Tuesday, 8 November 2011

Scoring the First Goal.

Football is in contrast to many other popular sports low scoring and that in part makes it a tense experience to participate in both as a spectator or a competitor.Therefore an opening goal when viewed within the context of a single game is very valuable to the scoring team.We can do a ball park calculation to demonstrate this importance.A game has on average 2.6 goals and lasts 94 minutes,so the first goal will on average be score after about 35 minutes (it's actually later).If we then assume a typical home side would be expected to score 1.5 goals and their visitors 1.1 we can calculate that the Expected Points for the home side if they score first would be 2.4 points and they would avoid defeat in 92 games out of 100.Even the visitors would return with points over 80 times per hundred games should they draw first blood and they'd average about 2 points per game where they took the lead.

Hajduk Split kick off against Stoke.
Having established that opening goals are hugely influential in individual games our next step is to try to come to a method that will indicate which team is more likely to score first.Probably one of the most famous opening goals was scored by San Marino against England in 2003.Although it didn't materially affect the result it was a surprising enough event to suggest that weaker teams are less likely to score first.We can test this intuitive assumption by collecting a large enough sample of games and record the identity of the first team to score and the pre game goal expectancies for each team.If we do this we find that in games where a first goal was scored the proportion of a team's pregame goal expectancy compared to the total goal expectancy for the game is near enough equal to the proportion of first goals that that team scores.

An example is probably in order.If a team would expect,on average to score 2 goals and concede 0.8 goals in a match or score 71.4% of the goals,it will also score about 71.4% of the opening goals in that matchup where a goal occurs.

Therefore we can say that scoring the first goal is a talent based skill,but the case of San Marino and England also indicates that short term outcomes and insufficiently large sample sizes can lead to less likely streaks appearing.If we imagine for a minute that the EPL is a coin flip league with no home advantage and each team has a 50% chance of scoring first in all 38 games.If we simulate 38 coin tosses and repeat it 20 times for each EPL team we would quickly find a team that appeared to score first much more frequently than 50%.But as we've used the same "coin" for each sequence we couldn't have found a team who specialised in scoring the opening goal,we've just hit an unusual sequence partly through a team's skill and partly through pure chance.Of course the EPL isn't a coin flip league,but it is still prone to short term streaks being mistaken for long term trends.

If we use pre game goals expectations to calculate the proportion of first goals the EPL teams were expected to score over a season and realise that 38 games are likely to produce sequences of games that have been more heavily influenced by good fortune that others,we can start to see which teams may have occupied false finishing positions in the final table.We saw here  that it was likely that Ipswich were lucky to score first as often as they did in their penultimate EPL season and their inflated finishing position of fifth was more of a mid to lower half team who got lucky,than a bona fide 5th place finish.A fact confirmed by their relegation the following season.

A Team's Likelihood of Scoring a Game's Opening Goal Compared to their Actual Record.2008-11.

TEAM. SEASON. % of Games where the Team Scored First. Predicted First Goal Scoring %. % Over
Points per
Arsenal 2008/09 64 67 -5.3 1.89
Arsenal 2009/10 69 69 0.7 1.97
Arsenal 2010/11 63 67 -6.6 1.79
AVilla 2008/09 49 57 -17.5 1.63
AVilla 2009/10 62 58 6.1 1.68
AVilla 2010/11 56 51 8.2 1.26
Blackburn 2008/09 39 47 -19.3 1.08
Blackburn 2009/10 42 46 -9.8 1.32
Blackburn 2010/11 54 45 17.1 1.13
Bolton 2008/09 44 44 0.3 1.08
Bolton 2009/10 47 44 6.8 1.03
Bolton 2010/11 46 46 -0.1 1.21
Chelsea 2008/09 74 72 2.1 2.18
Chelsea 2009/10 76 74 2.2 2.26
Chelsea 2010/11 53 70 -32.6 1.87
Everton 2008/09 61 53 12.5 1.66
Everton 2009/10 56 56 -0.8 1.61
Everton 2010/11 47 56 -18.6 1.42
Fulham 2008/09 53 46 13.8 1.39
Fulham 2009/10 36 45 -23.8 1.21
Fulham 2010/11 49 48 1.0 1.29
Hull 2008/09 31 42 -37.6 0.92
Hull 2009/10 33 39 -17.1 0.79
Liverpool 2008/09 67 69 -3.4 2.26
Liverpool 2009/10 70 69 1.0 1.66
Liverpool 2010/11 65 62 4.4 1.53
ManUtd 2008/09 83 72 13.1 2.37
ManUtd 2009/10 68 71 -5.1 2.24
ManUtd 2010/11 76 70 8.4 2.11
ManCity 2008/09 58 54 6.7 1.32
ManCity 2009/10 49 62 -27.7 1.76
ManCity 2010/11 74 66 10.2 1.87
Middlesbro' 2008/09 44 45 -2.0 0.84
Newcastle 2008/09 29 46 -56.5 0.89
Newcastle 2010/11 51 48 6.7 1.21
Portsmouth 2008/09 41 48 -16.6 1.08
Portsmouth 2009/10 26 40 -55.4 0.74
Stoke 2008/09 44 38 13.9 1.18
Stoke 2009/10 59 44 25.9 1.24
Stoke 2010/11 38 46 -21.6 1.21
Sunderland 2008/09 50 45 10.0 0.95
Sunderland 2009/10 53 46 12.9 1.16
Sunderland 2010/11 50 47 6.0 1.24
Spurs 2008/09 45 54 -18.8 1.34
Spurs 2009/10 69 63 8.2 1.84
Spurs 2010/11 44 59 -33.7 1.63
WBA 2008/09 29 40 -40.1 0.84
WBA 2010/11 32 46 -45.6 1.24
WHU 2008/09 50 46 8.0 1.34
WHU 2009/10 40 46 -15.1 0.92
WHU 2010/11 39 44 -13.2 0.87
Wigan 2008/09 47 47 -0.2 1.18
Wigan 2009/10 47 45 4.7 0.95
Wigan 2010/11 37 41 -10.4 1.11

