is they know their roles and stick to it.
Toronto’s biggest failure offensively was their inability to find a true attacking midfielder and player who could be responsible for set pieces. The result being Toronto was one of the worse teams in terms of converting on set play opportunities. Going into the season for me, the answer was Dwayne De Rosario, who the model predicted five goals, six assists, but also 1788 minutes of playing time. De Rosario only ended up playing 508 minutes scoring one goal and played almost more of a forward role, and because of this he had one of the highest rate of shots and shots on target per 90 minutes. His failures this season was partly due to his age, although it seems that for the most part coaches, especially Ryan Nelsen, were unwilling to play him.
Through the summer, Toronto also brought in four players. Luke Moore was the most positive and recorded almost the exact same stats as projected, although with his original club Chivas USA, the key for TFC next year is to making sure Moore is on a similar contract to truly bring value to the club. Dominic Oduro struggled in terms of scoring, but did play well as a winger with Toronto FC in providing opportunities and was also fouled 46 times with more than half of them in dangerous positions. Like Moore, cost would be concern regarding him staying with the club next season. To me, Collen Warner was the most confusing move this season. While he is fair value for money under his current contract, the confusion is mainly because him and Bradley were very similar. Warner also struggled with his shooting and providing balls from set plays and his poor tackling led to him fouling or not closing the play down. In many ways Warren Creaville is just a standard MLSer, and for me there is greater fear of him being wasted on the bench than being a useful player for TFC in the future.
The question remaining is how does a team go from a predicted goal ratio of 52-39 +13 with expectations of the playoffs, to a reality of 44-54 -10 and missing the post season for the eighth straight year? I feel there are a number of reasons. First, goalkeeping was expected to be better and players such as Bloom and Hagglund weren’t expected to get so many minutes. Second, Bradley was supposed to be more effective than the -15 +/- goal differential he ended up having on the season. Thirdly, a better, more concise offense would mean greater ball control and also put games out reach taking pressure off the defense. Lastly, a true attacking midfielder would mean the wingers wouldn't have to be relied on creating chances and instead could have helped concentrate on their defensive duties.
The 0-4-1 October with a -7 Goal Difference made the team much worse stat-wise than the team was most of the season, however, I do feel these issues existed throughout the season. One could argue the firing of Nelsen could have cost TFC a playoff berth, but based on how things looked from the start, the team should have fired or bean to look for an alternative to Nelsen at the beginning of the season or in June when managers from around the World became available. In the end, you had a team that despite its talent was mismanaged in terms of players signed and how they were used. You also had two key strikers, one with off field issues and the other with on field, but both couldn't put the team on their backs like other DP players have done in the past for other MLS clubs.
The expected stats were tabulated last February and this one, along with other models, can evaluate a season as well as its progresses. However, the thing that worries me about Toronto FC for a number of years is what work like this is supposed to provide.
From what I’ve seen in the past eight seasons, right up to present day, is that rarely anyone connected to the club ever has the right answer, or even more importantly knows how to ask the right questions.
|