EPL 3-4 October 2015

EPL 3-4th October 2015

A brief look at the shot statistics which underlay the weekend's games:

In [1]:
%matplotlib inline
import league_analysis
from IPython.display import display, HTML
epl = league_analysis.epl
league_analysis.display_matches(epl, '3/10/15', '4/10/15')
Home Away
Team Aston Villa Stoke
Goals 0 1
Shots 13 12
SOT 2 5
Home Away
Team Bournemouth Watford
Goals 1 1
Shots 12 11
SOT 6 2
Home Away
Team Chelsea Southampton
Goals 1 3
Shots 10 13
SOT 3 5
Home Away
Team Crystal Palace West Brom
Goals 2 0
Shots 21 7
SOT 8 2
Home Away
Team Man City Newcastle
Goals 6 1
Shots 23 6
SOT 11 4
Home Away
Team Norwich Leicester
Goals 1 2
Shots 17 16
SOT 2 5
Home Away
Team Sunderland West Ham
Goals 2 2
Shots 12 20
SOT 3 6
Home Away
Team Arsenal Man United
Goals 3 0
Shots 12 9
SOT 5 5
Home Away
Team Everton Liverpool
Goals 1 1
Shots 14 13
SOT 4 4
Home Away
Team Swansea Tottenham
Goals 2 2
Shots 11 19
SOT 2 11

Wins for Manchester City and Southampton

These were, by shot counts at least, largely deserved victories, if perhaps a little flattering on the scoreline. Southampton have been doing fine all season and not quite being rewarded, so this win does not come as such a large surprise as it might have done at the start of the season, even disregarding Chelsea's poor form.

Southampton

Chelsea have been notably, and widely discussed, performing worse than last season. So how well did Southampton do to out-shoot Chelsea?

In [2]:
epl.compare_average_team_games('Chelsea', 'Southampton', '03/10/15')
Game Type Shots For Shots Against SOT For SOT Against
This Chelsea game 10 13 3 5
Avg Chelsea game 15.625 13.875 4.5 6.375
Avg Southampton game 16.125 9.625 5.75 2.5
Avg Chelsea home game 17.25 13.25 5.5 5.75
Avg Southampton away game 14.0 10.0 4.75 2.0

So, Chelsea had over 5 shots fewer than is typical for them this season, but 10 shots is the closest integer to the 9.6 shots Southampton have generally allowed and bang on for how many Southampton allow when playing away from home. Southampton had 13 shots, 3 fewer than their average per game, but also just under the 13.9 shots Chelsea have generally allowed.

Sort of similar shots on target statistics, Chelsea had fewer than they would normally have, but a little more than the average Southampton have allowed. But Southampton had 5 shots on target which is lower than both their average per game and lower than the 6.375 Chelsea allow on average. Even if we look only at Chelsea's home games, they are allowing 5.75 shots on target against them at home.

What all of this tells us is that this was an entirely expected performance from both teams. The fact that such an average performance arguably justifies a Southampton win at Stamford Bridge simply confirms that Chelsea are, so far this season, not great. In contrast, Southampton are, so far this season, looking good to at least repeat last season's 7th, if not better.

I stress that there is a long way to go, but at this point, it wouldn't seem unreasonable, to me at least, to suggest that Southampton will finish above Chelsea.

Manchester City

An obviously pretty pleasing victory after two league defeats on the trott, however much those defeats may have been unwarranted. Again comparing against the 'average' game between these two sides:

In [3]:
epl.compare_average_team_games('Man City', 'Newcastle', '03/10/15')
Game Type Shots For Shots Against SOT For SOT Against
This Man City game 23 6 11 4
Avg Man City game 20.5 9.125 7.875 2.875
Avg Newcastle game 8.375 18.125 3.0 6.25
Avg Man City home game 21.5 7.25 8.0 2.5
Avg Newcastle away game 7.75 19.75 2.5 7.0

So, in other words, Newcastle are a bit worse than the average team against Manchester City and Manchester City are a bit better than the average team that plays Newcastle. Well considering these teams are top and bottom of the league that is about as expected. Generally speaking Newcastle need to change something and quick, otherwise it is the division that will be changing from underneath them.

Arsenal 3-0 Manchester United

Well, on the face of this, it would seem a bit of a flattering result for Arsenal. It quite possibly was, but this game more importantly displays one of the major limitations of shot statistics. This is, that it does not take into account game-state. Arsenal scored with their first three shots on target, which is certainly arguably fortunate. However, it does render the rest of the game less representative of the two teams' strengths.

Usually we expect a team that scores early to be under-represented by their shot statistics, because, essentially, there is less urgency on them to shoot. Their bigger concern is preventing shots from the opposition. This could work in either direction, it could be that Manchester United would have typically had more shots, but were effectively shut out by a team with little offensive aspirations. This is arguably particularly true of a team that profits from fast counter attacks.

In any case, multiple early goals are a bit like early red cards, in that it renders the rest of the game as mostly useless in comparing the strengths of the two teams.

