Monday, June 29, 2015

What Wins Games (Part 1)

In the last possession article, I concluded that possession is pointless although there was a small sample size. Recently, I built a scraper to scrape the match stats from every MLS match this season. The results point towards the same conclusion.



Possession
Win % with 55% or more possession27.03%
Win % with 45% or less possession41.67%

So if not possession, they what? Shot Dominance, the ratio of a team's shots to its opponents' shots and Shots on Target Ratio, the ratio of a teams shots on target to the total shots on target are also tools used to evaluate match outcomes. The results of the tests of these two metrics are below



Possession
Win % with 2.5X or more shot dominance58.33%
Win % with .4x or less shot dominance21.43%



Possession
Win % with .75 or more STR70.59%
Win % with .25 or less STR7.89%

In conclusion, STR is the best predictive metric available readily on MLS Match Reports.

Analyzing the Amarikwa-Harden trade

Recently, San Jose and Chicago exchanged Ty Harden and Quincy Amarikwa, which looks to be a good deal for both sides. In every trade, there does not have to be a direct winner and loser: on the surface, Chicago recieved center back depth while San Jose acquired a starting forward. And salary wise, both are on the cheap: Amarikwa earned $60,000 guaranteed in 2014; Harden, $71,665.

First, the Whoscored and Squawka Indexes:


Harden(W)Harden(S)Amarikwa(W)Amarikwa(S)Adailton(W)Adailton(S)Jahn(W)Jahn(S)
20156.94736.56566.811756.59-41
20146.682117.02256----5.86-56
20136.98756.48110----6.60122

Harden seems like he is around Adailton's level. Amarikwa seems to be an inconsistent performer, playing excellent in 2014 but okay in 2013 and 2015. However, he seems better Jahn. Jahn twice achieved negative ratings and is only slightly ahead of Amarikwa this year. Amarikwa will eventually start performing better and Jahn worse as they regress towards the mean.

Another tool for player evaluation is radars, pioneered by Ted Knutson. Here are this season's comparisons for the Amarikwa and Jahn below:

Amarikwa in Red, Jahn in Blue


The radars paint a wholesome picture. Amarikwa is better in passing and dribbling, having more key passes and a higher passing accuracy. However, Jahn contributes more defensively, and is dispossessed less. But since both seem to be having off years, here is the comparison for their best years (2014 for Amarikwa, 2013 for Jahn)
2014 Amarikwa in Red, 2013 Jahn in Blue
The same trends hold true: Jahn with good defensive work and ball control and Amarikwa with better passing. Note that Jahn played less than half of the minutes Amarikwa played - 1001 vs 2550.

Saturday, February 21, 2015

The Correlation between Expansion Draft success and actual performance

This year, people seemed to be all over Orlando City's seemingly mediocre Expansion draft. But the folks in Orlando beg to differ - according to the Orlando Sentinel, "While studying past expansion drafts, Orlando City leaders saw the most successful clubs had focused more on filling out the depth within a roster."

So this gave me the idea - did teams previously use the expansion draft for depth? How many starters were recruited via the expansion draft ? Do good expansion draft picks lead to first year success? I have summarized my findings in the table below, but before that couple of points: First, if a player recruited via expansion draft has played more than 2000 minutes, I consider him to be a starter; if a player has played more than 1000 minutes, I consider added depth. Any player who has not played 1000 minutes, is probably waste of a draft pick.

Note: I've used PPG as a measure of success because of the number of games in a season varies


YearTeamTotal minutes# of players who
played 1000+ mins
# of players who
played 2000+ mins
PPG
2005Real Salt Lake~7000610.63
2007Toronto354000.83
2008San Jose8964521.10
2009Seattle9412331.57
2010Philadelphia12146531.03
2011Portland4211211.24
2011Vancouver6519300.82
2012Montreal9633521.24

Clearly, not many starters were recruited via expansion draft, in fact not many depth players either - only 4 (out of 8) expansion teams were able to recruit 5 or more of their depth players via expansion draft.

Seattle was clearly the most successful team, recruiting three starters via expansion draft - James Riley, Nate Jaqua and Brad Evans. Philadelphia, on the other hand, was just middle of the pack in terms of PPG, though they also recruited 3 starters -  Stefani Miglioranzi, Jordan Harvey and Alejandro Moreno. Philadelphia recruited depth players too from the draft; they were clear leaders in terms of minutes played by their expansion draft recruits. Portland managed to find only one starter and just one more depth player, still they did pretty well in their first year, second behind Seattle and tied with Montreal, who did a decent draft (2 starters and 3 depth). So, clearly, recruiting via expansion draft is not the deciding factor of putting together a new team.

Seattle depended on overseas recruitment for their starters including prominent stars as Freddie Ljungberg and Fredy Montero and used their USL team as well for a few starters like Sebastian LeToux. Seems Orlando also believes in that strategy with signings like Kaka and Bryan Rochez in addition to USL star Kevin Molino.

Wednesday, January 14, 2015

5 year performance of MLS teams

Here is a table I made to rank the overall team performance of all MLS teams over the last five years. After some deliberation, I included Portland, Vancouver and Montreal. Even though they have not played all five years, at least all of them played more than 100 matches. Due to the inclusion of the three, I used points per game to evaluate the teams' performance.

