Quantifying Delap's Long Throws.

One of the central ideas behind  the "Moneyball" approach to running a successful sports team is the identification of undervalued player skills that can contribute greatly to winning matches and then purchasing these skills cheaply and creating strategies that help your team be successful.It can be argued that one of the most high profile and productive examples of such a strategy  occurred in the English Premiership when Stoke manager,Tony Pulis secured English born,former Irish international,Rory Delap from Sunderland on a free transfer.

The Prelude to a low,flat delivery

Delap's career started at Carlisle,before he spent four seasons at Derby.Throughout his career he was never a prolific goalscorer and the eight EPL goals he claimed to make him leading scorer for the Rams in 1999/00 was more a reflection on their lack of firepower than his scoring prowess.Never the less,a season later he made his one and only big money transfer move when Southampton paid in excess of £4 million for his services.Once again his employers received good,honest work rate in return for their outlay and a goal a season for five seasons.However,even with such a mediocre scoring rate,Delap still managed to attract the attention of the wider footballing public by scoring the best overhead bicycle kick goal of all time against Spurs in 2004,eclipsing even Rooney's effort nearly a decade later.

A short and unproductive free transfer to Sunderland saw him surplus to a new manager's requirements and within just three seasons he found himself wearing his third different set of red and white stripes at Stoke.Firstly on loan,but with a view to a permanent switch.When a 51/49 touchline collision,ironically against Sunderlnd saw him sidelined for the year only 95 minutes into his loan period,Stoke could have easily turned their back on the proposed deal and sent their stricken loanee back to Wearside,but instead the Potters immediately indicated their intent to honour the deal and sign Delap.Their actions were widely interpreted as an example of old world "word is my bond" business ethics that had led Stoke's owner to create two hugely successful enterprises,with the benefit of hindsight a determination to secure a rare,if not unique talent was also at the heart of the deal.

During a dozen seasons of his professional career, Delap's longthrows had produced one goal at Carlisle,one Derby and three at Southampton,but his move to Stoke was about to take the tactic to whole new levels of production.The Potters hadn't yet narrowed their playing area to maximize Delap's delivery,but he bettered his career assists total in his first full season for his new club.Six of his eight assists broke the deadlock in games,three came away from the Britannia Stadium and Stoke were unbeaten in all eight matches.His throw assisted scores added over six Expected Points to Stoke's points haul in their promotion year and while Expected Points are a partly abstract concept (it quantifies by how much the goal has advanced a team's chances of securing either a win or a draw),it seems very likely that Stoke would have fallen into the playoffs without Delap's contribution.

Expected Points Value of Goals Scored from Delap Throws,2007/08 Championship Season.

Opponent.	Expected Points Pre Goal.	Expected Points Post Goal.	Pre Goal Scoreline.	Increase in Expected Points.	Game Result.
Plymouth	0.583	1.572	1-2	0.99	W 3-2
QPR	1.812	2.477	0-0	0.67	W 3-1
@ Plymouth	1.290	2.071	0-0	0.78	D 2-2
Hull	1.766	2.531	0-0	0.77	D 1-1
Preston	1.837	2.515	0-0	0.68	W 3-1
Scunthorpe	0.352	0.951	0-2	0.60	W 3-2
@Norwich	1.264	2.337	0-0	1.07	W 1-0
@Colchester	2.060	2.627	0-0	0.57	W 1-0
Expected Points added per Appearance	0.14	Total Expected Points added.	6.12	Expected Points added per assist.	0.76

Delap's long throws became an international phenomenon on Stoke's return to the Premiership.A narrow but legal pitch combined with one high profile scalp in Arsenal ensured that the tactic remained on the back pages.A high leverage winning assist in Stoke's first home game of the season turned a single home point into a virtually assured three and settled the nerves of a team who had been comprehensively outclassed at Bolton on opening day.The win loss record when Delap provided an assist was slightly inferior to the previous year in the Championship,but there were two high leverage winning assisted goals at home to fellow relegation candidates,Sunderland and Middlesbro'.In terms of keeping Stoke competitive during a difficult first season Delap's throws were invaluable,especially at home.

Expected Points Value of Goals Scored from Delap Throws,2008/09 Premiership Season.

Opponent.	Expected Points Pre Goal.	Expected Points Post Goal.	Pre Goal Scoreline.	Increase in Expected Points.	Game Result.
Aston Villa	1.004	2.816	2-2	1.81	W 3-2
Eveton	0.079	0.366	0-2	0.29	L 2-3
@Portsmouth	0.235	0.846	0-1	0.61	L 1-2
Sunderland	1.261	2.531	0-0	1.27	W 1-0
Arsenal 1st goal	0.605	1.298	0-0	0.69
Arsenal 2nd goal	2.122	2.772	1-0	0.65	W 2-1
Bolton	2.528	2.924	1-0	0.40	W 2-0
Middlesbro'	1.248	2.744	0-0	1.50	W 1-0
Expected Points added per Appearance	0.21	Total Expected Points added.	7.22	Expected Points added per assist.	0.90

The 2009/10 season marked the first signs of a steady decline in the effectiveness of the tactic.Five additional assists in Stoke's run to the FA Cup quarter finals partly masked the decline in productivity in the league matches.The value of the goals in the context of the scoreline when they occurred decreased markedly from the previous two years as did Delap's assisted Expected Point added per appearance.Teams actively prepared for the challenge posed by the longthrow,advertising boards were moved closer to the pitch resulting in just one assist in the league away from the Britannia and referees began to pay more attention to the aerial bombardment that opposing keepers faced.