Swansea 2-2 Tottenham

Tottenham are doing just fine. They will be disappointed not to take all three points, but they had 11 shots on target and Swansea scored with their only 2. Last week we determined that there had only been three games so far in which one team had taken 12 or greater shots on target, but all three of those games involved that team winning. First off, all matches in which one team has taken 11 or more shots on target so far:

In [4]:
matches_so_far = league_analysis.get_matches(epl, '01/08/2015', '04/10/2015')
high_sot_matches = (m for m in matches_so_far if m.HST >= 11 or m.AST >= 11)
league_analysis.display_given_matches(high_sot_matches)
display(HTML('There have been {0} matches in total'.format(len(matches_so_far))))
Home Away
Team Arsenal Stoke
Goals 2 0
Shots 29 9
SOT 12 4
Home Away
Team Leicester Arsenal
Goals 2 5
Shots 16 26
SOT 7 12
Home Away
Team Liverpool Aston Villa
Goals 3 2
Shots 21 10
SOT 12 4
Home Away
Team Man City Newcastle
Goals 6 1
Shots 23 6
SOT 11 4
Home Away
Team Swansea Tottenham
Goals 2 2
Shots 11 19
SOT 2 11
There have been 80 matches in total

So only 5 in 80 games, two of which coming this week. A couple of observations:

  • In the other 4 games, the team taking 11 or more shots on target won
  • In the other 4 games, opponents had 4 or more of their own shots on target
  • 3 of the other 4 games had the opponents Stoke, Villa and Newcastle all in the bottom 5 for sot against, the only other game, had the opponents Leicester who at least managed 7 of their own sots.

So all-in-all a very good chance creating game for Tottenham that they probably deserved to win. As a side note Tottenham fans will also like to see Chelsea joint worst for shots on target against.

In [5]:
league_analysis.display_statistic_rankings(epl, 'sot_against')
Position Team sot_against
1 Chelsea 51
- Sunderland 51
3 Newcastle 50
4 Stoke 47
5 Aston Villa 42
- West Ham 42
7 Crystal Palace 41
- West Brom 41
9 Arsenal 39
- Norwich 39
11 Leicester 38
12 Everton 37
- Swansea 37
14 Watford 32
15 Man United 30
16 Tottenham 29
17 Liverpool 26
18 Bournemouth 24
19 Man City 23
20 Southampton 20

Everton 1-1 Liverpool

Honours even seems fairly reasonable. Whoever takes over at Liverpool though will be taking on a side with some pretty reasonable looking numbers. As the table above shows only 3 teams have allowed fewer shots on target. Liverpool's shot ratios are beginning to stabalise in about 4th. Everton fans should probably be a little more worried even if they are currently ahead on points.

In [6]:
after_game_no_dicts = league_analysis.collect_after_game_dicts(epl, '01/08/2015', '04/10/2015')
league_analysis.plot_changing_stats(epl, after_game_no_dicts, 'tsr', rankings=True,
                                    teams=['Liverpool', 'Everton'])
league_analysis.plot_changing_stats(epl, after_game_no_dicts, 'sotr', rankings=True,
                                    teams=['Liverpool', 'Everton'])
league_analysis.plot_changing_stats(epl, after_game_no_dicts, 'points', rankings=True,
                                    teams=['Liverpool', 'Everton'])

Palace and Leicester

Palace created a lot of shots with a decent number on target against West Brom and won fairly convincingly. Leicester, less so and in particular Norwich did manage to create a lot of shots, they just did not get them on target. Worth looking at how the shots statistics have changed for these four teams:

In [7]:
teams = ['Leicester', 'Norwich', 'West Brom', 'Crystal Palace']
league_analysis.plot_changing_stats(epl, after_game_no_dicts, 'tsr', rankings=True, teams=teams)
league_analysis.plot_changing_stats(epl, after_game_no_dicts, 'sotr', rankings=True, teams=teams)

So Norwich lead out of these four teams in total shot ratio but drop to third for the ratio of shots on target. Crystal Palace are the opposite, they are third in the total number of shots but do better in shots on target ratio. Helped, it must be said by this previous game against West Brom who are seemingly worse at creating chances than the other three teams. It must be said Tony Pulis teams have defied shot numbers in the past.

Aston Villa 0-1 Stoke

Arguably similar to the Leicester game, Villa have created more shots but fewer on target than Stoke and Stoke have gone on to win. I'd be more worried about Aston Villa than Norwich. Norwich do seem to be creating chances, but Villa not so much. Both teams only managed 2 shots on target at home this week, but that seems to be more the norm for Villa than for Norwich, who have generally had 10 more shots on target than Villa.

In [8]:
league_analysis.display_statistic_rankings(epl, 'sot_for')
Position Team sot_for
1 Man City 63
2 Arsenal 58
3 Tottenham 52
4 Southampton 46
5 Leicester 41
- Swansea 41
- Liverpool 41
8 Crystal Palace 40
9 Chelsea 36
10 West Ham 35
- Norwich 35
12 Man United 33
13 Everton 32
- Bournemouth 32
- Stoke 32
16 West Brom 26
17 Aston Villa 25
18 Newcastle 24
- Watford 24
20 Sunderland 23

Bournemouth and West Ham

Both teams could feel a little hard done by not to have won. West Ham did go down two goals so will probably feel okay with a draw, particularly since it involved a Sunderland red card. West Ham's numbers are generally not great, and they are over-performing in terms of points (they have more points than we would expect based on their shot totals). But it does not take much to beat Sunderland this season.

Comments

Comments powered by Disqus