Archive for the ‘AI Series’ Category

Dublin v Tyrone 2018 All Ireland Final

September 6, 2018

This may seem like a ridiculous statement when reviewing a game in which the four time Champions won by six points but Dublin were just better. We’ll go into the various components below but they had more primary possession, more shots and better accuracy on those shots.

More ball plus more efficiency in using that ball equates to a relatively straightforward win.

When Dublin had the ball

What makes the margin of victory more impressive was the hole that Dublin had dug for themselves. After 16 minutes they were 12 – 9 behind on the possession count, 10 – 7 behind on the shot count and four points behind on the scoreboard. They were losing the primary possession battle and returning a hitherto fore unseen 14% (0-01 from 7) Conversion Rate. This also included Dean Rock missing two frees. The same Rock who recorded an 83% (0-25 from 30; Expt Pts of +3.15) Conversion Rate on frees in All Ireland finals and semi-finals in 2016 and 2017. It is lost in the talk of five in a row but that was a dreadful start reminiscent of their start against Mayo in 2016.

Of course combining the end result with that terrible start (and it was terrible) means that when Dublin removed the handbrake post the 15th minute they then proceeded to destroy Tyrone in the following 65 minutes winning the possession battle 41 – 32, producing five (25 to 20) more shots and returning a Conversion Rate of 72% (Expt Pts + 5.24 and a points per possession of 0.54). Utterly devastating. The below time sequence chart shows the gap, in terms of score and Expt Pts, just widening from that first goal.

The controlled nature of the second half display (0–10 from 14 (71%) and winning all 16 kickouts) is very reminiscent of the 2017 final when they went 0-12 from 16 attempts (Conversion Rate of 75%) and secured all 11 kickouts. Something similar was in effect in the semi-final against Galway (time series chart in the Appendix). Had Galway taken their chances at even an average rate they would have been with Dublin well into the second half – but still Dublin powered away with excellent second half shooting.

So how did Dublin do it?

First and foremost they converted their goal chances. They had three shots at goal returning 2-00. Even when Morgan saved Costello’s attempt Rock popped up with the subsequent 45.

Speaking of Rock he had a sterling day scoring 0 – 07 from 9 including 0 – 03 from 3 from play (Expt Pts of +1.33). That is even better than the bare numbers when we consider that his three points from play came in the first half after missing those aforementioned early frees. Ciarán Kilkenny clipped over 0-03 from 5 (Expt Pts of +0.69) whilst Brian Fenton popped over his two attempts.

The Dublin shot chart is in the Appendix. The shooting wasn’t as tight as was highlighted in the preview (only 9 of the 23, or 39%, were from the artificially created central channel) but it was tight enough with next to no wild attempts. Whilst Tyrone have to be given credit for keeping Dublin outside of this central channel it is an indication of Dublin’s clinical nature that of the 14 point attempts from “outside” they converted 57% (0-08 from 14)

Not only did Fenton score 0-02 but he was also very active in the attack with four primary assists. These assists are probably more a testament to his conditioning, and relative cool, as three of them came in injury time at the end of the second half. The epitome of this Dublin team – the right option executed properly when under the twin pressures of fatigue and an opponent attempting to manufacture a comeback.

Having said all that the unsung hero for Dublin was Con O’Callaghan. He was the primary assist for seven separate shots including for Scully’s goal and winning three of Dublin’s five frees all in the first half when the game was there for the taking.

When Tyrone had the ball

In the preview it was highlighted that whilst, in the run up to the final, Tyrone’s Conversion Rates were on a par with Dublin there were reasons to believe that theirs was a false position being propped up by an unsustainable outing against Roscommon. And so it proved.

Tyrone scored 1 – 14 from attempts that the average intercounty team would return 17.22pts. The fact that every team doesn’t face Dublin when those averages are being compiled gives Tyrone some leeway – so let’s say they returned in and around what was expected. Tyrone did not lose because of their wides or due to any perceived inefficiency. Their Conversion Rate, excluding the Roscommon game, coming into the final was 52%. Their average from ’15 – ’17 was 52%. They shot 50%. It was always going to be thus.

Tyrone lost because they didn’t produce enough shots (or alternatively reduce Dublin’s volume of shots) knowing they would be up against the most efficient team we have ever seen. Or as stated in the very first sentence because Dublin are just better.

How Tyrone compiled their returns is slightly at odds with how we would expect. We knew coming in to the game that despite having four goal attempts in the recent Super8 contests against Roscommon and Dublin they were struggling to create chances. They had only manufactured three goal attempts in their three previous games against Dublin. Now we can make it four in four with two of those being penalties. They did maintain their recent fine record of converting goal chances. In the five games from the Super8 onwards they have had 10 attempts at goal scoring 8-01. That’s a phenomenal return (the average over the years has been creeping up to about 40%) but is it indicative of a too cautious approach at an average of 2.0 a game? Only taking the shot when it is absolutely on? Dublin were 4.0 a game in the same period.

Frees were an issue coming into the game (averaging out in the mid 60% range) though given Dublin’s defensive discipline there was the possibility it wouldn’t be an issue. Dublin did maintain their discipline (again only offering up five shots at goal from frees with only two in the first half) but Tyrone were excellent here converting all five using the rota of Harte, McAliskey & Lee Brennan.

So six shots at goal from frees and deadballs returning 1-05 (Dublin were 2 – 04 from 10). What let Tyrone down was their point taking. At a macro level they were 38% (0 – 09 from 24) with an Expt Pts of -1.92. Dublin ended up with 57% (0-13 from 23; Expt Pts of +1.88) despite their very poor opening.

Their two strike forwards of Bradley & McAliskey were good combining for 80% (0-04 from 5) with an Expt Pts of +1.79. Of course that means everyone else was very, very poor combining for 26% (0-05 from 19; Expt Pts of -3.44). That’s just not going to work against Dublin.

In the preview we highlighted how Tyrone’s shot location was hampering them. Here Tyrone had some poor efforts (three shots in a knot at the 20m line out by the right sideline) but overall 45% of their point attempts came from the aforementioned central channel. Right place but just poor execution; think of Lee Brennan’s snapped shot wide with his right @ 67:40, Cathal McShane @ 52:29 with his left about 16m out directly in front of the goals

Tyrone will have regrets, and thus something to build on, as their shooting was an absolute mixed back. At a macro level their accuracy was not a problem – they hit the average. But unlike previous form their free taking held up so their accuracy from point taking was an issue. But even therein their strike forwards were accurate – everyone else was very poor.

Tyrone’s shot location was another mixed bag – more shots from central locations should have helped the Conversion Rate but they had some very poor attempts from “inside” mixed in with some absolute haymakers from out wide.

Kickouts

In all the above what we have yet to touch on is just exactly how Dublin built up +6 on the possession count. And it is quite different than a “normal” Championship game.

Tyrone won the turnover battle quite comprehensively 23 – 13 which in and of itself is striking. So how do you lose a game by six points if you gain 10 more possessions from turnovers? By getting cleaned out on kickouts.

The possession count from kickouts was 36-19 in favour of Dublin. Whilst remarkable using a term like “cleaned out” may be hyperbolic as 13 of those 17 extra possessions came from short kickouts – and once again Tyrone dealt very well with Dublin’s short kickouts allowing just 0.23pts per kickout won. Tyrone allowed 0.28pts in the three games prior to the final. This is against the 0.47pts allowed by other teams in the run up to the final.

Tyrone, it would appear, can handle Dublin’s short kickouts. What we saw in the three previous games was that they couldn’t handle the longer ones with Dublin scoring 2-13 off 34 won past the 45 (0.59pts per kickout won – compare that to how they stifled Dublin on short ones). Again here Dublin scored 2-04 off just 14 won past the 45; 0.71pts per kickout won.

