J O’Donoghue – a review

When compiling the accuracy charts from play it became apparent that James O’Donoghue’s point taking accuracy was freakishly good.

 

 

But how exactly did he achieve such a high Success Rate?

 

Table A; O’Donoghue’s attempts for points from play

			Shot Frequency		Success Rate
Sector	Shots	Points	Brogan	Average	Brogan	Average
4	4	2	9%	23%	50%	37%
5	11	8	25%	25%	73%	50%
6	9	6	20%	18%	67%	35%
7	4	4	9%	13%	100%	42%
8	13	11	30%	13%	85%	70%
9	3	3	7%	9%	100%	46%
Total	44	34			77%	45%

Note; – for a representation of the pitch segments see the graph in the Appendix

 

These are a remarkable set of numbers. Up until the semi final replay this year O’Donoghue had not missed an attempt at a point from inside the 20m line in nigh on three years. Not one. Even after that game he is running at an astonishing 90% Success Rate for all shots inside the 20m line.

A very high percentage of O’Donoghue’s point attempts come from Sector8; 30% versus an average of 13%. This is the most productive area, and an argument could be made that this would inflate your Success Rate but this is not always the case – as we have seen with B Brogan. Plus O’Donoghue is converting those “simpler” chances at a much higher rate than the average.

It is not just in Sector8 that O’Donoghue shows above average levels of accuracy – it is all across the pitch. For point attempts out wide inside the 20m line the average return is c44% – O’Donoghue has converted all seven of his attempts.

It continues outside the 20m line. O’Donoghue rarely shoots from the right wing (Sector4), only taking four shots over the three year period (possibly a point to note for opposing defences!) but when he does shoot from his favoured sectors (5 & 6) he converts at a combined 70%. The average? 44%.

59% of O’Donoghue’s shots occurred in the 2014 Championship when he compiled a 73% Success Rate. His shot chart is below. Given an overall Success Rate of 77% for the three year this year’s Championship was actually a down year (I jest – sort of!)

 

 

APPENDIX

 

Advertisement

B Brogan – a review

Following the review of point taking it became clear that B Brogan was an outlier, both in terms of volume of shots and Success Rates.

 

 

The volume of shots is not a real surprise given the fact that he has been the main marksman on the most attacking team over the past three years. Nearly every Dublin game being televised doesn’t hurt the volume either. His accuracy however, defined by both his Success Rate and his weighting, is a surprise.

 

Table A; Brogan’s attempts for points from play

			Shot Frequency		Success Rate
Sector	Shots	Points	Brogan	Average	Brogan	Average
4	11	4	19%	23%	36%	37%
5	10	5	18%	25%	50%	50%
6	13	5	23%	18%	38%	35%
7	3	–	5%	13%	0%	42%
8	15	7	26%	13%	47%	70%
9	5	1	9%	9%	20%	46%
Total	57	22			39%	45%

Note; – for a representation of the pitch segments see the graph in the Appendix

 

Brogan’s shooting returns outside the 20m line (Sectors4, 5 & 6) are bang in line with the average. His shot frequencies are split quite evenly across the pitch whilst his Success Rates, for such relatively small numbers, are remarkably in line with the average – nothing to see here really.

It is when we get inside the 20m line that issues arise. Firstly his shooting is concentrated in Sector8, directly in front of goals, with 26% of all shots coming from here. This is against the average of 13%.

Whilst his shot selection may appear correct (getting more shots from the most productive area) Brogan is only converting 42% (7 from 15) of his shots from Sector8 as against an average of 70%.

At this vantage it is hard to say why it is that Brogan is so poor from such a productive Sector. His misses have come against all forms of defences (Mayo, Donegal, Kildare, Laois, Meath) not just massed defences whilst it has occurred throughout all three years as well; 4 in 2012, 1x 2013 & 3x 2014. It could be that given Dublin’s propensity to go for goal he is holding onto the ball too long awaiting a goal opportunity allowing the defences to get back but whatever the reason it is this very poor return that is damaging both his Success Rate and his weighting.

As an aside when Brogan does try a shot from inside the 20m line, but not centrally, his returns are also very poor.

He is only converting 13% (1 from 8) combined from Sectors7 & 9 though those volumes are very small. It is perhaps more an indication for opposing defences – if you push him wide close in to goal he is probably more likely to offload than shoot.

Below is Brogan’s shot chart for 2014 – it is very indicative of what he has produced over the past three years.

 

 

APPENDIX

 

King of the Point Takers

Following on from the piece reviewing the deadball specialists it is now time for the point takers.

 

 

The first thing to note is the groupings by team.
• Given their propensity to be on TV, and the fact that their attack has been so prolific of late, it is unsurprising that six Dublin forwards have had >=25 point attempts.
• Kerry players sure can kick a point – their three most prolific shooters over the past three years, Cooper, O’Donoghue & Declan O’Sullivan, are all in the top 4 in terms of accuracy.
• Donegal’s big three, Murphy, McFadden & McBrearty, all performed below average when the three years are combined.

As well as that there are two obvious outliers in B Brogan (volume & weighting) and J O’Donoghue (Success Rate). How they achieved their respective returns is dealt with separately here (Brogan) and here (O’Donoghue).

