Pages

Wednesday 31 August 2011

Blasts from the Past.....Newcastle 4-4 Arsenal.

An extraordinary graph with Newcastle's EP spending over 80 minutes at 0.08 or below before leapfrogging Arsenal's EP for the last 7 minutes.This game more than any illustrates that the current EP merely takes you to the current position in the game and the size of that EP is an expression of what happens afterwards ON AVERAGE.

Diaby's 50th minute red card certainly made Newcastle favourites to score more than Arsenal in the remainder of the match ,but the four goal cushion should have been sufficient to enable the Gunners to hang onto the points.The dismissal advanced Newcastle's EP from 0.001 to 0.01 and Arsenal would have won the game from that position 794 times out of 800 and draw it on 5 other occasions.They'd have lost just once.
Countless Geordie fans had already made this mental calculation for themselves and left the ground at half time.

Even when Barton converted a 68th minute penalty the EP barely moved and Best's 75th minute second still saw the respective Newcastle and Arsenal EP at 0.07 and 2.879.

Crucially the clock was now favouring the 10 men visitors and just 5 minutes after Best's strike Newcastle's EP had halved,but Barton's second penalty suddenly made the impossible merely improbable.It had taken 82 breathless minutes to return each teams EP to the values they'd had after Walcott had opened the scoring after a minute.

Tiote's strike was worth just under an EP of 1 (almost 3 and a half times the worth all thee previous goals and the sending off combined) and propelled the home side to marginal EP favourites,but Arsenal held out to ensure that this game wasn't the 1 in 800 when a team gives up a 4-0 half time lead.

Eight goals in total and everyone of them integral to the final result,Tiote's strike claiming the EP plaudits.


EP is just part of the Narrative.

Before we get carried away with the significance and application of EP in soccer,it's worth pointing out the limitations and pitfalls that are out there.A goal is usually a fairly significant event in a relatively low scoring game such as soccer,but it is just one event amongst many that occur during a game.Even the simple ticking of the game clock alters the EP from one minute to the next.So it's worth noting all the event that influence the course of a soccer game and try to quantify their relative importance.

Relative team ability.
The obvious first factor is the strength of each team.Often one is going to be a lot better than the other,even more so if the better team is also the home side.ManU are going to be favoured to beat WBA at the Hawthorns and they're going to be even bigger favourites in the reverse fixture.

Time Elapsed.
It's a fairly trivial problem to work out how many goals on average a team is likely to score against today's opponents at today's venue.It's also fairly easy to work out how that expected goal average decays as the game progresses.Armed with this information the relative win and draw percentages,and hence the EP can be tracked for both teams.

Red Cards.
Relatively infrequent (unless you are Arsenal),red cards can have a massive effect on a team's EP.Timing and current score is obviously a factor,but for example a team being reduced to 1 men just before half time can expect to score around half a goal less and concede half a goal more compared to the card not being shown.

Goals Conceded.
Another obvious contributor.A goal for the opposition not only raises their team's EP,but it also reduces that of their opponents.

So hopefully we can see that EP changes resulting from goal are just part of the story of the game.In relatively uncomplicated games such as smallish margin wins to nil against similar "on the day" strength of opponent,it can give a very accurate picture of how that goal has contributed to the seasons points total for that side.
However,if we use the raw EP from games where the lead is constantly changing then EP numbers become meaningless unless kept in context.

Let's look at last week's game of the week,Hamburger SV v FC Cologne.

Hamburger were relatively strong,home favourites.At kick off they would expect to pick up 1.79 points from this fixture if it could be replayed over and over again.After 10 minutes the game was still 0-0,Hamburger's pre game goal expectancy had reduced slightly,the draw had become slightly more likely,the home win slightly less and their current EP as now about 1.76.Time was elapsing and with the score still tied the home side was accumulating small amounts of negative EP.
Their 11th minute goal catapulted their EP from 1.76 to 2.45.Now a goal to the good Hamburger merely had to "draw" the remaining 80+ minutes of football to record a match win ("winning" the next 80minutes would course also get them a match win).This goal cushion saw their EP gradually rise as the game time elapsed.By the 20th minute it had risen to 2.46.They were now accumulating small amounts of positive "in game" EP.
Cologne then immediately equalised and Hamburger's EP went from 2.46 to 1.73.

