Never back Each Way in GAA

Never back each way (EW) in the 1^st goalscorer market with Paddy Power. And I mean never. For the uninitiated EW is explained here ( https://en.wikipedia.org/wiki/Each-way ) but it is essentially two bets; half your stake on your selection scoring the first goal and the other half on them scoring one of the first “x” goals as laid out by the bookmaker in the place terms. For GAA Paddy Power make this “x” to be the first three goals. Below is a screenshot of the goalscorer market for Tyrone v Donegal in the Ulster semi-final (2019) with the place terms highlighted.

Why not bet EW? Because the answer is there on the screenshot … instead of putting €1EW on a player put €1 on the same player in the 1^st goalscorer market and the other €1 on the Anytime market. Why? Let us use Peter Harte as an example.

Harte is 7/1 to score the first goal. If you put €1EW on Harte to score first, and he does so, you get €10.75 back (calculations below). If his first goal is the 2^nd or 3^rd goal in the match you get €2.75 back. If his first goal is the 4^th goal or later (or indeed he doesn’t score) you get nothing

So:

1^st goal – +€8.75

2^nd/3^rd goal – +€0.75

4^th goal onwards, or no goal – -€2.00

 

Now instead of backing Harte €1EW lets split our stake to be €1WIN and €1 ANYTIME. The returns for same are

1^st goal – +€9.50

2^nd/3^rd goal – +€1.50

4^th goal onwards +€1.50

No goal – -€2.00

 

On every permutation the ANYTIME return is better than the EW return because the place terms (for the EW portion) are just so poor. And on top of that you get paid for the 4^th goal onwards in the ANYTIME market. This holds true for every player.

Never back EW in the 1^st goalscorer market with Paddy Power. And I mean never

 

Calculations

EW market

€1 WIN @ 7/1 = (€1 *7 + original stake back) which = €7 + €1

€1 PLACE = ((7/4) *1) + original stake back) which = €1.75 +€1

ANYTIME market

€1 WIN @ 7/1 = (€1 *7 + original stake back) which = €7 + €1

€1 ANYTIME = ((5/2) *1) + original stake back) which = €2.50 +€1

Expected Wins; how teams fared versus their odds

Once September rolls around only one or two teams will deem their year as being successful. In 2015 Dublin had a year of years winning the league, Leinster and the All Ireland (do we throw in the O’Byrne cup?). Monaghan winning Ulster made for a successful season whilst there is an honourable mention for Fermanagh with promotion to Division2 and the quarter final appearance. But what about the rest?

If the league is a means to an end for the majority, and the All Ireland and Provincial championships are regularly shared by the same teams, how do we measure the remainder’s performance? Or indeed how do we judge a team like Tyrone that got relegated, fell short in Ulster but rallied to get to the All Ireland semi-final? One way is to compare a team’s results against how bookmaker’s thought they should fare.

Bookmakers give odds on all games. The main markets are match odds and handicap. Any bookmaker worth their salt will tell you that though all odds can be converted into a percentage chance of winning this is not their primary aim when setting the line. They are not trying to exactly predict the likelihood of an outcome but rather set a line that will encourage multi way action on the game. This then enables them to have relatively evenly split betting on all outcomes and they can take the built in margin.

Still these lines are a very good proxy for how a team is expected to perform and the cumulative odds can thus be used to extract just how many games a team won above, or below, what was expected. Thus we create an Expected Wins (Exp Wins) metric.

Expected Wins

All odds for a game were converted to an Exp Win (see methodology in Note2 below) and then teams ranked according to how many wins they obtained in the League & Championship above this mark

It comes as no surprise that seven of the top ten teams in pure win percentage appear in the top ten based on Exp Wins. Fermanagh and Monaghan are up there given their aforementioned successful seasons. Longford also had a good year winning 9 of their 13 games. In fact on pure winning percentage they finished second in the country behind Dublin’s 75%.

But what of the remainder? The biggest surprise by far was Limerick. They only won three games in total, ranking them in the bottom third on pure wins alone, but were 7th when compared to their Exp Wins. How so?

They were the outsider in all seven of their league games but won three. From those seven games the bookmakers expected them to win 1.87. They outperformed their expected wins by more than a full game. In the Championship they lost by two points away to Clare in a game that had Clare favoured by two and then walked into Tyrone in the first round of the back door. The positive Exp Win total they accumulated in the league was not too badly dented by these two losses – especially the Tyrone one where they were huge outsiders.

Sligo were a bit of a surprise given that they only won four games but again they were quite large underdogs when beating Roscommon in the Championship and complete outsiders in the next two games against Tyrone & Mayo. Given the very low combined Exp Wins from those three games (0.39) that one victory against Roscommon puts them in positive territory for the Championship alone.