In the deadball piece the above chart was followed by one showing the Success Rate by weighting. This showed a clear separation between Cluxton & O’Connor and the other deadball experts – they were converting harder frees at a higher rate. This chart is reproduced below but you will notice a high correlation between it and the one above.

 

 

This high correlation is due to the fact that the Success Rates – which are the main drivers of the weightings – don’t tend to differ from play as much as they do from deadballs. Therefore being prolific from one section will not necessarily increase your weighting. A bit convoluted but perhaps the following chart will illustrate the point

 

 

The majority of the players are bang on the trendline. Apart from some obvious exceptions (Flynn, B Brogan) the weighting is driven by the accuracy. If you are bang on average in terms of Success Rate then you are more than likely going to have an average weighting. This is not the case for deadballs where you can convert a high portion of very simple frees to achieve a high Success Rate but an average weighting.

Deadball accuracy has increased – why?

Overview

• There was an increase in deadball accuracy during 2014 from 67% to 73%
• This increase was not influenced by a change in the mix of deadballs attempted but rather a jump in the Success Rate of free kicks from 70% to 76%
• This increase occurred despite the fact that the volume of free kicks attempted outside the 45 rose.
• The increased accuracy was noted in all sectors inside the 45

How was the increase achieved?

Since starting the blog one of the main findings has been the deadball accuracy is quite predictive. Deadballs returned Success Rates of 66.6% & 67.1% for the past two years. I rarely reference 2010, the year I started tracking performance, but even then the returns for deadballs was 66.3%.There have been some fluctuations in the Success Rates for various types – 45s, frees, penalties & sideline attempts – but overall you could set your watch by it.

Except this year something different happened. The Success Rate for deadballs, from 25 games, was 72.9%. This is surprising for two reasons – the first being that there was any deviation from 67% and the second that the deviation veered towards the positive. Throughout the year it appeared that more frees were taken from outside the 45 (Niall Morgan, Rory Beggan, Michael Murphy et al) which are the most difficult to convert. If this were the case surely the overall returns should have decreased?

Table A; Success Rates by Deadball Type

	2012		2013		2014
	% of all deadballs	Success Rate	%of all deadballs	Success Rate	%of all deadballs	Success Rate
Free kicks	85%	71%	84%	70%	85%	76%
45s	12%	40%	13%	50%	10%	52%
Penalties	1%	100%	2%	67%	3%	82%
Sidelines	1%	50%	2%	33%	2%	17%
Total	32%	85%	28%	85%	31%	94%

Although there is fluctuation in the Success Rates for Sideline attempts and penalties this is almost entirely due to low sample sizes – there is a combined 38 instances of these deadball types over the three years.

There has been an increase over the years in the Success Rates for 45s but the big jump occurred between 2012 and 2013. The accuracy increased slightly again in 2014 but not in any way that would change overall percentages.

So that leaves free kicks. The volume of free kicks per game has dropped (13.1 in 2103 to 11.4 in 2014) but as a total of all deadballs it has remained rock steady at 84-85%. What did change in 2014 was the accuracy. 76% of all frees were converted in 2014 compared to 71% and 70% in 2012 & 2013.

Table B; A yearly breakdown of free kicks

	2012		2013		2014
	% of all deadballs	Success Rate	%of all deadballs	Success Rate	%of all deadballs	Success Rate
Outside the 45	13%	43%	11%	43%	15%	35%
Between the 21 & 45s	55%	68%	61%	67%	54%	77%
Inside the 21	32%	85%	28%	85%	31%	94%

The Success Rates for 2012 & 2013 were remarkably similar. So much so that it didn’t even enter my thought process that it could be a coincidence. I took the year on year similarity as evidence that deadball accuracy was a very predictable entity. The 2014 returns show us that it is not.

As had been expected the volume of frees taken outside the 45 increased (from 11% last year to 15% this). The fact that the Success Rate decreased is probably a function of just how far out some of those frees were. Michael Murphy tried some absolute monsters this year.

The real change however is in the accuracy from 45 metres in. Using the tried and trusted Sector approach all frees inside the 45 were further subdivided into 6 Sectors.

Where was the increase achieved?

Table C; Success Rates for free kicks, by sector, inside the 45

Sector	2012	2013	2014
4	65%	52%	72%
5	82%	83%	87%
6	56%	61%	68%
7	73%	86%	89%
8	94%	88%	100%
9	75%	75	86%

A couple of things to note here

• All six sectors produced returns above anything recorded in 2012 or 2013. There was not one particular facet of free taking that helped increase the overall accuracy. It was an overall general rise in standards.
• The two sectors that showed the biggest improvement on 2013 were Sectors 4 & 8. These are two of the top three sectors for where the volume of frees were taken. Increase accuracy in the sectors with the greatest volume and you’ll have the greatest impact on overall numbers.
• A Success Rate of 100% in Sector8 was aided by the fact that none of the frees were a desperate last minute shot for a goal. In 2013 four such attempts were missed – the Success Rate without these would have been 94%. Similarly there was one goal attempt in 2012 which was missed – without that the Success Rate would have been 96%. So whilst 100% is obviously impressive it is not *that* unexpected.