And so it continued.Hamburger's EP drifted gently upwards or downwards in the periods of goalless play,while falling or rising precipitously when goals were scored by the sides.

Below is a table showing the amount of EP accumulated by both sides through the different phases of the game.


TEAM
Pre
Game EP.
EP
accumulated through Time Elapsed.
EP
accumulated through Goals Scored.
EP
lost through Goals Conceded.
Total
Game EP.
Hamburger SV
1.79
-0.24
2.72
-4.27
0
FC Cologne
0.96
0.39
4.18
-2.52
3

From these figures it becomes apparent why EP for individual goals can bear little resemblance to the points a player helps to gain for his team over the course of the whole season.EP is very strongly context driven,it can be very informative for individual players when applied to a single game,it shows how big a chance he gave his team to win that match.However,it's also more of a longterm measure of the type of side he's playing for.

Forwards playing for teams with poor defences can expect to accumulate EP seemingly in excess of those who don't.Continually playing from behind gives you greater opportunity to score high EP value goals.Hamburger SV's three goals were worth a combined EP of nearly three,but their points haul at the end of the day was zero.

Similarly a team by virtue of the lead constantly changing hands can score goals which have a combined EP that is in excess of the 3 points available for winning as in the case of Cologne verses Hamburger.

By contrast,Rooney has to score the first goal in a comfortable ManU win against inferior opponents if he wants to score a reasonably large EP.Once the first goal has gone in any subsequent scores will require a goal or two input from the opposition to make them worth much more than an EP of 2 or 3 tenths.

To make a player's goal EP a truely useful statistics this anomalies need to be addressed.Simply averaging a player's EP per goal for the season can be unfair to excellent players who have limited opportunities to score high value EP goals because their team spends large amounts of the game in very commanding positions.It is also plainly illogical that a player can gain more total EP in a game than there are points on offer,as would have been the case if one player had scored all of Cologne's goals against Hamburger.

High EP goals are certainly special goals,but whether that confers anything extra special onto the players who score them is open to debate.Try to look at EP more as a numerical representation of the crowd reaction to the goal.

Sunday 28 August 2011

Individual Goal Values for the EPL 28th August 2011.

WBA 0-1 Stoke.
Stoke's record of 20 wins,8 draws and one defeat against WBA over the last 20 years was added to with a high value last minute strike from a full back playing as a makeshift striker.Tight,low scoring games are always going to lead to high value goals and Shotton's strike,coming as it did in the last minute was one of the highest you'll see all year.




ManU 8-2 Arsenal.
Big margin blowouts for favoured teams rarely produce very interesting graphs as the team EP quickly approaches 3.This game was very slightly more interesting than the usual rout.United led 1-0 when Arsenal won a 27th minute penalty and a successful conversion would have dropped United's EP from 2.67 to 2.04 and raised the Gunners' from 0.2 to 0.7.However,following the early season trend,Van Persie failed from the spot,60 seconds later the Red Devils were 2-0 up and their EP had climbed to 2.92 and Arsenal's was down at 0.05.A converted penalty may not have changed the ultimate result,but the miss led to a massive EP swing compared to what might have been.
Neither Rooney's hat trick,nor Arsenal's two goals registered much on the EP scale and neither team's EP showed any movement for most of the second half. The 4th red card of the season for Arsenal was meaningless in the context of game EP.



Tottenham 1-5 Man City.
Man City were not as big a favourite as their near neighbours,so there was more EP on offer for a comfortable victory. Dzeko's first three goals easily outscored Rooney's hat trick later in the day.Any potential to accumulate  significant EP had vanished by the 55th minute barring a concerted Spurs comeback and that was never on the cards.



Newcastle 2-1 Fulham.
A fairly standard example of a comfortable home win involving two evenly matched teams was briefly interrupted by a small blip bought about by Dempsey's late consolation.



How much a Team's Expected Points were increased by each Premiership Goal.