Against the Spread

Another way of tracking a team’s performance is to see if they covered the bookmaker’s handicap; or what their ATS (against the spread) was in American parlance. We would expect some cross over with the best performers in the Expected Win list but crucially you don’t have to win a game to beat this performance metric – only play above an expected standard

Again six of the teams that appeared in the Exp Wins top ten re-appear. A number of the teams, such as Limerick, Sligo, Fermanagh & Monaghan we have touched upon previously but there are a few surprises. Mayo, despite being a very high profile team, would have been a profitable one to follow on the handicap. Cork, for all the negativity following the losses to Kerry & Kildare, were also profitable but it is London & Leitrim that jump out. Between them they won four games all season but it could be argued they had a pretty good year; their performance exceeded expectations in 12 of their combined 18 games.

London only won one of their nine games all year but managed to cover the handicap on six occasions. Narrow that further and they covered the handicap in five of their seven league games including all three that they played away. You would never state that London had a good season but from a performance perspective we should probably cut them some slack. They performed well above expectation.

Worst Performances

Originally the above table was going to be the bottom five but I expanded it to catch two of the bigger fish.

Some of the lower lights – Carlow, Wicklow & Waterford – being down here is not really a surprise given just how few games they won. However it does indicate that perhaps the bookmakers were generally over rating them despite their poor form.

Laois were particularly poor but looking purely at their Championship form they beat Carlow when their Exp Win was 0.86 so get very little credit for that and then had a further three games failing to win any of them when the combined Exp Win was 1.75.

Given they were relegated from Division 1 with just the one win from seven it is perhaps no surprise to see Tyrone down here.

Kerry won seven games throughout the year but were expected to win eight. Creating a league/Championship split Kerry had an Expected win of -0.81 in the league and -0.19 in the Championship. Their Championship was slightly less underwhelming than their league (I kid – sort of!)

Three of those that appeared in the worst Exp Win table re-appear when we look at the worst performances against the handicap. Wicklow and Waterford not only failed to win enough games but also played poorly in their losses covering a combined four handicaps over 18 games. Given that they won seven games but were only an outsider once during the year – and that a slight outsider in the final against Dublin – it is no surprise that Kerry are again represented.

They had, all told, a good year but were consistently over valued by the bookmakers. Or conversely the bookmakers kept their odds short as the public’s perception of Kerry was that they were performing better than they actually were.

APPENDIX

Note 1; there can be quite a difference in bookmaker’s odds. The odds used for this piece were taken primarily from Paddy Power rather than taking the best prices available across all bookmakers. The main reason for this was laziness on my part as it meant just one source rather than hopping around sites.

When you take the price can also be important. Lines do move. However they were generally taken on Saturday or Sunday morning when any early moves had been accounted for.

Note2; generally speaking the margin on GAA match odds is 109% with lesser games getting up to 112%. A typical line in a close game would be 10/11 (home team), 15/2 (draw) & 6/5 (away team) which equals a book of 109.6%. To make this, and all games, come in at 100% – and remove the bookmaker’s margin – I extracted 3% from each outcome. There is a valid argument that this should be more nuanced (take less off the draw perhaps) but for now it’s fine.

The home team has a 52.4% chance of winning on the odds. We know this is inflated to account for the bookmaker’s margin. Take 3% away from each of the three outcomes to account for this and the home team now has a 49.4% chance of winning. So using the above quoted odds we get an Exp win of 0.49 for the home team (priced at 10/11) and 0.42 for the away team (priced at 6/5).

Do this for all games for a particular team and you have created an Expected Wins metric.

Data Request

Update: Thanks for all the help. One of the traders from Ladbrokes has been on and provided the data

Quick post looking for data

During 2015 I tracked the betting odds on all league and Championship games. Or at least I thought I did. The plan was to do a piece comparing all teams’ expected wins versus how they did and see who outperformed their odds throughout the year. Or underperformed.

Having transcribed the data into a spreadsheet I realised I missed a number of games. If anyone has the odds, or the handicap (I can work out one from the other), from the below games I would be very grateful. Drop me an email at dontfoul@gmail.com or leave them in the comments

Thanks
James

Division 4
London – Leitrim
Longford – Offaly (the final)
Dublin – Kerry in the AI Final

(yes yes I know. I tracked all the games then forgot to note the biggest game of the year)

Dublin’s 2015 Goal Attempts

Dublin have always gone for goal at a higher rate than other teams. Things were no different in 2015. They made up 13% of the competitors in the 26 games recorded but were responsible for 23% of all goal attempts. The attempts were not scattergun either as at a Conversion Rate of 53% (18 from 34) they maintained the average whilst attempting much more than anyone else. So is there anything we can learn by reviewing their 2015 attempts?

Origination

Where do Dublin’s goal attempts originate from?

19 came from possession gained on a kickout; nine from their own and ten on the opposition’s. 13 attempts came from turnover ball with the remaining two coming from Dublin shots that went astray – both from Brian Fenton incidentally (McManamon’s scramble against Mayo in the drawn game & Fenton’s cross shot – in the replay – that was guided in by Brogan).

What was noticeable just watching the goals back to back was the speed at which Dublin break. Of the thirteen turnovers that produced an attempt nine began inside their own 45. Add these to the nine from their own kickout and that is 53% (18 from 34) of their goal attempts starting from a position that the opposition should be in a position to defend.