Note – all the above metrics are for Championship games shown on TV (25 games for each of the three years) and exclude any extra time within these games for ease of comparison.

Weightings. A good comparison tool but not without its faults

With a bit of a lull in games to review now seems as good a time as any to post this.

I received a text from Brendan Coffey, a journalist with the Kildare Nationalist who is well worth a follow on twitter (@coffeybrendan), after the Dublin-Donegal game querying how it was that Ryan McHugh’s weighting was greater than Paul Flynn’s. Both converted four from five from play yet McHugh’s attempts were much closer to goal. With Flynn’s attempts being harder to convert surely his weighting should be higher?

And there is the rub. The weighting is not a finely tuned precision instrument but rather a roughly hewn prototype. I believe it gives a better sense of a player’s accuracy than any other measurement out there but due to its rough nature there will be inconsistencies.

A weighting is essentially a value for a shot based on historical Success Rates for similar shots. It is broken into four main components
• Is the shot from play or a deadball?
• What type of deadball is it?
• Is the shot for a point or a goal?
• Where on the pitch was the shot taken from?

Is the shot from play? This is easily classified but there will be differences of opinion as to what constitutes a shot. Anything classified as a shot gets a weighting. This is irrespective of game scenario (Dublin’s two goal mouth scrambles at the end of the Donegal game are counted as shots on goal as if they were one-on-ones in the 15th minute).

When creating the weighting originally the aim was to have no subjectivity. As time passes it is obvious that not all shots are created equally. Colm McFadden’s goal in the semi final is not the same as Connolly’s attempt in the 66th minute. To give them separate weightings however it would be necessary to (a) create subjective definitions over how hard an attempt was, the angle and distance of the shot etc and (b) collate enough examples to create a meaningful Success Rate in each new sub category.

Type of shot? Again this is easily classified however the weighting to attribute to each is not. Take an attempt from a sideline. Ideally we would have an individual weighting for this type of shot but in the database there are less than 20 shots from sideline balls. A weighting could be created from these twenty but given the low volumes the Success Rates could be way off the “real” returns. Instead I have decided to treat attempts from sidelines the same as frees from that sector. Now it stands to reason that the majority of frees are easier than a sideline therefore when a player misses a sideline attempt their weighting will be quite harsh.

A subjective decision (to treat sideline attempts the same as frees from that sector) has an impact on the objective outcome (the weighting). In hindsight perhaps it would be best to go with the objective return (take the Success Rates from the 20 sideline attempts and just go with that. Something similar is at play with penalty attempts where I have treated them as goal shots from play rather than use the weighting from their actual Success Rates.

Is the shot for goal? There is some subjectivity here but I would say that 90% – 95% of shots at goal are easily identified as such. The others can be harder to discern i.e. when the ball scoots over the crossbar – was he drilling it for a goal or drilling it high hoping for a goal? Currently whether you get a goal or a point the weighting is treated the same for a shot on goal. The player got a score. Again this is probably being generous to those players who aim high and get the point as opposed to those who try for the goal only.

Where on the pitch? This is relatively straight forward but again you can have different camera angles, poor camera tracking, a low lying sun obstructing your ability to see the lines on the pitch (and gauge where the shot was taken from). 99% of shots are easily placed on a pitch diagram but there will be that 1% that requires some guesswork. And that doesn’t touch on the subject of how big the sectors are. Only having six sectors inside the 45 is too few – the sectors need to be refined. Once they are the weightings will be adjusted accordingly.

So there you have it. The weighting is a good tool for shooting accuracy comparisons between players, teams and games. It is robust for 90% of shots and scenarios but when we get to the margins (sidelines & penalties, grading goal attempts) it is not fool proof.

And so we come to the original query – how was McHugh’s weighting better than Flynn’s? Three of McHugh’s shots were from Sector8 but all three were shots for a goal. He may have scored 2-01 but as outlined above these are all treated the same for weighting purposes. Those three shots gave him +1.778 (+0.5927 for each score). His first shot was from Sector5 which went wide (-0.4987) whilst his other point was from the left in Sector6 which has a weighting of (+0.6408). In total +1.920.

What of Flynn? His long range shooting was exceptional but unfortunately (for him) the sectors are too big to account for this. Thus his four points came from Sector4 (2 x 0.6407) and Sector 5 (2x 0.5013) whilst he also missed one from Sector5 (-0.4987). A total of +1.785.

Over the years players have converted point attempts from Sector5 more frequently than they have goal shots from Sector8. As such, from a weighting perspective, McHugh’s conversions were harder (taking all the above points into account) than Flynn’s were.

What Lies Beneath – Success Rates for Various Shot Types

This was originally published over on livegaelic before the season started – I meant to put it up here but have only gotten around to it now.

Introduction
I have been gathering data on Championship games since 2010. Unfortunately I do not have any games from 2011 however for the other three years 86 games have been charted with 6,593 attacking possessions and 4,805 shots recorded.

There have been minor adjustments over the years as to how the data is gathered. Where this is the case any extrapolation will be highlighted. We also have to take into account the fact that all the games are charted manually so there will be errors and inconsistencies however I am confident that they are minimal as (a) I don’t do games live and have the benefit of time to review and check and (b) I do all the games myself so how I interpret actions (a shot, or pressure for example) will be relatively consistent.