SCORER TEAM Goal Time
(mins)
Pre Goal
Points Expectation.
Post Goal
Points Expectation.
Increased
Points Expectation.
SHOTTON Stoke 90 1.05 2.81 1.76
BEST Newcastle 48 1.52 2.44 0.92
DZEKO Man City 34 1.40 2.27 0.87
WELBECK ManU 22 2.08 2.67 0.59
DZEKO Man City 41 2.30 2.79 0.49
BEST Newcastle 66 2.53 2.92 0.39
YOUNG ManU 28 2.67 2.92 0.25
DZEKO Man City 55 2.85 2.96 0.11
DEMPSEY Fulham 88 0.01 0.11 0.10
ROONEY ManU 41 2.93 2.99 0.06
ROONEY ManU 64 2.96 3.00 0.04
WALCOTT Arsenal 45 0.006 0.037 0.031
AGUERO Man City 60 2.98 3.00 0.02
KABOUL Tottenham 68 0.001 0.007 0.006
NANI ManU 67 3.0 3.0 0
PARK ManU 70 3.0 3.0 0
ROONEY ManU 82 3.0 3.0 0
DZEKO Man City 90 3.0 3.0 0
VAN PERSIE Arsenal 74 0 0 0

Game of the Week. Hamburger SV v FC Cologne.

The archetypal 7 goal thriller,featuring first half domination by the favoured home team,three changes of lead,massive swings in team EP and finally a late come from behind win for the pre game underdog.





SCORER. TEAM. Goal Time
(mins).
Pre
goal

Points Expectation.
Post
goal

Points Expectation.
Increased

Points Expectation.
PETRIC HAMBURGER SV 11 1.76 2.45 0.69
CHIHI FC COLOGNE 21 0.33 0.98 0.65
NOVAKOVIC FC COLOGNE 49 1.02 2.03 1.01
RAJKOVIC HAMBURGER SV 60 0.52 1.54 1.02
SON HAMBURGER SV 63 1.55 2.56 1.01
CLEMENS FC COLOGNE 84 0.17 1.04 0.87
McKENNA FC COLOGNE 88 1.03 2.68 1.65

Individual Goal Values for the EPL 27th August 2011.

Firstly a brief description of the methodology.

The win probability for each team is tracked on a minute by minute basis throughout each EPL game.At any point in a game a team will have a probability of winning the game outright and an associated probability of drawing the game.By multiplying the probability of winning the game by 3 and the probability of drawing by 1 and adding the results together you will get an expected longterm average for the number of points that team will achieve from that position or Expected Points (EP) for short.

By comparing those totals immediately before and immediately after a goal is scored it is possible to assign a value to that goal.Similar significant events that occur during a game,such as dismissals,goalkeeping saves and referring mistakes can be treated in the same way.
All probabilities account for the respective abilities of both sides.An early goal by a vastly superior side is for example less "valuable" than one scored by a team in a more evenly matched encounter.
        ..............................................................................................................................






















Wigan 2-0 QPR.
A comfortable looking victory as Wigan enhanced their pre game superiority with two Di Santo goals either side of the interval.Although the first strike was worth over twice the second,it was the second goal that virtually guaranteed the three points.This indicates that the raw EP of a goal isn't the whole story.A combination of El Habsi and the woodwork ensured that the game didn't take a very different course on two occasions.


























Blackburn 0-1 Everton.
A much more interesting game than the bare scoreline would indicate.The early season trend to miss your penalties continued apace as first Hoilett and then Fomica failed to score from the spot for Blackburn.As slight pre game underdogs,the Blackburn EP spent most of the game hovering just above one.That EP would have received a healthy,but not decisive boost if the first kick had been converted,but Formica's very late failure was much more costly in terms of potential EP gains.Arteta  predictably scored from Everton's 90 minute penalty,providing an object lesson in converting under pressure and also gaining one of the biggest EP goals it's possible to score,Blackburn's EP crashing from just above 1 to 0.09 in the process.




Swansea 0-0 Sunderland.
Little evidence of excitement from the graphs and Swansea remain goalless after three games.Vorm was by far the more involved of the two keepers and fortunately for the Welsh newboys he looks an excellent buy.A definite candidate for a graph of his own to chart the contribution he's making to Swansea's quest for points.