But it’s the speed of transition that does for teams. The average for these 18 attempts, from gaining possession to taking a shot, is 20.3 seconds and 5.4 passes. We have nothing to compare this to but next time you are watching an intercounty side gain the ball inside the 45 count to 20 seconds (or 5 passes) and you will soon see how quick that is. And that is the average!

Dublin will not be that quick with every turnover, or kickout won, but the intent is always there. And when it is on they go. This is where McCaffrey’s transition speed, and Kilkenny’s accurate foot passing in the middle third, are hugely beneficial.

Speed of transition is further emphasised by the ten attempts generated off the opposition’s kickouts. Again they will not always be this quick but the first recipient has his head up looking for the forward ball. On the ten attempts the average time elapsed was 11.4 seconds incorporating 4.2 passes.

Results

Below are the outcomes of the 32 attempts from play; the original 34 included two penalties that were converted.

 

There are two things to the above. The first is the very nice cluster of goals Dublin had on the edge of the small square. The second is to note that this reflects a mixture of individual accuracy as well as team play. Of the 32 goal attempts five are fisted whilst another four are scrambles where the ball shot was instinctive rather than planned.

It says a lot about Dublin’s general attacking intent, and support play, that there are players in a position to fist the ball in or to be the first onto these scrambles. But if we are trying to decipher the Dublin players’ accuracy we need to remove these. Below is what the goal attempt chart looks like with these nine removed. A much reduced return of 35% on 23 shots.

 

The more I do this, and the more granular data we get our hands on, the more obvious it becomes that averages hide a lot. So any outcomes – whether it be weightings or Expected Points – used on the blog needs to be always challenged. In the last four years 36% of all goal shots were converted but what proportion of those attempts were fisted? Under pressure? Scrambles? Is 36% a fair representation of shot accuracy?

Post script – anything else on the Dublin shots?

• Thirteen different players had a shot at goal across the seven games
• It is hard to say from the camera angles how many were on target but only 1 of the 32 attempts from play went wide. Three were blocked, six saved, 1 hit the post, 4 went for a point whilst another was diverted in (the aforementioned Brogan toe poke on Fenton’s cross shot)
• Outside of the goal McMahon bundled into the net against Mayo he took two further shots. And scored a point with both – keep the ball down Philly
• Excluding fisted attempts & scrambles (the 23 attempts in the second chart above) only six (26%) were attempted under any form of defensive pressure

A few non game related updates/musings

Amendments
I had completely forgotten that people get blog entries through email. It wasn’t until my brother texted me to tell me that he never visits the blog (thanks bro!) and gets everything through email that I remembered there was an email list.

Why this is important is that I have gotten into a bad habit of putting entries up quickly and then fixing them on the fly – but the first, usually incorrect, entry goes by mail. Apologies for that – I’ll try and take an extra hour or two and get the first entry right.

Definitions
I have updated (finally) the definitions page (here). Feel free to rummage and find any errors/omissions.

This weekend (July 18th/19th)
One of the issues with this being a hobby is that real life gets in the way. I am going to be out of the country over the coming week so won’t get to this weekend’s games except on a delayed basis. There are three huge games in the Provincial finals. Given the history we have to date it is likely I will do the Cork – Kerry game first and then choose between Mayo – Sligo and Donegal – Monaghan depending on how the games go. Though keep an eye on the Ulster GAA website as they have an excellent (and much cleaner looking!) breakdown of all their games. Donegal – Monaghan should be out mid-week.

Using the stats
I don’t mind people using the stats. I don’t. Anyone who has interacted with me will tell you that – people don’t have to ask to be honest. Just credit the original source as it does take some time to collate.

Generally what I do is not proprietary. It is basically glorified counting and segmentation (apart from the weighting) thus anyone could do it. And that’s the problem. There have been a few instances where you *know* the data has been lifted from the blog and I look like an attention seeking teenager looking for notice when you query where the numbers have come from. Then there are times when the person has obviously gone through the game and come to the same numbers/conclusions. It can become tiresome (a) trying to decipher who has lifted the numbers and (b) seeing yourself as an attention seeking teenager. Tis many, many years since I was that!!

If you want to use any of the data go ahead. Just credit the source – please!

Colm McFadden – a review

The last time we saw Donegal was in the league semi-final against Cork which was fairly unremarkable apart from both teams’ ability to get shots off at a high rate – in the mid 80% against an average of 77%. It is only one metric but a fairly large indicator of a game lacking intensity.

One thing that did peak my interest however was the performance of Colm McFadden. In a high scoring open game he only managed 0-02 from play and did not attempt any frees. I had a vague recollection of noting a declining performance throughout 2014 so decided to have a look. Thanks to the BBC’s coverage of the Ulster Championship we have all bar one of Donegal’s game over the past three years and below are McFadden’s returns over that span.