As we can see within just 38 games there is considerably variation between a team's most likely rate for opening the scoring and their actually real life figures.Chelsea dropped over 30% below the figure you would have expected for a team of their ability in 2010/11.They weren't quite the side they had been in the previous two seasons,but by failing to score first in just under half of their games,they saw their points per game total fall below 2.They were closer to 4th than they were to 1st.In short Chelsea partly through luck found themselves trailing more often than previously and because "luck" has no obligation to immediately compensate them once the goal had been scored,they almost finished 4th.So far this year they've enjoyed the kind of scoring record more akin to their ability,netting the first goal 70% of the time one has been scored and are gaining 2 points per game,although sample size is an obvious caveat.

The sample size issue is also illustrated if we combine the three season records of the original Big Four,Liverpool,Chelsea,ManUtd and Arsenal to produce a sample equivalent of 12 seasons.If Chelsea were unlucky when it came to awakening the vidiprinter last year,fortune smiled on United in 2008/09 when they pipped Liverpool by 4 points.However,overall the Big Four combined for the last 3 seasons were expected to score 69% of first goals in non goalless games and that's just what happened.

Spurs were possibly fortunate to grab their Champions League spot in 2009/10.Random acts of elevated 1st goal scoring gave them a similar actual first goal percentage to rivals Liverpool despite Spurs arguably being inferior to the Reds.Second city rivals,Man City were probably as good a side as Spurs,but unlike them they underperformed in the 1st goal stakes.Tottenham grabbed the last Champions League spot by 3 points.

If luck plays a part in deciding the title,it can have much more serious consequences around the foot of the table.As Ipswich discovered,if your base talent isn't as high as you think,a few bad or simply neutral bounces can see you relegated from a previous position of apparent strength.Newcastle and WBA have been relegated in seasons where they could have reasonably expected to score first more frequently than they did,they were probably better than their points haul indicated and both maintained enough quality to return to the top flight.The same could well be true at West Ham.

Alarm bells should have been ringing pre season at Blackburn,who scored the opener in 54% of games in 2010/11 compared to an expectation of 45%.There is of course no guarantee that they won't get close to last year's figure this term,but the likelihood is that with little change in squad quality they will trend towards a lower percentage and find themselves at risk of relegation.So far this season they have just over 0.5 points per game and have scored the opener in only 27% of games.Opening scoring percentage correlates very well with points per game and a scoring % of 36 or below over a season puts a team on or below the magic 40 points mark that separates the EPL survivors from the dispossessed.

Graph of Points per Game verses Frequency of which a Team Scores the First Goal.

A team's opening goal record does not define a season,but it can hide frailties or abilities that reveal themselves in subsequent seasons.Ipswich should have toasted their 5th spot,but prepared for the next campaign as though they had finished 15th.When team management comes to decide how much squad strengthening is needed to maintain standards or push on to greater things,a quick look at the previous year's table is rarely sufficient.

No comments:

Post a Comment