Apart from only breaking even (winning 8 to Dublin’s 7) on their own kickouts past the 45 the other glaring number from a Tyrone perspective is that they only got two shots from the nine short kickouts (22%) they reclaimed. This seems to have been a one off as in the three games against Dublin they created a shot on 64% (18 shots from 28) of their short kickouts

Appendix

Time series chart v Galway

Dublin shooting chart


disc = scored, x= missed, yellow = deadball, red = goal attempt, black = point attempt 1st half, white = point attempt 2nd half

Advertisements

2018 All Ireland preview

August 28, 2018

Possessions

It goes without saying that the first primary objective of any team is to gain control of the ball. In the three most recent meetings between the protagonists (2017 AI semi-final, 2018 league and 2018 Super8) the total possession count is 136–135 in Dublin’s favour. It is somewhat surprising to see things so evenly matched however when we review first halves only Dublin are 69–62 ahead so game state has definitely had an effect on this metric.

Still Tyrone have been better in the last two encounters trailing the possession count at half time 24–23 in the league game and 22–20 in the Super8 contest. The concern here is that both of these contests were held in Omagh with all the built in advantages that entails. It will be imperative for Tyrone that they replicate these Omagh starts in the slightly less friendly environs of Croke Park and, at the very least, get on as much ball as Dublin. And get on it early.

Possession can be garnered in two primary ways. The first is through kickouts and the second is through turnovers (see note1 below). Kickouts are covered in depth below but focussing on turnovers there is definitely a trend favouring Tyrone. They, in their three games against Dublin, lead the turnover count 75 – 64 however that advantage dissipates when it comes to the prime counter attacking turnovers garnered outside your own 45. Both teams have managed to get their hands on 17 turnovers apiece outside the 45 with Dublin producing more (which will become a common theme) scoring 2–04 from these 17 possessions compared to 0–04 for Tyrone (see note2 below).

Tyrone’s turnover advantage is blunted by the fact that they have received the ball more often inside their own 45, giving Dublin time to reset the defence.

Once control of the ball is established the next step is to manufacture a shot. This is the oft referenced transition stage and is generally something that we in the GAA are bad at quantifying. We know, just from visual observation, that Dublin are excellent in this phase but just how good are they? And how do you quantify that?

Using shot totals as one measure Dublin are not that much better. In their three meetings Tyrone, despite having one less possession, actually end up with one more shot than Dublin (80 – 79). That even includes two shots that Dublin got from rebounds (so can’t really be attributed to “transition”). Now there is great disparity game to game (Dublin had six more shots in the 2017 semi-final, Tyrone had eight more in the league game whilst Dublin had just the one more in the Super 8 encounter) but Tyrone have, on the whole, created chances.

As further evidence of Tyrone’s ability to create chances; in their four most recent games (3x Super8 contests and the semi-final v Galway) Dublin progressed 68.9% of their possessions to a shot whilst the opposition combined for 69.3% on this metric. In their last four games, against much the same opposition as Dublin, Tyrone progressed 73.9% of their possessions to a shot. Of late Dublin have not really stopped teams, be it Tyrone or others, from transitioning to a shot.

If there is no discernible quantitative difference between how the teams use the ball then, given that Tyrone have lost the last three encounters by a combined 20 points, there must be quite a large qualitative difference.

When Dublin have the ball

Going into the 2017 final Dublin were recording a quite remarkable Conversion Rate of 61%. That was from five games which included three in the Leinster championship so theoretically that 61% should have been somewhat bloated by facing lesser opposition (that theory was dented when they then went out and hit 68% against Mayo in the final but still …. )

In their last four games, which on paper should have been collectively harder than the 2017 run in, Dublin have returned a quite incredible 64% (see Table1 in the Appendix for detailed player breakdowns). To try and put this into some form of perspective the average Conversion Rate from 2015 to 2017 was 53% whilst everyone else in the 2018 timeframe (Super8s and semi-finals) has hit 55%.

So how are Dublin achieving this?

Frees

Perhaps the most obvious place to start is with their free taking. Dean Rock has been the leading light in terms of accuracy of late. From 2015 to 2017 Dublin converted 82% of their frees (0–93 from 113) with Rock accounting for the vast majority of this; he took 81% of all their frees in this timespan recording a whopping 88% (0–81 from 92) Conversion Rate. Every other free taker was a combined 72% during this period.

Within the four games outlined above Dublin have converted 82% (0-18 from 22) of their shots from frees with everyone else from the Super8s onwards combining for 74% (0–95 from
128). Dublin would appear to be maintaining their free kick superiority.

It is not as clear cut as all that however. With the Super8s including long range experts like Michael Murphy, Rory Beggan and Niall Morgan it could be argued that Dublin should be hitting a higher Conversion Rate as they are collectively attempting easier frees.

Dublin have taken 55% of their 2018 frees from central locations (an imaginary line up from the D out to the 45 in the above chart) with only four being attempted from outside what I consider to be the optimal range. By contrast only 36% of the other teams’ frees were from this central alley with a not insubstantial 7% of all frees coming from outside the 45. So Dublin’s frees have been appreciably easier.

This is where the Expected Points model comes into its own. On the 22 frees Dublin attempted they were expected to score 18.42 points. They scored 0–18 thus were below average (Expt Pts of -0.42) for these attempts. The other teams had an Expt Pts of 91.26 on the 128 they attempted so the 0–95 scored was 3.74pts above average (Expt Pts of +3.74). Dublin have aided their 2018 Conversion Rate with easier frees but they have not been better free takers per se.

Tyrone however cannot rely on this. Dean Rock has stood up on the big stage before recording a whopping 83% (0-25 from 30; Expt Pts of +3.15) in All Ireland finals and semi-finals in 2016 and 2017. The balance of probabilities is he will deliver again in this final.

There is nothing Tyrone can do to affect Rock’s Conversion Rates (I have no doubt they will try all sorts of verbals, and running across his eye line, but if Lee Keegan’s flying GPS unit in injury time of a final can’t put him off I doubt anything Tyrone do will!). All they can do is cut down the number of opportunities they afford him from frees.
In their most recent non Dublin games Tyrone have given up an average of 6.75 shots at goal from frees (6.5 if we exclude those frees from outside the 45 which Dublin tend to ignore). In their last three games against Dublin they have allowed an average of 4.67 (4, 5 and 5) shots from frees. That is pretty good. This discipline needs to be maintained.

Point attempts

Something similar, in terms of Conversion Rates, is identifiable when it comes to point attempts. Dublin in 2015 – 2017 converted 54% of all their point attempts whilst everyone else was at 46%. In the last four games Dublin have scored on a remarkable 61% of their point attempts (see Table1). Everyone else in that time span has converted 50%. What gives?

Unlike frees – where the main influence on Conversion Rates is from where on the pitch the free was taken from – you can have more constituent parts here affecting the outcome. Location on the pitch is still a deciding factor; as is pressure (see note3) being applied to the defender. Weather can be a factor whilst game state (reviewed here) is also an issue.

Here we are going to focus on two of those – pitch location & pressure applied – but in truth all the above factors can feed in.

Pitch Location

Using our previously constructed central channel as a starting point Dublin again – much like their free taking – help themselves by taking more shots from here. 48% of their point attempts in the last four games come from this region as opposed to 38% for everyone else. But here the similarities end.

Dublin have converted 67% (0–29 from 43) of all point attempts from this central channel and 55% (0–26 from 47) from elsewhere. Everyone else, from the Super8s onwards, is 59% (0-113 from 192) centrally and 45% (0-135 from 303) from elsewhere. Dublin may take easier shots more often but they also convert them at a higher rate.

You may, correctly, say that all these percentages are within any margin of error given the sample sizes. But this phenomenon is not new. In 2017 Dublin were 63% on point attempts from the central channel and 54% elsewhere for an overall Conversion Rate of 58%. Everyone else was 57% centrally, 42% elsewhere and 47.5% overall.

The one trend that is very noticeable this year is just how much Dublin have focussed on shooting centrally. As noted above, and which is very noticeable from the shot chart, 48% of their point attempts have come from the central channel this year – that was consistently in the low 30s the last three years (32% in 2015, 34% in 2016 and 31% in 2017).