	Attempts	Scores	Success Rate	Weighting
Deadball	53	40	75%	+3.499
Play	52	23	44%	-0.196
Total	105	63	60%	+3.303

Not bad. His shooting from play is bang on average but given the volume of shots he takes that’s ok. Average does not equal bad! His deadball accuracy however is very good – perhaps only below Cluxton and O’Connor in terms of per kick weighting. In truth there’s nothing to see here but the more you do this the more you understand that averages can hide a lot.

	Attempts	Scores	Success Rate	Weighting
Deadball
2012	27	22	81%	+4.566
2013	13	8	62%	-0.770
2014	13	10	77%	-0.297
From play
2012	24	13	54%	+2.226
2013	11	6	55%	+1.987
2014	17	4	24%	-4.409

McFadden was stupendous, both from deadballs & from play, in Donegal’s march towards Sam in 2012. But he was so good on that run that when we take the average from the last three years it is masking a very big drop off in both categories.

McFadden’s deadball accuracy has dropped from the unsustainable highs of 2012 to just below average over the last two years. Although his combined Success Rate for 2013 & 2014 is high at 77% the negative weighting shows you that he is taking easy, or at least easier, frees – he should actually be converting closer to 80% (similarly the high weighting in 2012 shows you that he was converting more difficult frees more often).

The real drop off has occurred with his shooting from play however.

	Attempts	Scores	Success Rate	Weighting
Point attempts
2012	21	11	52%	+1.240
2013	8	4	50%	+1.001
2014	14	3	21%	-4.395
Goal attempts
2012	3	2	67%	+0.986
2013	3	2	67%	+0.986
2014	3	1	33%	-0.014

McFadden was never prolific from play but he was always above average in his returns. That was until last year. During the six 2014 games McFadden had 14 attempts at a point – and only converted three. And all three were simple attempts from around the penalty spot. He did not convert one point attempt from outside the 20m line in six games.

McFadden’s 2014 point attempts

It was not as if he was trying harder shots in 2014 – the opposite actually. Of his 14 point attempts in 2014 ten, or 71%, were in the optimal shooting zones of 5 & 8 (see the shot chart above). In 2012 & 2013, when he was converting at ~50%, only 35% of his shots were coming from these central areas. There was nothing in terms of pressure to explain the drop off. Last year 57% of his shots were taken under pressure – in 2012 & 2013 that figure was 66%.

The data on file begins in 2012 which is unfortunate for McFadden. It was, as Ciaran McMonagle of the samsforthehill blog pointed out to me, a career year for McFadden when Donegal were purring and they were playing to his strengths. 2013 was a bit of a washout for Donegal but in this case was instructive as to how easily Drew Wylie handled McFadden in the Ulster Final. Monaghan were on the lookout for McFadden’s favourite play – the loop inside from the right hand side curving the ball over with the left – and completely snuffed it out.

2014 was a down year but was that McFadden or the way Donegal set up? The loop play disappeared but a lot of that was due to the fact that Murphy spent considerable chunks of time in the middle allowing closer attention on McFadden. Alongside that we saw the further emergence of two left footers in MacNiallais & McBrearty. McFadden’s space narrowed, his foil was missing and he’s not blessed with speed. Not a great mixture.

What does 2015 hold? Usually when you have big swings in data like this you’ll get a regression to the mean but will McFadden be given the space, and as importantly the game time, to rack up the shots? Interesting year ahead.

Dropping shots into the goalkeeper’s hands – is it that bad?

Something that appears to drive both coaches and fans nuts is a forward dropping a shot into the opposition goalkeeper’s hands. The apparent calamity of a missed opportunity is enhanced by enabling the opposition to attack you when your defense isn’t set and you are vulnerable. It’s a double blow.
But is it really that bad?

Firstly just how often does it happen? In 23 Championship games last year I recorded a shot dropping into the goalkeeper’s hands on 80 occasions. That is less than two shots per team per game – 1.74 to be exact (80/(23*2))) – not exactly huge.

In total the opposition received the ball from a turnover inside their own 20m line on 428 occasions. These turnovers are broken into three types for the purposes of this piece

• Shots
(a) that drop into the goalkeeper’s hands
(b) that somehow end up with an opposition player other than the goalkeeper (a block picked up, a shot dropping short & the opposition picking the ball up, off the post etc.)
• other turnovers (tackle, pass) where the opposition picks up the ball

The outcome of the possessions from these specific turnovers are outlined below

	# Turnovers	Return	Pts per t/over
Shots – into goalkeeper’s hands	80	0-22	0.28
Shots – other turnovers	94	3-23	0.34
Other T/overs inside 20m	254	2-60	0.26

Ignoring all other factors a ball kicked into the goalkeeper’s hands is no more dangerous than any other shot type that the opposition gets their hands on – and is only on a par with all other turnover types received inside the 20m line. Somewhat surprising. Averages hide a lot – and we are dealing with small volumes – but perhaps this is more of a reflection on goalkeepers than the attacking potential of the turnover. Goalkeeper’s, in the main, are not springboards for attacks but rather keepers (pun intended) of the flame whose primary job is to hand pass the hot potato to the first back that looks for the ball.