Overall

Year	Games	Possessions	Avg possessions	Shots	Shot Rate	Scores	Success Rate
2010	36	3,013	83.7	2,041	68%	1,016	50%
2012	25	1,764	70.6	1,351	77%	698	52%
2013	25	1,816	72.6	1,413	78%	713	50%

There are a few things that jump out from this one table alone. The main one is the contrast between 2010, in terms of average possessions per game and Shot Rates, and 2012 & 2013. They are quite a way apart given how consistent ’12 & ’13 have been.

There were 44% more games in 2010 and given that these would all have been earlier in the Championship I did wonder if there was something in the quality of the games recorded that was producing this result. As such I extracted the same seven games from each year (4 quarter finals, 2 semi finals and the final) and compared them

Year	Games	Possessions	Avg possessions	Shots	Shot Rate	Scores	Success Rate
2010	7	579	82.7	413	71%	209	51%
2012	7	492	70.3	377	77%	204	54%
2013	7	510	72.9	388	76%	205	53%

The Shot Rate gap does shrink but the Avg. possession rate remains stubbornly different. We do know that football is a copy cat game – 2011 saw the emergence of the Donegal wall as well as a tempered Dublin. It could be that following both those teams’ success an emphasis was placed on minding the ball. Recycling the ball and attempting to take more quality shots, rather than a greater quantity of shots, seems to be in vogue.

As an aside it is interesting to note that the higher profile games – quarter finals onwards – where one would expect more evenly matched teams brought about higher Success Rates. Despite, in theory, these games possessing better defences the Shot Rates remained relatively static in ’12 & ’13 but the Success Rates increased. At a very, very generalized level the better forwards outperformed the better defences.

Shot Outcomes
From Play
Before we move on to looking at the various shot types below is the outcome of all shots from play.

	2010	2012	2013	Total
Blocked	7%	6%	7%	7%
Goal	4%	4%	4%	4%
Goalkeeper	7%	6%	5%	6%
Point	41%	42%	40%	41%
Saved	2%	3%	4%	3%
Short	6%	5%	7%	6%
Wide	34%	30%	31%	32%
Framework		3%	2%	1%

There is a remarkable consistency to these outcomes. 44-46% of all shots end in a score (goal or point) whilst 33 – 34% of all shots were wide each year (note that in 2010 all shots that hit the framework were considered wide … these were separated in ’12 & ’13). Now we know that the type of shot will have a big bearing on the Success Rates however if you are a member of a club then by simply tallying the outcome of your team’s shots you will know how far off inter county standards your team’s shooting is.

The only real “trend”, and it is a minor one overall, is the fall in the number of shots dropping into a goalkeeper’s hands. It has fallen from 7.4% of all shots in 2010, to 5.8% in ’12 and 5.4% in ’13. I would be surprised if this was due to anything other than teams preaching a mantra that emphasises minimizing turnovers; the belief that it is better to kick the ball wide, and set for the kickout, than let your opponent start a counter attack when the team is not ready. It would be interesting to see how many shots a team gets from possession garnered from balls dropping into the keepers hands versus from kickouts. Another one to add to the ‘to do’ list.

Knowing that 45% of all shots from play end in a score is useful. It can be used as a convenient benchmark. However there are a lot of other factors that feed into whether a particular shot will be successful or not; strength of the wind, where on the pitch the shot was taken from, rain, strength of the opposition, whether a defender was placing the shooter under pressure to name but a few.

A lot of these are subjective (pressure?) whilst others are difficult to gauge watching games on tape (how do you grade various wind strengths). One factor we have been able to consistently capture however is where on the pitch a shot was taken from.

I have always used the above segmentation and whilst it has its drawbacks it has two major strengths. One it is objective as all pitches are marked the same and two it is readily transferable across grounds and broadcasters. No matter the camera angle, or trajectory of the sun, you will rarely struggle to place a shooter in the above grid using the pitch markings.

Having said that how do Success Rates for shots for points for the various segments line up?

Sector	2010	2012	2013	Total
1	100%	67%	0%	83%
2	31%	56%	22%	38%
3	60%	0%	40%	44%
4	34%	40%	35%	36%
5	49%	52%	50%	50%
6	37%	36%	34%	36%
7	41%	42%	45%	43%
8		67%	75%	71%
9	40%	49%	38%	42%

The returns for 2010 from Sector 8 are blank as there was no distinction between shots for goals and shots for points in that year. Alongside the returns it is instructive to review where the volume of kicks were taken from (see Appendix).

For shots from outside the 45 (Sectors 1-3) there is a lot of volatility; this is purely due to low volumes however. Only 2% of all shots – 77 shots in 86 games – in the three years occurred outside the 45. It is very difficult to extract anything meaningful from this data.

From year to year there have been fluctuations on the returns from the wings (Sectors 4 & 6) however over the three years there is no difference. Whether kicking from the left or the right the Success Rates are 36%. As a grouping Sectors 4 & 6 combined produce the most shots (38% across the three years) but the returns are relatively poor.