Pressure

Looking at the pressure applied to the point attempts we can see that strong or severe pressure was applied to 41% of Dublin’s point attempts and 49% of everyone else’s. It is hard to know if that gap is purely based off Dublin’s fluid movement and decision making or something else. One factor that could aid the gap is game state.

Dublin tend to be involved in more lopsided games than anyone else. Intensity, and as a consequence pressure on shots, drops off in such games. We are getting into very small volumes here but if we only look at point attempts in the first half, when that intensity drop should not be in evidence, do we get the same results? It does narrow the gap; Dublin face strong or severe pressure on 48% of their first half point attempts as opposed to 51% for everyone else.

Dublin’s high Conversion Rates are aided by centrally located shooting and a drop off in pressure in low key second halves. But that shouldn’t belie the skill in evidence – they still convert those centrally located shots at a higher rate than anyone else and manage to take more shots, under less pressure, despite shooting more often from the same crowded central zones.

What about Tyrone’s defence? Is there anything in their recent history that suggest they can affect this? In the seven games under review (see note4) Tyrone have allowed a 50% Conversion Rate on point attempts. About average. This overview masks a pretty big differential however.

Dublin have converted 56% of their point attempts in their three games against Tyrone whilst Monaghan, Donegal and Roscommon combined have converted 45% in their four games.
That 45% is very good. The 56%, when compared to 2017 and 2018 Dublin averages of 58% and 61%, is quite good. Tyrone will need more than quite good here however.

In fairness Tyrone have progressed game on game allowing Conversion Rates of 62% (0-13 from 21) in the 2017 AI SF, 57% (0-08 from 14) in the league encounter and 50% (0–10 from 20) in the Super8 contest. They can allow a high conversion rate off a low base – the league encounter, or a low Conversion Rate off a high base – the Super8 contest. Just not a high Conversion Rate off a high base which is what happened in last year’s semi-final.

Goals

Dublin are known for going for goals. The last four games, excluding penalties, has seen them return 8–02 from 17 shots at goal for a 47% Conversion Rate and 1.53 points per attempt (ppa). Everyone else from the Super8s onwards returned a 40% Conversion Rate (22-09 from 55) and 1.36 ppa.

In 2015 – 2017 Dublin converted 44% of their goal attempts for 1.43ppa whilst everyone else was 38% and 1.23ppa. Dublin are better at taking their goal chances and have been consistently so over the last few years. Quelle surprise!!

Dublin’s higher Conversion Rate is not the real story however. What is really devastating is that they maintain these above average returns whilst going for goal more often. In their last four games they have gone for goal once every 11.4 possessions whilst everyone else has gone for goal once every 18.6 possessions. From 2015 to 2017 they went for goal once every 14.9 possessions versus once every 19.0 possession for everyone else.

There is also the belief that Dublin go for goal early. This was definitely a “thing” in 2017 where the time of their first goal attempt in each game was 09:33 v Kildare (which they scored and then proceeded to goal again on their very next possession), 00:49 v Westmeath, 08:40 against Monaghan and then O’Callaghan’s two salvos against Tyrone @ 04:33 and Mayo @ 01:22 respectively. Five games all with a goal attempt within ten minutes and three with a goal attempt within five minutes.

This has not been as evident in 2018. It was 14 minutes before their first goal attempt against Donegal, their one and only attempt up in Omagh was in the 39th minute and 26 minutes had elapsed against Galway before they had a shot at goal. Only the non-entity of a game against Roscommon produced an early attempt (Costello had his saved on Dublin’s very first possession).

In the seven games in focus Tyrone have allowed 17 shots at goal, or 2.43 per game. In and of itself this is fine however again there is a Dublin/non-Dublin split. In the three Dublin fixtures they have allowed nine goal attempts. That’s three a game, or one every 15.4 possessions, to Dublin and two a game, or one every 23 possessions, to other opponents.

Again there has been an improvement. In their two 2018 meetings with Dublin they have allowed four shots at goal; two a game and one every 23 possessions. You feel that Tyrone will need to replicate this to have any hope but the nagging fear is that, as stated previously, both of those 2018 games were in Omagh. In their Croke Park meeting last year they allowed five shots at goal – can they replicate the Omagh form in Croke Park?

When Tyrone have the ball

Somehow, whilst everyone has been focussing on the blue marvels in the capital, Tyrone have amassed a very nice 60% Conversion Rate in their last four games (see Table2 in the Appendix). Whilst, like Dublin, they were greatly aided by Roscommon (Tyrone converted 79% for their 4–24 whilst Dublin hit 74% in their 4-23) the fear is that, unlike Dublin, the non-Roscommon return of 52% is a more accurate reflection of Tyrone’s level. To this point their 2015 to 2017 shooting saw them return a combined Conversion Rate of 52% with no individual year standing out as particularly accurate (50% in 2015, 49% in 2016 and 56% in 2017).

If Tyrone are not able to do any more than stem the Dublin tide can they show a hitherto fore unseen level of accuracy? In a one off game of course they can but their recent collective history is not so bullish.

Frees

One of Tyrone’s big downfalls in recent years has been their free taking. 2015 to 2017 combined saw them return 65% (2015 = 53%, 2016 = 55%, 2017 = 80%). This is against the backdrop of Rock hitting 88% in the same period. The 2017 mark of 80% looks like they may have hit on a solution however that elevated return was due entirely to Séan Cavanagh who is no longer there. He scored 0–12 from the 13 frees he took last year leaving everyone else to convert 67% (0–08 from 12).

2018 hasn’t seen much progression either. Combined Tyrone returned a below average 71% (Expt pts -0.68) in the last four games and have converted just 69% (0-09 from 13) in their three games against Dublin.

Tyrone know it is an issue (hence why they have had five different free takers in the three games against Dublin). Dublin know it is an issue. The crowd know it is an issue. As such the pressure will get ratcheted up with every miss (think of the pressure on McCaliskey in Ballybofey as he attempted one just before half time after Tyrone had missed three of their first four). At the very least Tyrone could really do with a clean first half here.

In the last three non-Tyrone games games Dublin have been very disciplined allowing the opposition an average of just 3.7 shots from a free with 1.67 in the first half of those games. They have allowed Tyrone just two frees in the first half of each of their three encounters.

Despite this being a weakness for Tyrone it might not manifest itself if Dublin maintain their normal discipline.

Point attempts

Although Tyrone have amassed a total Conversion Rate of 60% in the last four games that has been aided by a 90% Conversion Rate on goals (8 – 01 from 10 shots). This in turn then deflates the Conversion Rate for point attempts to 53% (0–48 from 90). Remove the Roscommon game and it drops to 46% (0 – 31 from 67). In the three games against Dublin it drops further to 41% (0-26 from 63).

There is just no sugar coating this. Tyrone’s point attempts from play have been poor. Very poor.
Part of the problem is shot location. If you recall Dublin are taking 48% of their shots from the central alley whilst the remainder are at 38%. Tyrone, in their three games against Dublin as outlined in the below chart, have only taken 24% (15 of 63) of their point attempts from this alley.

This isn’t a Dublin defensive thing either. In their three games, excluding Tyrone, Dublin have allowed 46% of point attempts from this central alley. Even removing the lacklustre Roscommon game it is still 43% for Donegal & Galway combined.

Just to prove it is more Tyrone than Dublin in their last four games, since the beginning of the Super8s, the volume of Tyrone shots from that central channel is 30% (27 of 90). An increase but still nowhere near the average let alone anywhere near Dublin’s returns.

Tyrone have been shooting themselves in the foot by ignoring the most productive sectors and shooting from what can only be described as sub optimal locations.

Goals

This may come as a surprise to those watching Tyrone of late but they have been relatively shy on the goal shot front. In the seven games under review they have just the 13 attempts with a sequence of 1, 2, 1, 4, 0, 4 & 1. That duck egg is the Super8 game against Dublin in Omagh whilst they have only managed three attempts in the other two Dublin games (a penalty in last year’s semi-final and two in the league game).