Of course there is an element to dropping a shot short that the above does not cover and that is the foregoing of a kickout. When a shot is dropped short you cannot gain any benefit from the possession that immediately ensues. This is most definitely not the case from a kickout.

Now we know that there are ways and means of controlling the kickout however at a macro level below are how teams fared in the above 23 games

	# kickouts won	Return	Pts per t/over
Kickout team	760	10-214	0.32
Opposition	371	14-107	0.40

The kicking team wins 67% of the kickouts scoring 0.32 points per kickout. The opposition wins 33% of all kickouts with a scoring rate of 0.40 points. If you drop a shot short it is this element that you are giving up.

Taking it all together; if a team drops 100 shots into the opposition goalkeeper’s hands they will give up 28 points (100 * 0.28). If they kick it wide they will give up a net ((67*0.32) – (33*0.40)) 8 points. That’s a rather large gap.

So there is a drawback – not necessarily from dropping a ball into the goalkeeper’s hands but giving up a turnover from a shot as a whole. It is not that you are giving a springboard for a counter attack when your defence is not ready but instead you are foregoing the ability to score off the subsequent kickout.

So the question now becomes do you change a player’s shooting mind-set to ensure all shots go dead so that you gain, on average, 0.35 (1.74 * (0.28-0.08)) points in a game? Or is the gain so marginal as to not be worth the risk of changing your player’s style?

Attack Origination

One of the new metrics introduced during the 2014 Championship was the tracking of where attacks originated. It was not captured in the expectation of finding something but rather an exploratory piece just to gain some more insight into how a game flows. It is important to state that what this does not do is track how effective teams are with the ball. It is more a measure of whether a team’s ability to defend an attack is dependent on where the attack originates.

Attack Origination

Origination	Attacks	% of attacks
Own kickout	451	31%
Opp kickout	253	18%
Own 3rd	426	30%
Mid 3rd	153	11%
Opp 3rd	54	4%
Other	99	7%

Own, Mid & Opp all relate to gaining the ball from turnovers in a specific area of the pitch. Other includes gaining the ball from retrieving your own team’s wayward shot, throw ins etc.
Own 3rd = inside your 45, Opp 3rd = inside the opposition’s 45 whilst Mid 3rd is the area in between the two 45s.

A couple of things immediately pop out.

Just under half of all attacks originate from kickouts with 31% coming from your own kickout. Kickouts are one of the few set plays within the game and an ability for a team to direct how primary possession is to be delivered. It is of course essential that teams win this primary possession – but they must also have a plan for what happens next. Too often the focus is on who “won” possession rather than what teams did with that possession.

The vast majority of turnovers that turn into an attack originate inside your own 45. This is in no way surprising as that is where the majority of attacks are engaged – it also includes all those shots that drop short – but it still shows the value of your transition game.

Shot success by Origination

	Attacks	Shots	Shot rate	Scores	Score rate
Own kickout	451	347	77%	187	54%
Opp kickout	253	200	79%	99	50%
Own 3rd	426	331	78%	169	51%
Mid 3rd	153	121	79%	54	45%
Opp 3rd	54	47	87%	30	64%
Other	99	86	87%	48	56%

The general homogenous nature of the shot rate (apart from when you turn over the ball inside the opposition’s 45) is a surprise.

I would have expected that balls turned over in the middle 3rd would have caught the defending team in less of a defensive set and thus made getting a shot off easier. This does not appear to be the case and is further confused by the fact that the shots that are taken are of a much poorer nature with only 45% converted. Could it be that after winning the ball rather than drive at the opposition the first thought is to secure the hard won possession and thus allow the defence to stream back? This would account for the similar shot rates but not necessarily the poorer score rate.

The shot rate for balls won from either your own or the opposition’s kickout is very similar. This in a way feeds into the strange homogenous nature of the shot rates. At a macro level you are going to gain possession from the opposition’s kickout further up the pitch than your own – plus if you win the opposition’s kickout they are more likely not to be in their full defensive set (anticipating winning their own kickout themselves). The fact that more shots are not eeked out from winning the opposition’s kickout is again slightly surprising.

Teams score 4% more of the time from shots that originate from the opposition’s kickout. Whilst an increase of 4% might be within any margin of error it does make sense in the aforementioned context of the defence not being as set on their own kickouts.

Shot success by Origination

	Kerry		Donegal		Mayo		Dublin		Other
	Attacks	% all attacks	Attacks	% all attacks	Attacks	% all attacks	Attacks	% all attacks	Attacks	% all attacks
Own kickout	44	22%	25	18%	31	19%	45	21%	110	15%
Opp kickout	50	25%	51%	36%	42	26%	60	28%	248	34%
Own 3rd	59	29%	51	36%	41	26%	54	25%	221	31%
Mid 3rd	28	14%	7	5%	21	13%	24	11%	73	10%
Opp 3rd	4	2%	1	1%	12	8%	14	7%	23	3%
Other	17	8%	7	5%	12	8%	16	8%	47	7%

As ever on these pieces Donegal are the outlier. As a result of not engaging outside their own 45 only 6% of their attacks originated from turnovers further up the pitch. This compares with an average of 13% with returns of 15% (Kerry), 21% (Mayo) and 18% (Dublin) for the other semi-finalists.