It is not a case that as a trainer or coach you tell your players not to shoot from here but you must know your players and who is likely to at least hit the average on these shots. Paul Flynn probably has licence to shoot from here all day – should Jack McCaffrey? Only if you know, from data gathered in games and training, what his Success Rate is like.

Inside the 20m line there is again volatility over the years but when combined there is practically no difference either. Shooting from the right or the left (Sectors 7 & 9) the Success Rates come in at 42%-43%.

Which leaves us with the centre. Just under a quarter of all shots come from Sector5 whilst 50% are converted. A conversion rate of 50% may seem low for shots more or less in front of goal but in the context of a 36% Success Rate from the Sectors either side of it this is quite a fruitful avenue.

Sector 8, straight in front of goals, surprises me somewhat. Remember these are deliberate attempts for points so goal shots are excluded. These should be the simplest shots in the game but yet a sizeable minority, 29%, are missed. This is where our lack of defensive reference points has a bearing. You would imagine that the vast majority of those missed shots came about due to defensive pressure.

From Deadballs
Success Rates for shots from deadballs

	2010	2012	2013	Total
Free	70%	71%	70%	70%
45	49%	40%	50%	47%
Sideline	13%	33%	33%	21%
Penalty	100%	100%	67%	88%
All deadballs	66.3%	66.9%	66.6%	66.5%

Again the level of consistency stands out. Over the three years the Success Rate from deadballs has remained static with the only fluctuations coming from those methods of delivery where volumes are low. Free kicks, as a percentage of all deadball shots, were 84.9% over the three years with a very narrow range (84.7% in ’10, 86.0% in ’12 and 84.1% in ’13). The average number of frees in a game across the three years is 12.4 – with a Success Rate of 70% your average free taker, in an average Championship game, will deliver 4.3 points.

I have always wondered at the low return rates for 45s however in the context that there is only, on average, 1.76 45s a game it is easy to see why players and coaches might deem it ancillary.
Frees only

Sector	2010	2012	2013	Total
1	25%	33%	50%	36%
2	47%	59%	42%	48%
3	42%	25%	38%	37%
4	55%	65%	52%	56%
5	86%	82%	83%	84%
6	58%	56%	61%	58%
7	78%	73%	86%	78%
8	93%	94%	88%	92%
9	94%	75%	75%	84%

Much like shooting from play there are distinct groupings of sectors here. From play we saw that there was a large uptick in Success Rates for shots down the centre inside the 45m. This is also evident in the free taking returns and is extended outside the 45 as well where, unlike shots from play, a large portion of frees emanate from (14% over the three years).

Again there is little difference in Success Rates for frees from the wings; 36% & 37% for frees from Sector1 & Sector3 and 56% & 58% for Sector4 & Sector6.

There is some variation in the outcomes for frees inside the 21m line with 78% of frees converted from the right (Sector7) and 84% from the left (Sector9). This may be due to right footed free takers taking some frees from Sector7 that more naturally fits a left footed free taker – but that is just conjecture on my part.

Pressure

One last piece. Previously we mentioned certain variables that could affect the Success Rate of shots. One that was mentioned was pressure. Although I have tracked whether a shot was taken under pressure since I started doing games I have never used it. The main reason for this was that it was a subjective analysis – put 10 GAA fans in a room and they would come up with ten different definitions.

Still the data is there so lets have a look at the outputs

Year	Shots	Pressure-yes	Scores	Success Rate	Pressure-no	Scores	Success Rate
2010	1,513	857	348	41%	649	315	49%
2012	1,007	622	263	42%	377	200	53%
2013	1,024	579	244	42%	438	205	47%

The results are, more or less, as expected. Each year there is a clear downshift, as against the mean, for those shots taken under pressure and an uplift for those taken without pressure. There is also quite a bit of consistency year on year which is quite comforting – it would lead you to believe that whilst subjective how pressure is being tracked is accurate. Definitely a subject worthy of its own piece.

APPENDIX
Breakdown of shots for points from play

Sector	2012	2013	Total
1	0%	0%	0%
2	2%	1%	2%
3	0%	1%	0%
4	21%	21%	21%
5	25%	23%	24%
6	18%	17%	17%
7	10%	15%	12%
8	14%	13%	13%
9	8%	10%	9%

Breakdown of where shots from deadballs taken

Sector	2010	2012	2013	Total
1	2%	4%	2%	3%
2	10%	6%	6%	8%
3	4%	3%	2%	3%
4	16%	16%	18%	17%
5	21%	20%	25%	22%
6	18%	19%	17%	18%
7	8%	7%	6%	7%
8	13%	18%	17%	16%
9	7%	7%	5%	6%

Going for, going for – goals

Following on from the remarkable stat in the recent Dublin-Tyrone league game where 40% of Dublin’s shots from play were attempts on goal I decided to revisit the ’13 Championship returns.

Team	% of all shots	Conversion Rate	Point shots per goal shots
Dublin	25%	32%	2.98
Mayo	15%	59%	5.86
Kerry	11%	40%	8.20
Tyrone	8%	13%	11.38
’13 Average	13%	31%	6.70
’13 Semi Finalists	16%	38%	5.23
’13 Non Semi Finalists	10%	19%	8.98

Perhaps to no one’s surprise Dublin were the most goal hungry of the four semi finalists, and indeed of all counties covered within the 25 games, with 25% of all their shots from play being attempts at goal. Considering the average as a whole is 13%, and Mayo were next best of the semi finalists at 15%, that is a huge gap. Even measured a different way, goal shots per shots for points, Dublin still considerably outstrip all other teams in terms of how often they go for goal.