All that adds up to 1.86 attempts at goal per game over seven games or 1 every 23.5 possessions. Even more starkly it is 1 per game and 1 every 45 possessions versus Dublin! They will have to do something to create more goal chances.

There is a chink of light however in that Dublin have allowed teams to get in behind them of late with Donegal having three shots at goal, Galway four and Roscommon five. All Roscommon’s five came in the second half when even the pigeons had had enough of the game but still Donegal and Galway combining for seven attempts shows that it is no forlorn hope to step up on their current record

Kickouts

The below table shows the result of kickouts from the three Dublin v Tyrone games. There’s bit to unpick!

First of all, at a macro level, both teams have used a similar enough kickout profile with approx. 50% going short (inside the 45), 33% mid (between the 45 & 65) and 17% going long (past the 65).

On the short ones the returns are broadly (we’re dealing in broad brush strokes here!) in line. Tyrone have lost just the one – which was a throw up after an infringement – but managed to get a shot from 64% of the remainder.

We have seen that Dublin’s Conversion Rates are quite a bit ahead of Tyrone so it is no surprise to see that they have scored 0 – 03 more from the short ones they won. It is a surprise however to see that they converted the ones they won to a shot at a much lower rate than Tyrone (47% v 64%).

That equates to Dublin scoring 0.28pts per short kickout won. Against Monaghan, Roscommon & Donegal combined (see Table3 in the Appendix) Dublin have manufactured a shot from 53% of their short kickouts and scored 0.47pts per short kickout won.

Against Dublin the short ones, both defending against and attacking from, have not really been Tyrone’s problem.

The longer ones are a different issue. At a primary possession level there is no great problem. There have been 62 kickouts that have landed past the 45 with Dublin winning the contest 34–28. At the start of any game you would probably take a 55-45% split against Dublin on kickouts that land past the 45. One step further and Tyrone match Dublin when converting those key primary possessions to shots; Dublin convert them to a shot 59% of the time (20 shots from 34 possessions) with Tyrone at 57% (16 from 28).

It is the scores that come from these longer kickouts that become the problem. Dublin have scored 2 – 13 to Tyrone’s 1 – 07 from kickouts won past the 45. That’s 0.56 points per kickout versus 0.36. This is not unique to Tyrone; against the three aforementioned teams Dublin have scored 0.61 points per kickout won past the 45 and allowed 0.30.
Initially I had thought it was something that Dublin were cleverly doing around Marks but that is not the case. Tyrone have “out-Marked” Dublin 10–6 in their three encounters and scored 0–04 to Dublin’s 0–03 from those Marks. Instead the issue is within those kickouts that land gently into a Dublin player’s stride, past the 45, with no pressure applied.

To break these kickouts down
• Dublin have won 34 kickouts past the 45.
• They have claimed 6 Marks scoring 0–03. That leaves 28.
• Of these 12 were claimed off Tyrone kickouts with Dublin only managing to score 0-01; Dublin may be able to win the Tyrone kickout but haven’t scored off it.
• That leaves 16 of their own kickouts won past the 45 where the ball has hit the deck … but Dublin have scored 1 – 09!

We are in the realm of extremely small sample sizes here and these numbers are not something Tyrone can base a game plan around. But they are unique to Tyrone. Against the other three teams Dublin have won 44 kickouts past the 45 scoring 2-21; 2-06 scored off 16 Marks, 0–10 scored off the opposition’s kickouts (x16) and only 0-05 off their own (x12).

Wrap Up

Dublin are the offensive juggernaut that Tyrone will have to stifle. To do this they will need to ensure they manufacture as many possessions as Dublin, especially in the first half, and maintain their recent discipline in the tackle by offering up no more than five scoreable frees to Rock. They will also have to avoid the dagger of an early goal though luckily for them Dublin may be compliant in this by not going for goal as early as they did in 2017.

After that (sure the first bit is easy!!!) every shot needs to be affected through pressure or location. They need to stay in the game.

Offensively Tyrone have to pick up. They cannot hope to win this game with 70% on frees and mid 40% on point attempts. Although they have not created many goal chances against Dublin we know from the Donegal and Roscommon games that they can. And more importantly that they have been clinical in taking them. If they cannot stem the Dublin attack then they will have to score whatever goal chances come their way.

note1; 95% of all possessions emanated from kickouts (48%) or turnovers (46%) in 2015 – 2017 (9,018 possessions). The remainder came from shots regained (thus the start of a new possession) or throw ins.

note2; As an aside Dublin scored 1-15 from turnovers garnered inside their own 45 (0.38pts per turnover) with Tyrone scoring 0-17 (0.29pts per turnover). Again a function of Dublin’s ability to convert rather than anything Tyrone are (not) doing in transition.

note3; One man’s pressure is another’s loose arm – there is no agreed methodology for tracking pressure but I subjectively give a points range of 0 (no pressure) up to 3 (severe pressure) on each shot to indicate how much the defence pressure shots taken. It is subjective but over time any errors or biases should be applied to all teams evenly so if it is wrong it should be wrong in a very fair and even manner.

note4; The seven games are the three against Dublin, the two non-Dublin super8 fixtures (v Donegal & Roscommon) and their two encounters with Monaghan this year (in Omagh & Croke Park)

APPENDIX

Table1; Dublin shooting by player (Super 8 & AI SF)

Table2; Dublin shooting by player (Super 8 & AI SF)

Table3; Dublin kickouts v Donegal, Roscommon (Super8) & Galway (AI SF)

Dublin v Mayo AI Final 2017

September 19, 2017

Unfortunately I won’t be in a position to do a half decent write up until much later in the week so below are the usual battery of stats/tables. I will do up a write up at some stage and post it here – mainly for reference when these teams meet in 2018!!!

Listening and reading to some of the coverage post the final what jumps out to me is

O’Connor’s free at the end was outside his range – see the preview. It is so harsh on O’Connor that he takes a hit for stepping up to cover his team’s deficiencies and in some way lessens what he does elsewhere – see his brilliant long range point to draw the game level at the end.

Take a look at the assists table for players who had hugely impactful, under the radar, game

Mayo’s shooting was not poor! It was better than both 2016 AI Finals … it was just that Dublin’s shooting was exceptional. At one stage in the second half they were 0 – 11 from 12 for a Conversion Rate of 92%. Collective ice in their veins.

Overview

Kickouts

Dublin attack

Shot Chart

Shooting Table

Assists

Mayo attack

Shot Chart

Shooting Table

Assists

Dublin v Mayo 2017 All ireland preview

September 14, 2017

Immediate post-game reactions and numbers from the 2016 final can be found here (drawn game) and here (replay). Both will be referenced heavily below

Dublin Attack

We’ll get into the constituent parts below but as a collective Dublin have returned a whopping 61% Conversion Rate so far in 2017. The 188 shots they have attempted has netted them 19 points more (Expt Pts +19.05) than the average intercounty team would have scored. This Dublin attack is rolling.

Point attempts from play

Table 1 in the Appendix shows the breakdown per player but as a collective over two thirds (+13.87) of Dublin’s positive Expt Pts has come from their point attempts from play. The average Conversion Rate over the past 5 years has been 46%; thus far Dublin have returned 56%. The question is not how good they are – we know they are exceptional – but whether they can maintain that level of production?

And it is an open question as we have seen this before. Coming into both the 2015 and 2016 finals Dublin’s forwards were also flying with pre final Conversion Rates of 57% in 2015 (Expt Pts +17.63) and 51% in 2016 (Expt Pts +8.25). In the 2015 final they were impressive in maintaining their high returns with a 57% Conversion Rate (0 – 08 from 14) and a one game Expt Pts of +1.57 (in poor weather it must be remembered). However in the two games against Mayo last year they stumbled to a very poor 37% (0 – 14 from 38; Expt Pts -4.72).