Expected Points Value of Goals Scored from Delap Throws,2009/10 Premiership Season.

Opponent.	Expected Points Pre Goal.	Expected Points Post Goal.	Pre Goal Scoreline.	Increase in Expected Points.	Game Result.
Burnley	2.516	2.873	1-0	0.36	W 2-0
Wolves	2.502	2.878	1-0	0.38	D 2-2
Blackburn	2.401	2.843	1-0	0.44	W 3-0
Arsenal	0.855	1.602	0-0	0.75	L 1-3
@Burnley	1.283	2.123	0-0	0.84	D 1-1
Expected Points added per Appearance	0.08	Total Expected Points added.	2.76	Expected Points added per assist.	0.55

The decline of the previous season was arrested in 2010/11,but the previous highs were never threatened and only four league goals came from Delap assists.However,three were relatively high value scores resulting in a healthy Expected Points per assist.Over half the goals scored following a Delap throw over the last two seasons have come in Cup competitions,reinforcing the thought that the surprise factor of the tactic no longer exists in the EPL.

Expected Points Value of Goals Scored from Delap Throws,2010/11 Premiership Season.

Opponent.	Expected Points Pre Goal.	Expected Points Post Goal.	Pre Goal Scoreline.	Increase in Expected Points.	Game Result.
Liverpool	1.110	2.182	0-0	1.07	W 2-0
@WBA	3.000	3.000	2-0	0	W 3-0
Sunderland	0.808	1.633	0-1	0.83	W 3-2
Aston Villa	0.980	1.802	0-0	0.82	D 1-1
Expected Points added per Appearance	0.07	Total Expected Points added.	2.72	Expected Points added per assist.	0.68

Delap continues to be a regular in Pulis' Stoke City sides in the current season,but the threat from his throws has failed to produce a single league goal so far.Ricardo Fuller,a major beneficiary of Delap's delivery has become a peripheral member of the squad and the expensive addition of Peter Crouch hasn't produced a score to add to the one the pair combined for seven years ago when both were Southampton players.

Expected Points Value of Goals Scored from Delap Throws,2011/12 Premiership Season.


None As Yet.

How EPL Teams Cope with Injury.

One of the more heart stopping moments during a football match is when your star player is slow to rise after a particularly hefty challenge.Teams invest heavily in their on field talent,so the prospect of a prolonged injury lay off can have season long implications for fans,managers and owners.When Stoke took long throw specialist,Rory Delap on loan from Sunderland,their thoughts were geared towards how his unique talent could help The Potters promotion push.Instead they lost him for the season just two game into his Stoke career following a double leg fracture,ironically against his parent club,Sunderland.Shorn of Delap's longthrows,Stoke finished two points shy of the playoff positions in 2006/07,but with the Irish International signed up and fit for duty in 2007/08 they grabbed automatic promotion partly on the back of eight goals directly from Delap assists.

Had Delap avoided a season ending injury on a cold October night at the Brit,he may well have made the difference that propelled Stoke into the playoffs,but his absence in 2006/07 and contribution in 2007/08 does highlight how teams can suffer when forced to play without their more productive players.Delap is a unique case,but more generally teams can be shown to underperform when either a leading goalscorer or their first choice keeper is missing.Sample sizes for individual sides are small and unrepresentative,but when pooled together teams score slightly less goals when without their top striker and concede slightly more goals when they play their second choice keeper.When translated into win probabilities,either absentee reduces his team's figures by about  five percentage points.

Canaries Fans hold their breath as top scorer,Grant Holt struggles to his feet.

 These kind of facts are interesting,especially as Premiership sides often rest their star player for less prioritized cupties,but they neglect the more common situation where teams have to deal with numerous,long and short term injuries to multiple squad members.NFL teams are obliged to provide a weekly injury report that details the likelihood of each roster player appearing in that weekend's game and increasingly football and rugby clubs are releasing accurate injury information.Sites such as the physioroom are particularly good at recording how many players days each club looses to injury each year.

I've taken four years of injury reports for the Premiership clubs during the 2000's and plotted the number of player days lost through injury for one season against the numbers for the next.This is to see if there appears to be any tendency for teams to be injury prone from one year to the next or if teams can stay consistently free of problems.