Considering the array of talent at Dublin’s disposal their goal finishing is disappointing at 32%. Much like in their breath taking performances against Cork and Derry at the back end of this year’s league they did not rack up goals because of deadly finishing – they racked up scores because they produced average quality finishing on a ridiculous amount of chances.

Is this a Gavin ploy or something Dublin do naturally? The finishing is consistent from year to year with Dublin converting 32% of their goal shots in 2012 however the volume of shots was running at 18% as opposed to 25% (or one goal shot for every 4.68 shots for a point). This year will tell us more as that drop could just be down to randomness or the opponents faced but it is significant. Given the returns in the 4 league games covered in 2014 it would appear to be Galvin led though.

The quality of the goal shot attempted (position on the pitch, goalkeeper position, number of defenders in the vicinity – or indeed in the way) is not tracked however looking at the very low conversion rates for the non semi finalists I would be willing to bet that those returns are an amalgam of (a) less skilled finishers and (b) teams chasing a game taking goal shots from less advantageous positions.

*as an aside the fact that Dublin take so many shots means that they could have had a greater impact on what the average is. Without Dublin the average conversion rate is 30.4%.

The Importance of the Gooch

When Colm Cooper went down in a heap in the All Ireland club semi final not only did the Dr. Crokes fans hold their breath but so did the majority of the GAA community.

Cooper has been a consistently excellent player for Kerry with his first half virtuoso performance in the 2013 semi final against Dublin a master class in peripheral vision and a player imposing himself on a game.

Given the importance of Cooper to Kerry trying to ascertain his worth through stats alone will inevitably undersell his impact however Kerry will have to replace his shooting; so exactly what will they have to replace?

Overall
There are twelve games spread across three years for Kerry in the database and Cooper played in all of them. The table below shows the shooting statistics for Cooper across those games lined up against the rest of the Kerry team.

	Shots	Scores	Success Rate	Weighting
Cooper	80	60	75%	+9.313
Rest of Kerry	249	122	49%	+0.989

We’ve always known how good Cooper has been but that is one hell of a hole that Fitzmaurice is going to have to plug. Cooper has taken 24% of Kerry’s shots in those twelve games and accounted for 33% of all their scores. He has been consistent throughout the 3 years as well.

From play

	Shots	Scores	Success Rate	Weighting
Cooper	37	24	65%	+5.856
Rest of Kerry	199	91	46%	+0.269

Again Cooper has been excellent and again he has consistently outperformed the rest of the Kerry team. The volume of shots has dropped as a whole (16% of all shots) but the quality – with a combined weighting of +5.856 (65% Conversion Rate) – is stellar.

Cooper’s numbers on their own are fantastic but they do not take into account who he has taken the shots against. In the twelve games he has faced Cork four times and the subsequent All Ireland winners in another two (Donegal in the ’12 QF & Dublin in the ’13 semi). Those numbers are not padded by easy games.

Fitzmaurice *may* be able to find a player (or players) to pick up the slack for the volume of shots that Cooper has taken but it does not appear that his quality of shots can be maintained. Of the 199 shots the rest of the Kerry players have attempted they have converted 46% (weighting +0.269). Essentially bang on average.

From Deadballs

	Shots	Scores	Success Rate	Weighting
Cooper	43	36	84%	+3.457
Rest of Kerry	50	31	62%	+0.720

Once again Cooper’s figures hold up. He has taken 43 free kicks in the twelve games converting 84%. The average is 67%. Now the combined weighting of these frees at +3.457, whilst good, show that the frees were more often than not on the easier side but the important point is that he still converted them. An average free taker would have missed three or four that Cooper converted.

The rest of the Kerry team converted 62% of the 50 deadballs (a mixture of frees & 45s) they took. These shots were definitely harder than Cooper’s – a positive weighting with a below average Success Rate shows this – however again the weighting shows that they were no more than average.

Impact
No one man is a team but some are much harder to replace than others. From frees Cooper gave Kerry a deadly accurate and consistent free taker from inside. From play his shooting was other worldly. On the flip side the rest of Kerry’s shooting was average. If Fitzmaurice has a brilliant shooter to replace Cooper’s accuracy you have to believe we would have seen him by now. If Cooper’s shots are spread across the rest of the team (and his replacement) the evidence shows that the returns will revert to the mean and Kerry will suffer.

No one man is a team. There is however only one Gooch.

Kickouts & why teams kick short

This article first appeared in LiveGaelic and was followed up by an interesting article from ex-Dublin goalkeeper coach Gary Matthews

The basics
Over 25 Championship games in 2013 1,095 kickouts were charted. Due to the vagaries of TV coverage not every kickout was captured however we are still left with a healthy 1,065 kickouts for which we have a complete picture. It is these kickouts that are referenced below.

The kickout team gained possession (won) 64.5% of the time and from those possessions – gained directly from their own kickouts – they managed a shot 75.9% of the time.