The question is whether this stumble was induced by facing Mayo or was it more of a Dublin blip? There is no way to categorically state either way but looking from the outside I would give Mayo a great deal of credit. Coming into the 2016 final Mayo, in their six games, had allowed a Conversion Rate of 41% (0 – 45 from 109) as opposed to the 51% Dublin had been putting up. Across the two games the 37% Dublin achieved was much more in line with what Mayo had been allowing.

We have a similar scenario this year. Dublin are flying with the aforementioned Conversion Rate of 56% but again over a nine game span Mayo’s defence have only allowed a Conversion Rate of 41% (which, as an aside, shows remarkable consistency year to year). Can Mayo repeat their 2016 trick and “drag” Dublin’s Conversion Rate down in the final?

In an effort to explain the drop in Dublin’s shooting I created a pressure index (see Note1) and noted that Mayo were consistent across the two games in applying strong or severe pressure to just under half of Dublin’s point attempts (45% in the drawn game and 44% in the replay). In 2017 – across the nine games – Mayo have applied strong or severe pressure to 46% of their opponents point attempts. This has risen to 59% in the last four games.

So we know that Dublin can shoot and that they can carry it over (2015 final) however we also know that Mayo stymied them in both games last year and that the pressure index for their 2017 season is trending upwards.

One difference, for Dublin, observed when comparing this year to the last two All Irelands is just who is taking the shots. Below are the top 5 point takers coming into the last three finals and whilst there was a change from ’15 to ’16 the new entrants then were generally known quantities in McManamon & Rock. With Connolly suspended and Brogan & McManamon mainly coming off the bench there is a newer, fresher, look to the Dublin strike force.

Mannion & Andrews are well known to the Mayo defenders at this stage but O’Callaghan – despite there being a mountain of video evidence on him – is new and may cause the Mayo defenders to sit off initially; remember it was the relatively unknown Costelloe off the bench that did the damage in last year’s replay.

Deadballs

Given their experience, both last year and in their run to the final this year, Mayo may believe that they can handle the Dublin forwards. But what they will have to be very conscious of is fouling. Rock had a poor day in last year’s drawn game (0 – 03 from 7 (43%); Expt Pts of -1.58) but was then devastating in the replay (0 – 07 from 7 with an Expt Pts of +1.58).

He, and Dublin as a whole, have maintained this form throughout the 2017 Championship returning a barely credible 94% Conversion Rate (0 – 33 from 35) with an Expt Pts of +6.43 (see Chart 1 in the Appendix).
A special note of praise for Rock here. Although he had an off day in the 2016 drawn final he has been phenomenal since basically being removed from free taking duties at the back end of 2015. In 11 Championship games since the start of 2016 he is 92% and as the below graphic shows he is not only consistent “inside” where he is 96% but also outside (75%). Not only has he accuracy but he has accuracy with length.

Mayo cannot hope that the Rock from the drawn 2016 final appears. Instead they must expect that the Rock we have seen from the start of 2016, and in the 2016 replay, shows up. They gave up seven scoreable frees in the replay; in 2017 they have given up an average of 6.6 shots at goal from frees. Unless defending a lead their target has to be to minimise Rock’s opportunities from frees to seven or less.

Goal attempts

Dublin have – surprisingly given their opposition and reputation – only been average on goal attempts. In 2015 they were averaging 6.0 shots at goal per game but had a phenomenal 60% Conversion Rate (18 – 04 from 30). In 2016 this reduced dramatically to 2.2 attempts & a 46% (5 – 00 from 11) Conversion Rate. So far this year it is back up to 5.0 attempts a game but with a basically average Conversion Rate of 36% (9 – 01 from 25). In the 2015 final they manufactured 4 goal attempts but couldn’t convert any whilst in the two 2016 games they only created four distinct attempts at goal – but did manage to score in the most unconventional manner!

It’s all a bit of a rollercoaster. I think it is fair to say that Dublin are a lot less goal hungry than previous incarnations but this year has seen them somewhat pick up their attempts per game.

Mayo attack

Whereas Dublin are on fire Mayo have been hotter and colder, both from game to game and within games, than any other team. They returned a Conversion Rate of 41% against Galway and 71% against Cork; they scored 1 – 09 from just 12 shots in extra time against Derry after producing 0 – 01 from 15 attempts in a 25 minute spell in the second half. Against Cork they scored 0 -14 from their first 15 point attempts from play.

Mayo are the walking embodiment of the fact that you can twist stats to back up any argument

Point attempts from play

Looking at the nine games in the round (see Table 2 in the appendix) however we can say that their point taking has been just above average (50% Conversion Rate; +2.21 Expt Pts). But this is a step up on their 2016 campaign (46% Conversion Rate; +1.62).

There’s a slight anomaly in the above figures in that Mayo’s Conversion Rate has increased but the Expt Pts has remained more or less static. That’s because Mayo have attempted much more shots from central locations this year compared to last year. In a most Un-Mayo like fashion they have made life easier on themselves!

If we run an imaginary line vertically from the outer edges of the D to the 45m line we create a central channel. In 2016 34% of Mayo’s point attempts came from inside this channel. In 2017 this has risen to 51%.

A lot has been made of Mayo’s reliance on Moran and C O’Connor. Between them they have attempted 32% of Mayo’s point attempts returning 34% of their scores however in truth this is not too far removed from Dublin’s spread. Dublin have relied on O’Callaghan & Mannion for 29% of their point attempts and 30% of their scores. Now the argument can be made that Dublin have more alternatives in Connolly, Brogan, Kilkenny, Rock, Andrews et al should you shut the front two down – and that’s fair – but there are only so many shots to go around. And as good a unit as Kerry’s defence couldn’t shut them down. Moran & C O’Connor took 28% of Mayo’s point attempts across the two semi-finals with a combined Conversion Rate of 72% and Expt Pts of +2.89

Another notable point re Mayo’s shooting is that the “back-up”, the next ten players by shot volume (see Table2 in the Appendix), have been as accurate as Moran & C O’Connor with a 53% Conversion Rate (0 – 55 from 104) compared to Moran/C O’Connor’s 54% (0-34 from 63). Have Mayo got two top shooters? Yes. Have they ready-made, volume based replacements if they are shut down? No. Can the workload be spread and the efficiency maintained if they are shut down? Yes.

What of the Dublin defence? Coming into the 2016 final Mayo were running at an about average Conversion Rate of 45% (Expt Pts +0.89). For the two finals their combined numbers were 0 – 14 from 31 (Conversion Rate of 45%; Expt Pts of +0.22). The sample size is small but Dublin’s defence had no real additional effect on Mayo’s efficiency – Mayo carried over their conversion rate. Where they did have an effect however was in limiting the attempts Mayo had. Coming into the 2016 final Mayo had taken 20.7 point attempts per game. In the final this dropped to 15.5.

This year so far Mayo have averaged (again with remarkable consistency) 20.5 point attempts per game. In 2016 Dublin – prior to the final – had allowed the opposition to take 16.6 point attempts per game but this has risen to 18.6 in 2017. So not only has Mayo’s efficiency on the same volume of shots increased but Dublin are allowing more shots and (in 2016 at least) didn’t have an effect on Mayo’s efficiency.

Deadballs

C O’Connor has been rock steady on frees throughout his career converting, when the pressure has been at its most intense, 86% (0-55 from 64 Expt Pts of +4.99) in semi-final and finals alone from 2012 to 2016. His range does appear to have shortened however.

The above chart shows C O’Connor’s 2017 frees in yellow and using the same boundaries as the Rock chart above we can see that O’Connor is as good as ever “inside”. He has recorded very similar numbers to Rock at 95% (0 – 35 from 37) however he has definitely struggled “outside” recording a very poor (relatively speaking) 35%.
The two black dots on the chart are Jason Doherty’s attempts in the semi-final replay versus Kerry whilst the red cross is the free that O’Connor missed at the death in last year’s replay. As discussed at the time you would not have wanted anyone else standing over that free given his performance in the most pressurised of stages previously but it was definitely on the outer edge of his range.

Given his range limitations, his record this year and how comfortable Doherty looked the last day surely there’s a case for handing over the longer range attempts?