TeamPoints 2010+11+12+13+14Points per game
LA Galaxy [1]59 + 67 + 54 + 53 + 61 = 2941.77
Seattle [2]48 + 63 + 56 + 52 + 64 = 2831.70
Real Salt Lake [3]56 + 53 + 57 + 56 + 56 = 2781.67
New York [4]51 + 46 + 57 + 59 + 50 = 2631.58
Sporting Kansas City [5]39 + 51 + 63 + 58 + 49 =2601.57
Columbus [6]50 + 47 + 52 + 41 + 52 = 2421.46
Dallas [7]50 + 52 + 39 + 44 + 54 = 2391.44
San Jose Earthquakes [8]46 + 38 + 66 + 51 + 30 = 2311.39
Houston [9]33 + 49 + 53 + 51 + 39 = 2251.36
Portland [10]X + 42 + 34 + 57 + 49 = 1821.34
Chicago [11]36+ 43 + 57 + 49 + 36 = 2211.33
Colorado [12]46 + 49 + 37 + 51 + 32 = 2151.29
Vancouver [13]X + 28 + 43 + 48 + 50 =  1691.24
Philadelphia [14]31 + 48 + 36 + 46 + 42 = 2031.22
New England [15]32 + 28 + 35 + 51 + 55 = 2011.21
DC United [16]22 + 38 + 59 + 16 + 59 = 1941.17
Montreal [17]X + X + 42 + 49 + 28 = 1191.17
Toronto [18]35 + 33 + 23 + 29 + 54 = 1741.05
Chivas [19]28 + 36 + 30 + 26 + 33 = 153.92

A few observations:
  • MLS is known for its parity and it shows. The top team has got about 1.7 times more points per game than the worst time. For the Premier League, it is about triple - the champion gets about 90 points where as the bottom team gets around 30
  • The top two teams are no surprise - they each have 3 DPs, including superstars like Keane, Dempsey, Donovan etc... However, Toronto landed at the bottom of the table. Lesson: It is not the money that counts but the smart money.
  • A relatively low budget team RSL has done surprisingly well to take the third spot. More over, they have shown remarkably well in terms of consistency: in their best year, they got 57 points, but their worst year they got 53 points, less than 10% down. Additionally, they got 56 points in each of the other three years.
  • DC United and San Jose Earthquakes on the other hand are most inconsistent teams. Both had very good year as well as very bad years. In their bad years, they got less than half of their good year points totals.
  • Most teams alternate between good and bad years
  • However, there are two teams that show year-over-year improvement for last 4 years: New England and Vancouver.

Saturday, January 10, 2015

Superdraft rankings of MLS teams, 2015 edition

As draft day 2015 nears, I thought I should update my last year's post on MLS teams' performance in the Superdraft. This time, while ranking the teams, I've taken into consideration of the number of first round picks the teams used to give teams with a low number of draft picks an even playing field. Otherwise, my basic methodology remains the same - to count the number of draft picks who've played 1000+ minutes in their first two seasons. There are six 2014 rookies who has played more than 1000 minutes in their first season itself, they have been included. The rational for the 1000 minute rule can be found at my earlier post here. Players will count for whatever team drafted them, regardless of trades, as this is about recruitment not development.

I included Portland, Vancouver and Montreal but did not rank them since there is not enough data for them, though Vancouver has done better than the other new entrants.

Like last time, New England and DC have remained as the top two teams in recruiting via Superdraft, while Philadelphia and Chicago has fallen behind - Philadelphia hasn't recruited a single player in the last two years, who met my 1000 minute criteria. Galaxy has done surprisingly well considering that they got low draft picks, except for 2009. Colorado have had the most improvement over the last two years, drafting Deshorn Brown, Dillon Powers, and Jared Watts. Also, Seattle and Columbus remain at the bottom of the draft recruitment rankings.

TeamNumber of draft picks who played
1000+ minutes in first two seasons
 Number of first round picks used  (2009-2013) Three players who clocked most minutes
(first two year's minutes)
Chicago  [B]             4                3 Austin Berry (5580)
Jalil Anibaba (5441)
Sean Johnson (3690)
Chivas             6                6Ben Zemanski (3599)
Zarek Valentin (3275)
Michael Lahoud (2556)
Colorado [C]             4                4Dillon Powers (2457)
Deshorn Brown (2169)
Tony Cascio (2222)
Columbus [D]             4                6Rich Balchan (1705)
Dilly Duka (1624)
Justin Meram (1475)
DC United [A]             8                5Perry Kitchen (5890)
Nick DeLeon (4087)
Chris Korb (3341)
Dallas  [B]             6                6Matt Hedges (4815)
Zach Loyd (4496)
George John (3300)
Houston [B]             5                4Will Bruin (4731)
Warren Creavalle (2504)
Kofi Sarkodie (1804)
LA Galaxy [A]             8                5Omar Gonzalez (5192)
A.J. DeLaGarza (3061)
Michael Stephens (3005)

Montreal

             1                3Eric Miller (1646)
Blake Smith (319)
New England [A+]            10                6Andrew Farrell (5653)
A.J. Soares (5075)
Kelyn Rowe (4630)
New York [B]             5                4Tim Ream (5670)
Tony Tchani (1722)
Ryan Meara (1620)
Philadelphia [B+]             7                5Raymon Gaddis (4278)
Michael Farfan (4262)
Zac MacMath (3555)
Portland             2                3Darlington Nagbe (4428)
Andrew Jean-Baptiste (2568)
Real Salt Lake [C]             4                4Chris Schuler (1728)
Collen Warner (1566)
Sebastian Velasquez (1537)
San Jose Earthquakes [C]             4                4Steven Beitashour (2295)
Brad Ring (2035)
Ike Opara (1399)
Seattle  [D]             3                4Steve Zakuani (4278)
Alex Caskey (1093)
Eriq Zavaleta (1354)
Sporting Kansas City [C]            4                5CJ Sapong (4306)
Teal Bunbury (3431)
Matt Besler (3252)
Toronto [B]            8               7Luis Silva (4043)
Joao Plata (2391)
Aaron Maund (1217)
Vancouver            4               5Jeb Brovsky (3742)
Erik Hurtado (2294)
Darren Mattocks (2211)