Aston Villa 0-0 Wolves.
Another fairly grim 0-0.Villa's EP drifted down to 1 from the higher starting point,so Wolves wil be the happier of the two teams as they continue to gather early points to stave of any threat of relegation come May next year.


Chelsea 3-1 Norwich.
A perhaps surprising graph showing Norwich's poor EP for virtually the entire match given that the game was only won late on.However,it is an accurate reflection of the gulf in talent between the sides.The much superior Chelsea side still had about 90 minutes of gametime ahead of them when Bosingwa scored.That supremacy was already factored into Chelsea's pre goal EP,so the Bosingwa goal was a relatively low value strike especially for an opening goal.Holt's equaliser increased Norwich's pre goal EP by almost 600%,but still didn't raise it's raw value above a point.

Liverpool 3-1 Bolton.
Liverpool's first goal gave the game a similar opening profile to their first home game against Sunderland.However,this time it was the Reds who grabbed the second and we ended up with the kind of graph we were seeing when Liverpool were bona fide members of the Big Four.




Expected Points Values for individual goals from Saturday 27th of August.


SCORER TEAM Goal Time
(mins).
Pre
Goal Points Expectation.

Post Goal
Points Expectation.
Increased
Points Expectation.
ARTETA. Everton 90 1.12 2.83 1.71
LAMPARD. Chelsea 82 1.72 2.99 1.27
DI SANTO. Wigan 41 1.43 2.34 0.91
HENDERSON. Liverpool 15 2.20 2.71 0.51
HOLT. Norwich 63 0.10 0.59 0.49
DI SANTO. Wigan 66 2.47 2.90 0.43
BOSINGWA. Chelsea 6 2.56 2.86 0.30
SKRTEL. Liverpool 52 2.71 2.95 0.24
ADAM Liverpool 53 2.95 2.99 0.04
KLASNIC Bolton 90 0.000 0.002 0.002
MATA Chelsea 90 3.0 3.0 0

Friday 26 August 2011

Individual Goal Values for the EPL 20th,21st and 22nd August 2011.


Firstly a brief description of the methodology.

The win probability for each team is tracked on a minute by minute basis throughout each EPL game.At any point in a game a team will have a probability of winning the game outright and an associated probability of drawing the game.By multiplying the probability of winning the game by 3 and the probability of drawing by 1 and adding the results together you will get an expected longterm average for the number of points that team will achieve from that position or Expected Points (EP) for short.

By comparing those totals immediately before and immediately after a goal is scored it is possible to assign a value to that goal.Similar significant events that occur during a game,such as dismissals,goalkeeping saves and referring mistakes can be treated in the same way.
All probabilities account for the respective abilities of both sides.An early goal by a vastly superior side is for example less "valuable" than one scored by a team in a more evenly matched encounter.
        ............................................................................................................................

Sunderland 0-1 Newcastle.
A feisty North East derby that was more interesting for what didn't happen than for what did.The home side were moderate pre game favourites,but Larsson's goalline handball which was missed by World Cup final referee Howard Webb would have changed all that.The red card alone would have improved Newcastle's EP from 1.1 to 2.08 and a successfully converted penalty would have moved it on further to 2.68 points.
Of course none of that happened,but justice was done when Taylor's second half strike advanced Newcastle's EP from 1.15 points to 2.29 and a belated red card sealed victory.























.......................................................................................................................


Norwich 1-1 Stoke.


Another fascinating game statistically speaking.Trailing by a goal after 63 minutes with their EP languishing at 0.34,Stoke won a dubious penalty and Norwich were reduced to 10 men.The dismissal bumped the Potters' EP up to 0.66 and if Walters had scored from the spot it would have increased to 1.84.He missed and it needed a 92nd minute Jones equaliser to boost their then EP by over 1400% and ultimately grab a point.

............................................................................................................................


Arsenal 0-2 Liverpool.
More self inflicted red card woes for Arsenal,whose disciplinary record is atrocious especially when you considered the small amount of time they spend without the ball.The Gunners EP dropped from 1.28 to 0.80 on Frimpong's dismissal.The own goal further reduced it to 0.1 and the game was virtually over before Liverpool actually had to contribute for themselves.