Apart from C O’Connor’s range – which does look to have a ready-made alternative should Mayo wish to employ Doherty – another area of concern would be frees from the right. Attempts from this area are sparse in the above chart (C O’Connor handed the ball to Moran for at least one free from this area against Kerry) as you would imagine C O’Connor is only too aware of his limitations. The issue was only exasperated by McLoughlin missing his two frees from inside the 20m on the right against Derry and Sligo.

Everyone knows O’Connor doesn’t want to take frees from wide right whilst the alternative, in McLoughlin, has, being kind, been shaky. Mayo know this is a weakness. We know it. You can be damn well sure that the Dublin defence knows it too.

Goal attempts

Mayo have been slightly below Dublin in their attempts per game at 4.2 however have been ahead of Dublin with a Conversion Rate of 42% (16 – 03 from 38 attempts). In the round that’s 5.6pts per game from both for their goal attempts (Dublin 5.60, Mayo 5.67).

Kickouts

One of the most eagerly anticipated duals will be the kickouts. How successful will Mayo be at getting their, what at times look extremely dicey, short kickouts away? Will they push up on Dublin? Once won how good will each team be at manoeuvring the subsequent possession into a shot and score?

Dublin Kickouts

See appendix for raw numeric tables

To date Dublin have gone short (see Note 2) on 66% (63 from 96) of their kickouts winning them all and managing to produce a shot from 73% which resulted in 3-27. That’s 0.57 points for every kickout won. Which is incredible. The returns for all other kickouts won is 0.45 – which doesn’t account for the kickouts the lost when the ball went past the 45 – whilst Mayo are a net (unlike Dublin they have lost 9 of their own short kickouts which has resulted in the opposition getting 0 – 03) 0.33 points per short kickout.

You have to imagine that Mayo will look to disrupt this wherever they can. To date 50% of the opposition’s kickouts have gone short in Mayo games with the opposition getting 0.31 points per kickout won. Mayo have gotten their hands on 10% (12 out of 120) of the opposition’s kickouts and when they do they go for goal – scoring 2 – 03 off those 12.
In the two finals last year Dublin had a similar split to this year in that 65% of their kickouts went short but Mayo had much more success keeping Dublin to just 0.12 points per short kickout won (0 – 03 off 25). And that’s without referencing the two short ones that Mayo won.

When we compare this to what Dublin have done to date in the 2017 Championship you have to believe that whilst Mayo may not be able to stop Dublin completely, and may not be able to clamp down as rigidly as they did last year, they should at least provide a much more substantial obstacle than Dublin have faced hitherto fore.

Mayo kickouts

Across both 2016 games Mayo lost five (17%) of their 29 short kickouts against Dublin including two at the death of the drawn game that surely had a part to play in Clarke being dropped for the replay. Clarke is now firmly ensconced as Mayo’s No.1 but the short kickout roulette can still occur at any moment as they have lost 9 across six different games.

Mayo have been somewhat fortunate in that none of those 14 lost short kickouts (5 against Dublin and 9 this year) have resulted in a goal – instead 0 – 08 has come from them – but it is only a matter of time. But Mayo will continue to roll that dice, even in the face of the Dublin pressure, as it has served them well overall. Whilst not as spectacular as Dublin’s 0.57 Mayo have returned 0.39 points for every short kickout won this year.

That does rise to 0.43 for their own kickouts that they win past the 45 however they only win 63% of these – for all the anxiety they give their fans on the short ones they are still winning 90%. It’s almost a no brainer. No matter how many palpitations they give you if your returns are the same on the short ones as on the longer ones but you win more of the short ones – well you keep going short. That is until you give away the inevitable calamitous goal … let’s hope for Mayo’s sake their luck holds for one more game.


Appendix

Table 1 – Dublin point attempts by player

Chart 1 – Dublin 2017 deadballs

Table 2 – Mayo point attempts by player

Table 3 – Dublin 2017 kickout overview

Table 4 – Mayo 2017 kickout overview

Note 1 – Pressure Index explanation; this is a subjective metric where the pressure applied to a shot from play is given a range from 0 to 3. Very generally
• 0 = no pressure applied,
• 1 = very little (e.g. a player running alongside but not tackling)
• 2 = strong (e.g. on the shoulder, catching the shooters eye by flying in to tackle)
• 3 = intense (e.g. a block, delivering a shoulder just as the shot is being taken)

It does come with a warning as it is subjective but seeing as there is only one person applying the metric you would expect that there would be a level of consistency when a large enough volume of shots is reviewed.

Note 2 – Kickout definition
• Short = ball landing inside the 45
• Mid = ball landing between the 45 & 65
• Long – ball landing past the 65

Again some judgement is required for contested balls around the 45/65 but on a large enough dataset any kinks will be lost to the averages

Dublin v Tyrone 2017 AI SF

September 5, 2017

Unfortuantely I haven’t been able to get to the game and at this stage there really is nothing I can add to the commentary that is already out there.

Instead I will just leave the raw outputs here in case anyone wants them

Game overview

Dublin attack

Shot Chart

Assists & shooting overview

Tyrone attack

Shot chart

Assists & shooting overview

Kickout overview

Mayo v Kerry 2017 AI SF Replay

August 28, 2017

Styles make fights. It’s an old adage but one that has endured because there is such an element of truth in it. How else to explain away how a game between two teams differs so much from one week to the next. In the drawn tie there were a total of 103 possessions with 47 turnovers. Kerry won the possession battle 54 – 49. Here there was a total of 87 possessions – with Mayo winning that battle 47 to 40 – and 31 turnovers. There was just 9 (9!!) turnovers in the first 35 minutes.

There are many minor elements throughout a game that will lead to such variances but the main variable that changes was in how Kerry set up. They started with an extra man back to cut off the kicked ball into Mayo’s forwards. The knock effect of this was that (a) Mayo’s kickouts were less frantic and (b) Mayo countered by playing more of a possession game. Both elements ensuring there were less turnovers.

More on the kickouts later but to my mind Mayo showed their collective football intelligence by changing their style to suit the additional element of Kerry playing one back. They played a game we don’t normally associate with them – keep ball. In the drawn game they had just the four (8% of all possessions) team possessions where there were at least 12 player possessions. Only one occurred in the first half of that game. In the replay that rose to eight (17%) with five in the first half.

Now 12 is an arbitrary number used to illustrate a point but that point is further bolstered by the fact that in the seven games prior to the semi-final that percentage (of team possessions with >12 player touches) was just over 8%. Mayo changed their natural game to play what was in front of them. And executed it with intelligence and no small degree of precision.

Mayo attack

A very clinical outing from Mayo scoring 0 -04 more than would have been expected from the shots they attempted. From an Expt Pts perspective the majority of the gains came from their goal attempts where three shots produced 2 – 00. Over the two games Mayo had eight shots on goal scoring 4 – 01 with an Expt Pts of +3.22 (Note; Expt Pts is less reliable on goal attempts than point attempts – as will be seen when we come onto Kerry). Even if they don’t manage to maintain the same high conversion rate Mayo will want to carry the volume of attempts through to the final as this was a weakness against Dublin last year. Over the 140 minutes of the replay and drawn game against Dublin they only manufactured three shots at goal returning 1- 01.

Another area that Mayo will be happier with now is their deadball striking. In the first game C O’Connor hit just
25% off a very low volume (0 – 01 from four; free x3 & 1 x 45) whilst his returns for the year badly lagged previous seasons (67% Success Rate & Expt Pts of -3.87).

Here he was 0 – 06 from seven attempts (86%) though with an Expt Pts return of +0.19 he would have been expected to score 0 – 06. His one missed attempt was again outside his range in being a 45 to the left of centre.

Jason Doherty’s two points (0 – 02 from 2; Expt Pts +0.66) were excellent – especially the first free from the left as Kerry had reduced the gap to four points at that stage by scoring the previous two and would have gotten a huge lift if Mayo had missed an opportunity through their main free taker being off the field.