Donegal were also very reliant on attacks from their own kickout. There is a concertina effect here as the relative lack of attacks from turnovers further up the pitch places more emphasis on the remaining categories however the ratio of attacks for Donegal from their own kickout to the opposition’s is 2:1. This is in line with the non semi finalists average of 2.3:1. The other top teams were much better at attacking the opposition’s kickout with ratios of 1.1:1 (Kerry), 1.4:1 (Mayo) and 1:3.1 (Dublin)

It is this, attacking the opposition’s kickout and getting opportunities from same, that might be one of the key strengths of the “better” teams.

Top performances over the last 3 years

The new weighting, taking into account the returns from the past three years, has been applied to the 74 Championship games in the database. Below are some comparisons of teams/players/games over that period.

Top 5 Shooting Performances

Player	Weighting	Game	Year
C O’Neill	4.065	Cork V Sligo	2014
R Munnelly	3.150	Dublin V Laois	2014
C O’Connor	2.885	Mayo V Donegal	2013
J O’Donoghue	2.555	Kerry V Galway	2014
J O’Donoghue	2.547	Kerry V Cork	2014

C O’Neill hit a wonderful 10 from 10 in Cork’s Rd4 qualifier against Sligo in Tullamore last year. He was 6/6 from play with three strikes coming from outside the 20m line whilst he converted 4/4 from deadballs including a 45. Yet it could have been improved upon – one of his conversions was a goal shot that went over the bar (just saying!)

Ross Munnelly’s game against Dublin stands out not only for its quality but also for the fact that he did it on a team that eventually lost by 11 points. He was immense that day.

We’ll touch on O’Donoghue anon.

 

Top 5 Overall performers

Player	Total shots	Success Rate	Weighting
J O’Donoghue	63	73%	12.937
P Flynn	48	56%	7.507
C O’Connor	119	68%	7.468
C O’Neill	50	64%	6.006
C Cooper	51	73%	5.534

Given that O’Donoghue has two of the top five one off performances it should be no surprise to see him sitting comfortably atop the overall table. Part of why he is so far ahead of everyone is the fact that he has only returned one game with a negative weighting – this year’s All Ireland final when he missed his one shot.

O’Donoghue is running at a remarkable 73% conversion rate – as was Cooper throughout ‘12 & ‘13. Scary thought for those trying to topple Kerry what with the imminent return of Tommy Walsh as well.

Flynn’s weighting is huge on such a low Success Rate. This is due to where he shoots from; he has had 41 point attempts charted and only two were from inside the 20m line. Sectors 4 & 5 are his playground; he has a Success Rate of 69% on these (20 from 29).

Re Dublin it shows at times how easy the mopping up duty has been for them. D Rock, who mainly has had cameo roles coming off the bench, is 7th on the overall list converting 80% (20 from 25) of his shots for a weighting of +5.196.
O’Connor’s accuracy, over so many shots is noteworthy. Only B Brogan & C McFadden are credited, alongside O’Connor, with more than 100 shots.

 

Top 5 from play

Player	Weighting	Game	Year
J O’Donoghue	3.041	Cork V Kerry	2014
B Brogan	2.929	Dublin V Louth	2012
C O’Neill	2.621	Cork V Sligo	2014
E Keating	2.601	Cavan V Donegal	2012
J O’Donoghue	2.555	Kerry V Galway	2014

Unsurprisingly three of the top5 overall games reappear. *If* we had adjustments based on the defence you faced then I’m sure Keating’s performance against the Donegal wall in 2012 would come out on top. He was genuinely remarkable that day hitting 5/6 with four of his five scores coming from outside the 20m line. He had one free from the right hand side out in Sector6 – had he converted that it would have been the 2nd highest scoring performance ever behind O’Neill.

 

Cluxton & 45s
Stephen Cluxton has converted 75% of his 45s (21 from 28). The Rest of Ireland combined is running at 39% rate (38 from 97). That difference is phenomenal.
Cluxton’s 28 attempts make up 22% of all 45s recorded – I can’t imagine any other player has such an impact on any other slicing of the database.

 

Top rated deadball games

Player	Weighting	Game	Year
S Cavanagh	2.382	Tyrone V Meath	2013
S Cluxton	1.964	Dublin V Cork	2013
C McFadden	1.687	Donegal V Cavan	2012
C O’Connor	1.674	Dublin V Mayo	2012
S Cluxton	1.623	Dublin V Mayo	2012
S Cluxton	1.568	Dublin V Kildare	2013

I’m obviously showing the Top6 just to get three Cluxton games in!