......................................................................................................................


Swansea 0-0 Wigan.


More penalty drama as Swansea welcomed Premier League action and sell out home crowds to the Liberty.The Swans' EP of 1.26 would have crashed to 0.31 had Watson converted Wigan's 73 minute spot kick,but Vorm made the save.










............................................................................................................................


Chelsea 2-1 WBA.
The opening campaign didn't get much easier for WBA.Long's early strike dragged down Chelsea's EP by a considerable amount and raised WBA's just enough to bring the teams close to parity.Chelsea spent over half the game as underdogs before a high value strike from Anelka and a late and equally valuable one from Malouda turned the game on it's head.




..........................................................................................................................

Aston Villa 3-1 Blackburn.
Standard fare both on the pitch and from a statistical viewpoint.

..........................................................................................................................

Bolton 2-3 ManCity.
Silva's opening goal was worth most in this match.It advanced ManCity's EP all the way to 2.5 points at a relative early stage of the game.Subsequent Blues scorers needed Bolton to make a game of it for their goals to be of a similar value and the Trotters couldn't quite oblige.Despite going down by the odd goal in five and being at home,Bolton's EP never managed to rise above 1 and the much superior Man City never saw their's fall much below 2.






















.........................................................................................................................


ManU 3-0 Tottenham.
Little out of the ordinary on show here.Despite the lateness of United's goals they still had a EP of just shy of 2 prior to Welbeck's opener.




























......................................................................................................................


Wolves 2-0 Fulham.
The first goal in this reasonably even contest was always going to be fairly valuable and so it proved.Doyle's goal advanced Wolves' EP from 1.43 to 2.35 and his celebrations cost him a booking.

..........................................................................................................................


Everton 0-1 QPR.
Smith's goal was a high EP value score in a week of high value strikes,flip flopping the pre match favourites in the process.Not a great month for Everton,who are behind in matches played and financially on their uppers.
































.............................................................................................................

A round up of the value of every goal scored this weekend.

SCORERTEAMPre goal
Points Expectation.
Post goal
Points Expectation.
Increased
Points Expectation.
Percentage
Increase
ARSENAL
OG
Liverpool1.262.611.36108.0
MALOUDAChelsea1.572.861.2981.8
ANELKAChelsea0.872.111.24143.4
TAYLORNewcastle1.152.291.14100
JONESStoke0.071.081.011476
DOYLEWolves1.432.350.9264.7
DE LAETNorwich1.382.280.9065.0
WELBECKManU1.862.750.8847.4
SMITHQPR0.711.530.82115.1
SILVAManCity1.832.540.7139.0
AGBONLAHORAston Villa1.942.570.6332.2
LONGWBA0.390.940.55142.3
JARVISWolves2.372.830.4620.0
DAVIESBolton0.160.550.39251.2
BARRYManCity2.552.890.3413.3
DZEKOManCity2.572.900.3413.2
HESKEYAston Villa2.582.890.3112.1
BENTAston Villa2.662.950.2911.1
KLASNICBolton0.070.270.21319.4
ANDERSONManU2.792.980.197.0
PEDERSENBlackburn0.040.230.19422.9
SUAREZLiverpool2.842.990.155.3
ROONEYManU2.993.000.010.19

Individual Goal values for the EPL 13th,14th and 15th August 2011.

Firstly a brief description of the methodology.

The win probability for each team is tracked on a minute by minute basis throughout each EPL game.At any point in a game a team will have a probability of winning the game outright and an associated probability of drawing the game.By multiplying the probability of winning the game by 3 and the probability of drawing by 1 and adding the results together you will get an expected longterm average for the number of points that team will achieve from that position or expected points (EP) for short.

By comparing those totals immediately before and immediately after a goal is scored it is possible to assign a value to that goal.Similar significant events that occur during a game,such as dismissals,goalkeeping saves and referring mistakes can be treated in the same way.
All probabilities account for the respective abilities of both sides.An early goal by a vastly superior side is for example less "valuable" than one scored by a team in a more evenly matched encounter.