 The average number of days lost to injury by EPL teams who enjoyed successive seasons in the EPL over the period was just under 1100 days.Although some teams such as Newcastle and ManC breach this number in multiple seasons and teams such as Bolton were consistently under the figure,overall there appears to be very little correlation between injury rates across years.I've no doubt that some players become more injury prone,especially if we were to concentrate on certain types of injuries such as pulls or strains,but individual teams in general appear able or lucky in avoiding stocking their rosters with such players.However,the ability or otherwise of a team's medical staff to keep players on the field and the risks associated with buying talented,but possibly injury prone players will have to wait for more detailed data.

Of more immediate interest is how a heavy injury load impacts on a team's performance.Simply regressing a team's seasonal success rate against it's injury burden is unlikely to produce any useful results.Team talent is such an overwhelming factor in determining season on season success that it is unlikely that the injury effect will be apparent over just four seasons of data.Instead I have run a multiple regression using a team's injury burden and success rate in year N and it's injury rate in year N+1 to see if these variables correlate to that team's success rate in year N+1 and if so how it is affected.

Over the sample all three variables are significant indicators of success rate in the following year.The presence of multiple variables means I can't plot correlation curves,but I'll use the regression equation to illustrate how a change in injury rate in terms of playing days lost can impact on the number of games a team can expect to win or draw over multiple seasons.

How Different Injury Burdens can Alter a Team's Success Rate in Subsequent Years.

Success Rate. Year N.	Injury Burden. Year N.	Injury Burden. Year N+1.	Predicted Success Rate. Year N+1.
0.50	Heavy	Light	0.61
0.50	Light	Heavy	0.38
0.50	Light	Light	0.52
0.50	Heavy	Heavy	0.47
0.40	Heavy	Light	0.53
0.40	Light	Heavy	0.30
0.40	Light	Light	0.43
0.40	Heavy	Heavy	0.39
0.76	Heavy	Light	0.83
0.76	Light	Heavy	0.62
0.76	Light	Light	0.74
0.76	Heavy	Heavy	0.69

I'm keen to avoid too many data heavy posts,so I've defined the injury burden as either light or heavy.The average days lost per season per club was around 1100 days and ranged into the 2000's and down to just over 200 for the more fortunate teams.Heavy and light were therefore pitched towards these two extremes.

0.5 teams are usually comfortably mid table teams,but if they achieve these numbers with a relatively injury free squad and their luck changes in the next season,they can find themselves pitched in with the relegation fodder.By comparison team's with success rates of 0.4 who manage to escape the drop can improve markedly if they stay fit in their next year.It also appears that consecutive seasons where injuries are less frequent can allow the majority of teams to prosper slightly from year to year,while heavy injury loads over seasons can be lead to a drop in results.

For those interested in more specific examples,Arsenal's success rate fell and rose in tune with their injury record from 2004 through to 2007.Liverpool had to contend with just 600 player days lost to injury in 2005/06 in recording a success rate of 0.75,but when those days nearly doubled in 2006/07 their success rate fell to 0.63.While Birmingham can possibly consider themselves unlucky to be relegated in 2005/06,having survived the previous year by 12 points,their days lost jumped from 900 to over 1500 in their relegation year.

These results are based on four seasons of data,so it's perfectly possible for teams to buck each trend because of factors that aren't accounted for here.The data is also very broad as no account has been taken about the number or position of players who comprise the injury days.But if as seems likely,injury rates are partly out of a side's control,they can contribute to a season of either toil or triumph where the final league position owes something to fortune,be it good or bad.

Messi's Five from Seven.

Two unusual events occurred around Lionel Messi on Wednesday night.Firstly,he scored five of Barcelona's seven goals against Bayer Leverkusen and secondly,there wasn't an unseemly fight between his opponents to swap shirts with him at the final whistle.

We don't have a great deal of stats on the latter event,but we should be able to put a ball park figure on the likelihood of the former occurring.

Barca where very short odds to win the second leg on Wednesday,I gave them around an 85% chance of winning the game on the night and that wasn't out of step with most other estimations.They were very likely to outscore their German opponents,but seven goals is usually an ask even for the very best and the chances of Barca scoring exactly seven goals would have been in the region of a 70/1 chance.

If we now move onto Messi.Over the last two seasons he's scored around a third of his club side's goals.So there's around a 30% chance that if Barca score,the scorer is Messi.However,the tie was virtually won for the hosts after the first leg,so there was the real possibility that Messi wouldn't be asked to play for the whole game.So to account for this we should reduce slightly the probability of a Messi goal in this game.We'll make it 25%.

If Barca do manage to score exactly seven goals and Messi is on the pitch while they are scored,there are 42 different combinations in which he could score five of the seven goals.We've decided that the chances of Messi scoring is 0.25,so the chances of someone else scoring one of the goals is 0.75 and with 42 different way for him to achieve 5 from 7 it's around a 40/1 chance that he'll do what he achieved last night.

So 70/1 for seven Barca goals and 40/1 for Messi to get five of them......does that mean he'll have to play at least 3,000 more games before it's likely to happen again :-).