Given O’Connor’s long distance travails the consummate ease with which he converted the second attempt, a 45, might give Mayo management a decision to make on the longer ones come the final.

In the drawn game Mayo’s point attempts were very good returning 52% (0 – 12 from 23; Expt Pts +1.03). They stepped back ever so slightly here returning 50% (0 – 08 from 16; Expt Pts of +0.78) but not to any degree that would be alarming.

Although their shooting form has waxed and waned throughout the campaign, and indeed throughout certain games, they have been very consistent at a macro level. In the seven games coming into the semi-final they were converting point attempts at a 49% clip (Combined Expt Pts of -0.40). In the two SFs here they were a combined 51% (Ext Pts of +1.83). In the final they will be looking to hold onto this 50% as a minimum – over the 140 minutes against Dublin last year they were 45% (0 – 14 from 31; Expt Pts -0.22). Marginal gains and all but … that uplift from 45% to 51% could be the difference in a one point defeat over 140 minutes and a one point victory in 70 minutes.

Kerry’s attack

Kerry’s numbers are an absolute contradiction with a very high conversion rate at 59% but a horrible Expt Pts of -5.29. The negative Expt Pts can be entirely explained through their goal attempts. They had four separate instances where they were through on goal, producing six shots, but came away empty handed. Of the six only Peter Crowley’s attempt – immediately after Mayo’s first goal – could be considered anyway clear cut. Geaney’s movement, in such a confined space, was nothing short of remarkable for the attempt that Boyle stopped on the line but it was far from what could be described as a “gilt edged” opportunity. Similarly the three attempts around the 45th minute that Clarke, and a plethora of defenders, kept out were, in the main, instinctive snap shots. Geaney had another attempt from out wide at the death but there again there was a whole host of Mayo bodies in the way.

Looking ahead to the final this will (should!) be an area of concern for Mayo. Although the quality of attempts here was low the six shots do add to the four Kerry had in the first game. We know how devastating Dublin can be when they scythe through teams – Mayo cannot afford to offer up four/five/six shots on goal to Dublin as they did to Kerry. (On the credit side of this argument is that Mayo only faced 11 goal attempts in their seven matches pre the semi-final and four in the two games against Dublin last year)

Leaving the goal attempts to one side Kerry were an incredible 74% (0 – 17 from 23; Expt Pts +1.97) on the remainder of the shots. We have seen that Mayo – on a combined 51% – had a combined Expt Pt close to Kerry’s with a 74% Success Rate. How so? Kerry achieved their 74% on very simple shots. The complete shot chart is in the appendix but below are all their 2nd half attempts – it comprises of goal attempts, tap overs and simple frees. There is only one shot from outside ~25metres. They went for goal too soon and too often.

Kerry struggled collectively the last day on their point taking (40%; 0 – 08 from 20) but coming into the semi-final they were on fire with a combined 71%. Yes the defences they faced were of a lesser variety than Mayo’s but they really should have been in a position to back themselves and keep the scoreboard ticking over.

Kickouts

Mayo (when compared with the outputs from the drawn game) wiped the floor with Kerry. They got 15 of their 23 kickouts away short with, unlike in the drawn game, quite a number coming under little or no pressure. That first game saw them score 1 – 06 from the 13 short ones that they won. They were less productive here scoring “just” the 0 – 04 on their own short kickouts but where they really dominated was on the mid/long range.

In total there were 19 kickouts that went past the 45 – with Mayo winning 13 (six on their own kickout to just one for Kerry and seven on Kerry’s kickout to just the five for Kerry). And when they won them they went for the jugular getting nine shots off and scoring 1 – 05. That’s 1 – 09 from kickouts both days. Why the turnaround? It is very hard to say watching TV footage but there appears to have been a convergence of Kerry getting caught between roles following their decision to play with an extra back plus Mayo going back to brass tacks (if you get a chance to re-watch Peter Canavan’s piece on Sky on how simple some of the “bunched” routines were it is well worth it).

One area that might give Mayo something to ponder is the fact that Kerry moved the ball quite well from their own short kickouts. Kerry won 11 converting eight to shots and scoring 0 – 06. Or 0.55 points per kickout won. In fairness this appears to be a blip as Mayo were giving up 0.28 points per short kickout in the prior eight games this year but still there are always those edges to be gained ….

APPENDIX

Game overview

Kerry’s shot chart

Mayo’s shot chart

x = missed, disc = score, yellow = deadball, black = 1st half from play, white = 2nd half, red = goal attempt

Mayo v Kerry 2017 AI SF

August 22, 2017

Kerry’s attack

Coming into the game one of the biggest disparities was in Kerry’s Conversion Rate versus what Mayo had allowed up to that point. In their three 2017 Championship games Kerry were running at a massively impressive 67% clip (both on total shots attempted as well as on point attempts from play) whilst Mayo were giving up 49% (a meagre 39% from play). Something had to give.

Kerry ended up with 2-14, a Conversion Rate of 50% and an Expt Pts tally of +0.37; not often you would consider letting in 2-14 a good result but Kerry’s returns were essentially average. On top of the overall lower conversion rate Kerry only hit 40% (0 – 08 from 20) on their point attempts. Advantage Mayo’s defence. Mayo did however give up eight shots from frees whilst Kerry created three clear cut goal opportunities (three shots on goal and Buckley’s converted follow up). So maybe more of an honourable draw with Mayo shading it.

Again coming into the game P Geaney & J O’Donoghue were the Kerry forwards’ standard bearers taking 40% of Kerry’s point attempts with a combined 64% Conversion Rate (Expt Pts of +4.44). Mayo managed to keep a lid on them as together Geaney & O’Donoghue “only” accounted for 30% of Kerry’s point attempts though their accuracy was as good as ever (0 – 04 from 6; 67% Conversion Rate & Expt pts of +1.44). That “keeping a lid” does come with a qualifier however; 0 – 03 came directly from frees where either were fouled.

There is plenty of commentary on the merit or otherwise of placing A O’Shea on Donaghy. Looking purely at the figures coming into the game Donaghy had taken just three shots in 124 minutes of action – scoring 1-2 mind – whilst also having eight primary and three secondary assists. Here in just over 70 minutes he scored just 0 – 01 but managed three shots and as can be seen from the assists table below he was Kerry’s puppet master with six primary and two secondary assists.

Kerry assist table

Undoubtedly the three shots were from less productive areas as Donaghy roamed given that Kerry had switched away from the high ball – but that is a minimal gain. Donaghy had a hand in all three goal opportunities and was the dominant Kerry presence in the forwards.

Mayo’s attack

Coming into the game Mayo had converted 53% of their shots (49% of point attempts) whilst Kerry had allowed 47% (40% on point attempts). Here Mayo had a combined 50% with 52% from point attempts. Advantage Mayo

Moran & O’Connor’s accuracy up front was excellent scoring 0 – 07 from 8 point attempts (88% Conversion Rate with an Expt Pts of +3.21) whilst overall, when including goal chances, they produced a combined 75% Conversion Rate (1 – 08 from 12) and Expt Pts of +2.37. You will go a long way to find better returns from a front two in such a big game.

Of course (there’s always a caveat!) if O’Connor & Moran were sublime, but Mayo were more or less bang on average, that means that there were issues elsewhere. The most noticeable was their deadball accuracy. They had three frees and one 45 with O’Connor only converting 1.

This is – worryingly for Mayo – now becoming a trend. Attached are all O’Connor’s deadball attempts this year with the ones in the Kerry game highlighted in black. Previously we highlighted that O’Connor has a definite arc outside of which he is vulnerable. That arc is as evident in 2017 as it ever was.

O’Connor 2017 deadball chart

Outside of Moran & O’Connor Mayo were 33% (0- 05 from 15; Expt Pts of -2.18) which is well down on how they were doing coming into the game (49% & an Expt Pts +0.97). It is hard to see Moran & O’Connor being as productive the next day but even if they do “slack” somewhat the expectation is that the rest of the team will improve.