                             .......................................................................................................................................


Liverpool 1 - 1 Sunderland.

Liverpool were strongly favoured pregame and started the match with a points expectancy of over 2 points.By the 12th minute their points expectancy had fallen slightly to around 2.1 points,but Suarez's goal bumped it back up to 2.67 points.Sunderland's EP was below 2 tenths of a point until Larsson's superb volley increased it to 8 tenths.Probabilistically Larsson's was just the more valuable of the two.
 





















........................................................................................................


QPR 0-4 Bolton.
Cahill's goal just before the break in a game where Bolton were marginal underdogs was the most valuable of the weekend,bar WBA's own goal.Worth almost of full point,it catapulted Bolton's EP from 1.16 to 2.15.QPR's self inflicted own goal was worth just over half of Cahill's strike.Klasnic scored just early enough to register a goal worth 0.1 of a point,but Muamba's goal was late and of negligible value.

.....................................................................................................

WBA 1-2 ManU.
Rooney's
strike was of similar value to Suarez's score for Liverpool earlier in the day.Such is ManU's ability compared to mid to lower teams like WBA,that Long's equaliser only increased WBA's EP from 0.2 to 0.7.However,each teams EP's were converging towards 1 when Young's deflected cross gave the defending Champions a boost of 1.36 points from 1.44 to 2.8 and ultimately all three points. 











...............................................................................................



Wigan 1-1 Norwich.
Wigan were just moderate favourites against newly promoted Norwich,so Watson's penalty was worth almost 0.8 of a point despite it coming relatively early in the game.Hoolahan's goal was identical in value to Watson's as it dragged the Canaries EP off the floor to just over one.Although Wigan remained probabalistically the most likely winners of the two both PE's converged to 1 at fulltime.

..................................................................................................

Blackburn 1-2 Wolves.
Virtually identical EP changes for the first two goals in this game and the Wigan Norwich encounter.Ward's  game breaker was worth almost a point,indicating the value of a goal in games between closely matched teams.

..................................................................................................


ManCity 4-0 Swansea.
The opening Dzeko goal didn't come until the 57th min,but that didn't make it close to being the most valuable of the weekend.There was still over half and hour to play and such was Man City's pregame superiority over the Welsh side there was still a good chance that there would be goals and the home side would be the team to score them.Welcome to the Premier League!




























...................................................................................................


Stoke 0-0 Chelsea.
No goals,but if Begovic hadn't tipped Anelka's late goalbound header onto the bar Chelsea's EP would have jumped from 1.56 to 2.67 with just 20 minutes to play and Stoke's would have dropped from 0.93 to 0.21.








......................................................................................................


Newcastle 0-0 Arsenal.
No goals here either and  Gervinho's red card was enough to ensure that Newcastle were just about favourites over the closing 14 minutes,but Arsenal 
hung on.






















..........................................................................................................


Fulham 0-0 Aston Villa.  
A dour,uneventful game.Fulham's pregame supremacy drifting gently down to meet with Villa's by full time.                                                                                                                                                                                                            

























...............................................................................................................

A round up of the value of every goal scored this weekend.

SCORERTEAMPre goal
Points Expectation.
Post goal
Points Expectation.
Increased
Points Expectation.
Percentage
Increase.
WBA OGManU1.442.801.3685.6
CAHILLBolton1.162.150.9988.4
WARDWolves1.122.100.9988.4
HOOLAHANNorwich0.351.120.77220.6
WATSONWigan1.602.370.7747.8
FORMICABlackburn1.602.3707647.4
DZEKOManCity2.062.800.7435.9
FLETCHERWolves0.381.110.72188.8
LARSSONSunderland0.180.800.62339.3
ROONEYManU2.112.670.5626.5
SUAREZLiverpool2.122.670.5626.2
QPR OGBolton2.252.790.5424.2
LONGWBA0.200.730.53269.8
AGUEROManCity2.812.980.176.0
KLASNICBolton2.882.980.113.7
SILVAManCity2.983.000.020.5
MUAMBABolton2.993.000.010.2
AGUEROManCity3.003.0000