On the 35.5% of kickouts won by the opposition they managed a shot 79.1% of the time. Therein lies the challenge; by their choice of kickout type can a team maximize the number of shots they create from kickouts whilst minimizing those they give to the opposition?

Definitions
First let’s look at the definitions used in measuring kickouts. Much like the shooting statistics on the blog the pitch is broken up into segments.
• Any kickout that lands inside the 45m line is considered to be short
• Any kickout that lands between the 45m & 65m lines is considered to be medium (referenced as “mid” )
• Any kickout that lands past the 65m line is considered to be long

A team is deemed to have “won” the kickout if they are the first team to gain possession from that kickout.

So where did the kickouts go?

Length	% of all kickouts
Short	21%
Mid	30%
Long	49%

Despite the predominance of the short kickout in commentary on the subject just about one in five kickouts go short. Half of all kickouts still go long with the remaining 30% landing between the 45 & 65m lines.

49% of all kickouts are won through broken ball whilst 43% are won cleanly. This however is a function of where the kickout lands as 99% of all broken ball from kickouts is won in the “mid” and “long” range whilst 55% of all clean wins occur from “short” kickouts.

Kicking Team Outcomes

Length	Won	Turned into a Possession	Turned into a shot
Short	94%	59%	81%
Mid	61%	66%	75%
Long	54%	75%	74%

The above table shows the results for the kickout team. At first glance the data intuitively make sense. The longer the kickout the less chance the kickout team has of winning the ball however the more likely they are to convert that win into an attacking possession. (a possession is where a team has control of the ball within the opposition’s 45)

	Kickouts	Won	#shots
Short	100	94	45
Mid	100	61	30
Long	100	54	30

Were we to use the aforementioned conversion rates in a sample of 100 kickouts from each segment you can see from the above table that a team will get 50% more shots from short kickout routines. Granted this is just one year’s data and as such is open to variances based on the teams actually covered in the 25 games however it is still noteworthy. 1,000 kickouts is a hefty sample size and 50% uplift in shots is quite significant.

Defending team outcomes
This of course only tells the story from the kickout team perspective. We have to balance the outcomes above with the returns that the defending team achieves.

Length	Won	Turned into a Possession	Turned into a shot
Short	6	100%	77%
Mid	39%	79%	80%
Long	46%	72%	78%

Similar to the kickout team’s returns the above table makes sense. The defending team win more kickouts the further from goal the ball travels, mainly because there is more of a contest, however when they do win them then the possession rate increases the closer to goal you go.

For both these scenarios (kickout & defending team) there is an argument to be made that the quality of shot will differ depending on the strength of the opposition, how strong the individual teams are at kickout strategy etc. but that is for further study. There is also the consideration that short kickouts won by the defending team may lead to more goal chances but the volumes are so low, that whilst an important consideration, it can be omitted for now.

Again if we convert the percentages to what would happen for every 100 kickouts we get the below.

	Kickouts	Won	#Shots
Short	100	6	4
Mid	100	39	24
Long	100	46	26

Combining the outputs give us a complete picture

	Kickouts	Won	# Shots	Won	# Shots	Shot Differential
Short	100	94	45	6	4	41
Mid	100	61	30	39	24	6
Long	100	54	30	46	26	4

(2nd & 3rd columns are for the kickout team; 4th & 5th for the defending team)

That’s a fairly conclusive table, (always with the proviso that it is based on one year’s data), in favour of kicking short.

It is not that this difference is driven by better teams either. The below table shows the actual results for short kickouts throughout the year.

	Short Kickouts	Possessions	Possession %	Shots	Shot Rates
4 semi finalists	117	68	58%	57	84%
Other	109	70	64%	55	79%

Converting these results into 100 short kickouts for the 4 semi finalists combined, versus all the other teams combined, the non semi finalists would have manufactured 1.7 shots more from their short kickouts.

What is noticeable however is the mix of kickouts that the more successful teams employed.

	Kickouts	Short %	Short or Mid %
Dublin	117	26%	73%
Mayo	114	38%	64%
Tyrone	83	19%	37%
Kerry	89	30%	69%
Other	662	16%	43%

Three of the four semi finalists (Tyrone being the exception) kicked a significantly smaller portion of all their kickouts long. So whilst the more successful teams are not necessarily better at the execution of short kickouts they do seem to have recognized its merits.

Though commentary is shifting, in the main kickout success rates are used as a proxy for whether a team is winning the midfield battle. In one sense this is not wrong as by far the biggest percentage of kickouts are hit straight and long (31% of all kickouts).

However kicking long is taking an increased risk with your own team’s possession. By doing so you are more than likely giving the opposing team extra opportunities to shoot that would not materialise were a greater portion of your kickouts short.

Am I advocating taking every kickout short? No. What is predictable is easily defendable. Plus no tactic or strategy should be built on one year’s data. However every team should have the ability, and confidence, to alter their strategy as the demands of the game require. And fans should probably show more patience and not be so quick to berate teams that kick short. “Hoofing” it down the middle is not always the best option.

Home field advantage

In their book Scorecasting economist Tobias Moskowitz and Sports Illustrated columnist L. Jon Wertheim looked at many truisms in sport and attempted to prove their existence or otherwise.