Areas of improvement for the replay

As already intimated Mayo will need more from the supporting cast from play whilst also getting more than just 0 – 01 from whatever deadballs they have. But there’s also the issue of their discipline. They gave up the eight shots from frees but some of those were, to use a technical term, just dumb – especially in the first half. After going ahead by 1 -01 they gave up an off the ball free just outside the D and then managed to give up an extra 13m for a free out towards the wing that brought the ball right onto the D. Two simple tap overs, lead halved and Kerry were up and running. This doesn’t account for the one where Vaughan dived at O’Donoghue’s feet from behind.

As an aside this is not the first time that 13m was tagged on to an important free – the equalising score for Cork came from a free just outside the 45 that had had 13m tagged on. Gotta cut out the dumb sh*t in the replay.

Kerry? They will be disappointed in their shooting. Dropping from 67% on point attempts in the three previous games to 40% here can be somewhat accounted for by the conditions and the step up in defensive quality however quite a bit of the drop can be credited to Kerry themselves.

Kerry pt attempts from play – pressure/no pressure

Of their 20 point attempts 45% were taken under little or no pressure (white in the above chart) with just a 33% Conversion Rate. The three converted were by P Geaney, BJ Keane and K Young (with his right!!) but it was the ones that were missed (Moran & Donaghy x2, Maher & Morley x1) that Kerry will need to tighten up on.

Kickouts

The main focus during the game was on Mayo’s kickouts where Kerry applied huge amounts of pressure pushing up on the short ones in an effort to force Mayo to go longer. When Clarke did have to go longer Kerry lorded it winning 8 of 11 (73%) that passed the 45. But the reason for Kerry pushing up so hard was evident in what Mayo did with the short ones they won. Of the 13 they got away they managed to get ten attacks, eight shots and scored 1- 06. Unbelievable returns.

Playing the short one as much as Mayo do – 109 so far in their eight games – they’ll always run the gauntlet of the calamitous one. Kerry got a point from the one short kickout they intercepted; to date Mayo have lost 9 (8% of all short kickouts taken) and given up 0 – 07. A goal will eventually come but the credit balance is such that Mayo really should just accept when it happens (though doing everything in their power to prevent it) and stick to the plan.

One point to note is how Mayo fared over the two halves. In the first half Mayo only got 46% (6 of 13) of kickouts away short but in the second half that rose to 73% (8 of 11). The high press game is very tiring and a combination of Kerry fatigue, and Mayo alertness to what they were doing, ensured a much more productive second half

Less newsworthy in the post-game reaction was how good Mayo were on Kerry’s kickouts. When Kerry went past the 45 Mayo essentially broke even (Kerry won 7 getting 5 shots; Mayo won 6 getting 4 shots from same) but they were more impressive in what they did to Kerry’s short ones. Kerry had 11 but only converted three to a shot.

APPENDIX

Game overview

Kerry’s shot chart

Mayo’s shot chart

x = missed, disc = score, yellow = deadball, black = 1st half from play, white = 2nd half, red = goal attempt

Mayo assists

Mayo v Roscommon 2017 AI QF

July 31, 2017

For those new to the blog, or who haven’t been here for a while, please find a refresher on the definitions and how the numbers are compiled here

Team Possessions Attacks Shots Scores Exp Pts
Roscommon 56 43 24 2 – 09 13.23
Mayo 54 42 29 1 – 12 14.46

Both teams had similar possession volumes (56 for Roscommon to Mayo’s 54) as well as Attack Rates (77% v 78%). Once inside the 45 there was a divergence however with Mayo getting more shots off (29 to Roscommon’s 25).

As an aside 110 possessions is high but not inordinately so. Across 54 games in 2015 and 2016 I have eight with a higher possession total including the 2015 final between Kerry & Dublin which was played in similarly wet conditions. The Connacht game ths year between Mayo & Galway had 113.

Mayo attack

As ever we start with Cillian O’Connor. One of his less auspicious days with a total Conversion Rate of 27% (0 – 03 from 11; Expt Pts – 2.91) and just 14% from play (0 – 01 from 7; Expt Pts -1.94). In eight games in 2016 he returned a combined conversion rate of 62% (0 – 44 from 71) with an Expt Pts tally of -0.93.

As ever his appetite for work and willingness to offer himself up as an option were evident as he took four of Mayo’s last six shots (Durcan took the other two) from the 65th minute onwards. One of these shots was a free from well outside his range whilst the other three were central enough however two were taken under strong or intense pressure. Indeed five of his seven attempts from play were heavily pressurised. If you’re Mayo you probably want O’Connor on the ball at the death but the opposition are well aware of this.

In a down game for O’Connor Lee Keegan almost single-handedly dragged Mayo back from the abyss scoring 1 – 03 from just five shots (80% Conversion Rate; Expt Pts +3.25). The goal was a prime example of split second decision making that can win/lose/decide games.

In the above Keegan’s (5) marker Enda Smith (9) gets sucked into the kickout melee unaware that Doherty has claimed a mark. Keegan sees it and immediately sets off behind Smith’s back. McDermott (Roscommon defender pointing) sees the danger but it is too late. Keegan is gone with nothing but open road in front of him.

The remainder of the Mayo team were average scoring 0 – 06 from 13 (46% Conversion Rate; Expt Pts of +0.10). None of the starting trio of McLoughlin, A O’Shea or S O’Shea managed a shot whilst only Tom Parsons produced one off the bench. Mayo will definitely be looking for more of a threat here the next day.

On McLoughlin & A O’Shea the fact that they didn’t get a shot off does not necessarily mean they had a poor game. Both were very involved higher up the pitch as evidenced by the assist chart.

Roscommon’s defence may be slightly disappointed in the fact that 46% (11 of 24) of Mayo’s shots were taken under little or no pressure however on the flip side they will be delighted in where they forced Mayo to shoot from. Mayo never really got through them or around the side – the vast majority of shots were from “outside”.

Roscommon attack

At a macro level Roscommon’s shooting was very good with them scoring ~1.7 points more than the shots they attempted would normally return. But their Conversion Rate was essentially average at 46% and therein lies a problem. Their high returns from goal attempts (2 – 00 from 2) masks how poor their shooting in general was. They had 16 point attempts from play and returned just 0 – 05 (31% Conversion Rate; Expt Pts -2.42).

It is quite possible, in a one off game in the replay, that they will again create more goal chances than Mayo and convert them all. It is much more likely however that they will revert to the mean on their goal attempt conversions and thus they will have to rely more heavily on the point taking (be that from play of from frees).

One point to note here is the affect that Mayo’s defence had on Roscommon. Above we showed how Roscommon helped themselves by keeping Mayo, in the main, on the outer perimeter. Roscommon got inside Mayo more often (see shot chart below) but the Mayo defense deserves credit as they heavily pressurised 75% (12 of 16) of Roscommon’s point attempts. This included a ~35 minute period post the second goal where they heavily pressurised 8 of Roscommon’s next 10 point attempts leading to Roscommon returns of 10% Conversion Rate (0 – 01 from 10) with a combined Expt Pts of -3.31.

Now undoubtedly there was “scoreboard pressure” as Mayo overturned the 7 point deficit that led to poor decisions and execution in the shots but Mayo definitely aided those poor decisions.

75% heavy pressure is high and at times they walked a bit of a tightrope offering up six attempts from a free with four coming inside the 45 – but you’ll take a combined return of 41% ( 0 – 09 from 22) from all point attempts.

Kickouts

Very even all told. 20 kickouts went passed the 45 with both teams winning 10 each. As with the general trend mentioned above Mayo were better at converting those possessions to shots.

In an interview post the Cork game Rochford mentioned, in assessing A O’Shea’s impact, how he had aided smaller things such as winning the throw in. Here Roscommon not only won both throw ins but scored off both as well.

APPENDIX
Roscommon’s shot chart

Mayo’s shot chart

x = missed, disc = score, yellow = deadball, black = 1st half from play, white = 2nd half, red = goal attempt