I guess it is no surprise to see Cluxton up there so often given his prowess from 45s though it is worth noting that he is only converting at 43% (9 from 21) for free kicks.

Cavanagh hit a rich vein of form in 2013 from frees. He was 3 from 3 against Kildare in Newbridge then hit 6 from 6 (all from Sector 6!) in Croke Park in the next game against Meath.

 

Top team performances

Team	Weighting	Game	Year
Mayo	6.813	Mayo V Donegal	2013
Cork	6.202	Cork V Kildare	2012
Mayo	5.408	Mayo V Dublin	2012
Mayo	4.629	Mayo V Galway	2013
Donegal	4.248	Donegal V Down	2012

No forwards how are ya! Mayo occupy three of the top five positions with the standout being their 2013 QF demolition of Donegal.

One of the oddest “oddities” from doing this was the fact that Kerry’s shooting in the All Ireland final was the 3rd worst out of 148 entries – they were truly appalling shooting wise

Team	Weighting	Game	Year
Kerry	-4.389	Kerry V Donegal	2014
Roscommon	-4.604	Roscommon V Mayo	2013
London	-4.697	Mayo V London	2013

 

Top 5 games

Game	Weighting	Year
Mayo V Donegal	7.337	2013
Derrv V Down	6.843	2013
Dublin V Mayo	5.487	2012
Kerry V Mayo	5.452	2014
Dublin V Kerry	4.793	2013

Sooo – hands up who had an Ulster QF down as being one of the Top5 most accurate games? Didn’t think so! The remarkable thing about Mayo’s annihilation of Donegal in 2013 was that shooting wise Donegal didn’t capitulate. Usually in these one sided hidings one team is just atrocious – not on that day.

Two modern day “classics” – the Dublin-Kerry SF in ’13 and the drawn SF between Kerry & Mayo this year – are represented. Sometimes the stats do back up what the eye sees. I’m still flummoxed how Kerry could be involved in those games and then produce what they did in this year’s final!

 

Worst recorded game

Game	Weighting	Year
Dublin V Wexford	-6.392	2012
Mayo V London	-5.058	2013
Mayo V Roscommon	-4.871	2013

No need to elaborate really. We have seen from above that Roscommon & London produced stinkers but here it shows that Mayo didn’t shoot the lights out to rescue the game. Given their lofty status in the top five games they were basically brought down by the paucity of the opposition. The Dublin Wexford game was just atrocious – both teams were as bad as each other ranking 120th and 144th on the overall list.

 

Most shots from deadballs

Game	Attempts	Year
Meath Tyrone	23	2013
Dublin V Meath	20	2013
Down V Monaghan	19	2012
Dublin V Laois	19	2012
Meath V Wexford	19	2013
Laois V Meath	19	2012
Donegal V Monaghan	19	2014

Ha. So four of the top seven games in terms of shots attempted from deadballs involve Meath. I was going to make a joke about the spirit of Mick Lyons living on but in two of those games Michael Newman had 11 attempts – showing that Meath were as much sinned against as sinner. Newman is the only player to have 11 attempts in a game (Brian Farrell & Cillian O’Connor managed 10) so it is likely that the high Meath counts are more an indication of Meath’s confidence in Newman’s distance than anything else.

Not sure what exactly the players in the 2012 Munster final were on but only five shots were attempted from deadballs in that one

Review round up

And so we come to the yearly comparison piece whereby this year’s performances, numbers wise, are stacked up against preceding years.

In previous versions of weightings 2010 data had been used, in conjunction with 2012 & 2013, so as to create more volume (and as the reasoning goes more accuracy). From here on all 2010 data will be omitted. There is no great reason for this, as the averages more or less stack up, but it just feels more complete to have continuous year’s returns rather than an outlier with a year’s hiatus in between. At this juncture, after taking a year out in 2011, there is no guarantee that the 2010 methodology was the same as is used now. 2010 was the first year of doing the tracking – I’m pretty sure issues were being ironed out throughout that year so 2012 data onwards just feels more robust.

With that said here are the numbers

Year	Attacks	Shots	Shot Rate	Scores	Success Rate
2012	35.28	27.02	76.6%	13.96	51.7%
2013	36.32	28.26	77.8%	14.26	50.5%
2014	39.83	30.88	77.5%	15.85	51.3%
averages
2012/2013	35.8	27.6	77.2%	14.1	51.0
all	37.0	28.7	77.7%	14.7	51.2%

Essentially Shot & Success Rates for 2014 stayed very close to 2012 & 2013 levels – once teams had the ball inside the opposition’s 45 the ultimate overall ratio of shots & scores from those possessions stayed the same.

The difference comes in the volume of attacks. There was a 9.7% increase per team in attacks on the 2013 returns. This is quite significant. Of course if the volume of shots and accuracy of shooting is constant on an increased volume of attacks you will get more scores; and that’s what happened with an increase of 1.6 scores per team per game.