One of the things they looked at was home-field advantage; did it exist and if so why? They took the five main sports in the US and worked out the percentage of games won by the home team in each. There was definitely a home team edge.

League	Home team win %
MLB	53.9%
NHL	55.7%
NFL	57.3%
NBA	60.5%
MLS	69.1%

Next up they took some of the more usually held truisms as to why a home teams win; crowd support, being more familiar with your surroundings, lack of travel etc. They effectively ruled out any of these as reasons for the home-field edge and instead landed on one main reason; referees.

Referees are involuntarily affected by the home support. They give 50:50 decisions to the home team to, again involuntarily, appease the home support. This makes sense in the case of the MLS having the biggest home-field advantage as in such a low scoring sport the referee can have a larger impact on the final score. This notion was supported by a paper on home field advantage in the Bundesliga by Thomas Dohmen in which teams with a running track around the pitch (thus the ‘pressure’ from the home crowd is lessened) have a lesser home-field advantage.

So. Knowing all that how does it stack up for GAA football?

Rob Carroll(@gaelicstats) & Kieran Collins(@DKieranCollins) did a review of home-field advantage in GAA football (see here) and concluded that there was an advantage in the League with the home team winning 56.6% of the time from 2001 – 2010.

With the league about to start I decided to review home-field advantage and see if we could delve a little deeper.

The below table shows the winning percentage of home teams, by Division, from 1999-2012 inclusive. Though probably not scientifically correct I removed all games involving Kilkenny & Carlow as I felt these *might* skew the lower league returns.

Division	Home team win %
Division 1	62.7%
Division 2	62.9%
Division 3	68.2%
Division 4	61.2%

The overall home-field winning margin of 64.1% is quite a bit above those recorded in the aforementioned review. The main explanation for this, which will be expanded upon below, is that the further you go back the greater the home-field advantage was.

A few initial thoughts

• An overall home-field advantage of c64% makes intuitive sense. If the referee is the biggest determinant then you would expect a sport with amateur referees to be highly affected.
• The referees have a big impact on GAA but again probably not to the extent that they do in soccer so having a home field advantage of c5% less than soccer again makes sense
• I was surprised at the fact that Division3 was so far ahead of other divisions

The Division3 spike intrigued me so I started to look through the data and it became evident that there was an issue with how I had created the Divisions. The streamlined League, as we now know it, came into being in 2008. Prior to that we had 2 Divisions split into As & Bs. I had thus made Division 1A = Division1; Division 1B = Division2 etc. Looking at the games it didn’t feel as if the Divisions were as representative of strength as they are now. Thus I broke the home-field advantage into new league format (’08 – ’12) which has been in operation for 5 years and the old league format for 5 years prior to that (’03 – ’07)

League games ’08 – ’12

Division	Home team win %
Division 1	54.3%
Division 2	53.6%
Division 3	62.9%
Division 4	52.4%

League games ’03 – ’07

Division	Home team win %
Division 1	64.3%
Division 2	66.0%
Division 3	67.6%
Division 4	61.4%

Again Division3 stands out and I cannot for the life of me fathom it. Not only does it stand out but it is a relatively recent phenomenon

Above that however is the sea change in home-field advantage that the new league format has introduced. In the past 5 years home teams have won 8.5% less games compared to the previous 5 years. That is quite a significant drop.

Perhaps it is not the new league format that has led to this change but rather how the league is now viewed by teams and managers. There has been a change in emphasis in recent years with managers like Mickey Harte & Conor Counihan seeing the league as a competition worth winning in its own right. Or perhaps it is a mixture of the equalisation of team strengths in the new League format, alongside a change in emphasis, that has led to this drop. Either way home-field advantage has dropped considerably during the past 5 years.

Appendix

Home-field advantage for each county since the introduction of the new league structure (’08-’12)

County	Home team win %
Cork	88%
Down	76%
Kerry	72%
Wexford	72%
Clare	67%
Dublin	67%
Sligo	67%
Tyrone	67%
Leitrim	65%
Monaghan	65%
Limerick	61%
Antrim	59%
Cavan	59%
Louth	59%
Offaly	59%
Kildare	56%
Donegal	56%
Longford	56%
Roscommon	56%
Tipperary	56%
Armagh	53%
Derry	53%
Meath	53%
Waterford	53%
Wicklow	50%
Westmeath	47%
Galway	44%
Fermanagh	44%
Laois	37%
Mayo	35%
Carlow	33%

Data
All the results were taken from gaainfo.com (a wonderful site). I did start to check the venues to make sure those teams listed as being at home were at home however venues start to become sketchy fro c2007 so gave that up. The results, and listing of home teams, did pass the test for the most recent years so that will have to suffice!

Pre ’03
For those of you eagle-eyed … the home field average for  ’08 – ’12 was 56%, for ’03 – ’07 was 64.6% yet the home field advantage from ’99 – ’13 was 64.1%. The reason for this is that the home-field advantage for ’99 – ’02 was 78.9%!!

League games ’99 – ’02

Division	Home team win %
Division 1	76.1%
Division 2	74.3%
Division 3	84%
Division 4	83.1%

The drop from ’99 – ’02 to ’03 – ‘07 is for another day!