So why have the volume of attacks increased so dramatically? Many put the increase squarely at the feet of the black card. It has undoubtedly had an affect but I’m a firm believer that this is a copycat game and Dublin’s successful swashbuckling style throughout 2013 & 2014 will also have had an impact on how team’s approach the game. Although smaller there was an increase of 2.9% without the black card in 2013.

If it was all the black card I would expect that what happened inside the 45 would be as affected by what happened outside. Players would have more time/space to shoot – less people pulling at their arm/jersey. The numbers don’t bear this out however. Yes the black card allowed for less attacks to be stopped by illegal defending up front but there was no boon to the shooters.

Deadballs

Another area where the black card has been given credit is the reduction in the volume of frees. Yes there was a decrease in the volume of deadballs per game in 2014 but this decrease was only back to 2012 levels. Is this decrease then the black card affect or simple variance?

I am not for one moment saying the black card is a bad idea. Or that it has had no effect. I would argue that there are many variables that lead to a shift in trends of which the black card is one. And trying to quantify the value of that one variable is incredibly difficult.

Year	Shots	Scores	Success Rate
2012	6.94	4.64	66.9%
2013	7.78	5.18	66.6%
2014	6.75	4.94	73.1%
averages
2012/2013	7.36	4.91	66.7%
all	7.16	4.92	68.7%

Some of the more seasoned observers will note the increase in the deadball Score Rate. One of the central planks of the data to date is that your deadball strikers will return a 67% Success Rate. Even 2010 returned a 66.3% Success Rate. This has jumped to 73% this year – for more detail on why this is please see here

So what of shooting from play?

Year	Shots	Scores	Success Rate
2012	20.08	9.30	46.3%
2013	20.48	9.08	44.3%
2014	23.79	10.77	45.3%
averages
2012/2013	20.28	9.19	45.3%
all	21.42	9.70	45.3%

Essentially there was no change in the accuracy; it came in bang on the 2012/13 average.

I did a quick comparison of shots for points versus shots for goal. Some of the noteworthy outcomes
• The Success Rate for point taking in 2014 was more or less the same as 2013 – 44.8% versus 44.7%.
• The overall increase in Success Rate from play was almost entirely due to goal shots being converted at a higher level; 48.6% in 2014 versus 41.9% in 2013
• One of the more volatile ratios is point attempts per goal shot; 2012 = 7.58, 2013 = 6.53 & 2014 = 7.13. Generally speaking the increase in shots from play was more weighted towards goal shots but not in any dramatic sense

Weighting

Some of the deficiencies in the weighting were outlined earlier in the year (here) – especially when it came to less frequent events like penalties, attempts from sideline balls and goal shots that go for a point.

Due to a lack of data, both in terms of not capturing data in 2010 and events occurring infrequently, some subjective judgement had been applied to the weighting in these cases. Although the rationale behind the judgement calls made sense (to me at least!) it is something I’ve always wanted to remove. With three consecutive years’ worth of data, where the same methodology has been consistently applied, that is now possible.

That is not to say the new weighting is complete – there are still some very small volumes on penalties, sideline balls etc. but it is better to just take the numbers and deal with any issues that arise thereafter rather than try to overlay a subjective value.

On top of stripping out any subjective overlay the 2010 data has been removed. The below tables show the differences between the weighting employed in 2014 and what will be applied going forward

From play

Sector	New Weighting	Old Weighting	Difference
1	0.667	0.707	-0.040
2	0.605	0.623	-0.018
3	0.750	0.698	0.052
4	0.628	0.641	-0.013
5	0.501	0.501	-0.001
6	0.653	0.641	0.012
7	0.579	0.575	0.004
8-point attempt	0.288	0.292	-0.004
8-goal attempt (goal)	0.662	0.593	0.069
8-goal attempt (point)	0.288	0.593	-0.305
9	0.538	0.683	-0.045

From deadball

Sector	New Weighting	Old Weighting	Difference
1	0.600	0.643	-0.043
2	0.526	0.519	0.008
3	0.731	0.629	0.102
4	0.379	0.453	-0.074
4 – 45s	0.524	0.565	-0.041
5	0.163	0.160	0.004
5- 45s	0.506	0.500	0.006
6	0.397	0.428	-0.031
6 – 45s	0.611	0.565	0.0.46
7	0.180	0.232	-0.051
8 – frees	0.064	0.087	-0.023
8 – penalties	0.182	0.593	-0.411
9	0.211	0.227	-0.016
Sidelines	0.688

The weightings have not moved all that much with the main differences coming around goal attempts.

Previously if a player went for goal and got a point he got the same weighting as if he’d scored the goal. This was slightly absurd so now in that instance a player will get the same weighting as if he’d gone for a point. Probably not ideal (he should get some extra bump for going for goal) but I think it’s better (or less bad!) this way. Similarly a player got the same weighting for scoring a goal from a penalty as from open play. Given that penalties are converted at 82% this over sold a converted penalty. The weighting now reflects this conversion rate.

In the coming days I will overlay this new weighting on all games from 2012 onwards and see what shakes loose.