Cúl runnings

There are many sports which have different scoring options with different rewards. In rugby, a player may score 5 points for a try (with 2 more for a conversion), versus just 3 for a drop goal. In basketball, a player can opt to score 2 points with a closer shot, or try their luck with a 3-pointer from long range. Similarly, hurling has two options: 3 points for a goal, versus just 1 for getting it over the bar. Each sport has a simple trade-off of risk versus reward: a tougher shot will get you more points, and an easier one will get you fewer. By having multiple options, it can keep the sport more interesting and unpredictable; hurling without points, for example, would likely result in players clustered around the goalmouth, rather than keeping the game open. Additionally, the risk of conceding three points in one possession can keep tension in a match where one team would otherwise hold a firm lead.

A common goal for sports analytics is to put a specific value on when the riskier option becomes worth it. The impact of this type of investigation can sometimes be enormous; long-standing questions over which option to take in which scenario have been answered, and sports such as basketball have seen certain types of shot nearly disappear at the professional level, in favour of shots that have been shown, statistically, to just be better options.

Bringing it back to hurling, we’re going to investigate the value of the goal versus the point. Though the difference in risk between attempting a point versus a goal is much more nuanced than taking a basketball shot inside or outside an arbitrary line, it’s still possible that there’s still a lot we have to learn about when we should take our points, and when the goals should come. 

A history of goals

Goals in the modern game aren’t rare occurrences, but games where the majority of the score came from goals certainly are. Before we get into the analytics of how often a player should go for goal, let’s first look at what we can learn from over a century of trial and error. One very important note, which I hadn’t realised before I began this article: the value of a goal hasn’t always been constant.

Hurling in the early days of the GAA was quite different. In the very first championship, goals were mostly what mattered: points were only used as a tiebreaker for teams on the same number of goals. In other words, in a game where one team had 1-0 and the other 0-25, the first team would win. Instead of 65s, there were also ‘forfeit points’, with each one being worth 1/5th of a point.

In 1892, this was changed, and goals became worth 5 points (and the number of players on each team was reduced from 21 to 17). Still, a very different setup to today’s game, which would drastically affect how often teams attempt to score goals versus points.

Finally, in 1896, after almost a decade of All-Ireland championships, we had the modern 3 point goal.

Let’s take a look at how teams have changed over time in how they play the game. It’s obviously difficult to find results on every match from the history of the GAA, so instead we took a small sample from every year of the All-Ireland: we checked the results to every All-Ireland final, semi-final and provincial final. From these, we can see trends in how teams scored over time.

‘Take your points and the goals will come’ appears to be a more modern invention than we realised. The chart above shows how much goals and points contributed to an average game’s total score. From the beginning of the GAA roughly until the late 20s, it seems that teams went for more and more goals. Following the rule change in 1896, points did briefly overtake goals as the primary contributor to the score, but this soon flipped dramatically, with goals typically taking over the vast majority of scoring, and the goal/point ratio almost the inverse of what we’d see today. This trend slowly reversed until points finally took over again the 50s. Ever since then, the gap between how much goals and points contribute has been widening. It seems that the decades of trial of error have reached a consensus, which is that you’re better off going for points almost all the time, taking goals only when a good opportunity arises. The result of this is that total scores have also trended upwards, so it looks like this prioritisation of points is working. 

How much harder is a goal than a point?

At a glance, the formula for whether or not you should go for a goal is simple enough: If g > 3p, where ‘g’ is the likelihood that you’d score a goal in the current situation and ‘p’ is the likelihood that you’d score a point in the current situation. In other words: if, in a given situation, it’s more than three times harder to score a goal than a point, you should take your point. The real difficulty is determining how to determine these likelihoods. It’s possible that part of why the three-point shot has dominated basketball in recent years is because it’s a sport where these likelihoods are easy to calculate: If a player makes more than two-thirds as many shots when behind a line as when they’re in front of it, they ought to take more three pointers. You can’t attempt a two-point shot from beyond the three-point line, or vice-versa, making it very easy to track attempts at each shot.

Hurling is much trickier: There can be ambiguity about whether a player was attempting a goal or a point. Goals don’t just come down to your location on the pitch. There’s no limit to where a goal can be taken from; though most will require getting into very close range, we’ve all seen occasions where a long pass drops behind the keeper and rolls in, or where a shot at a point catches the wind and drops low at the last moment. Similarly, an attempt at goal instead becomes a point if it goes too high, either by the shooter or from a deflection. To make the above calculation more accurate, we’d really have to factor in the likelihood of a goal attempt going over the bar for a point, or a point attempt dropping low for a goal, rather than assuming every missed attempt will be wide or saved.

Even getting past all this, there’s the more practical issue that this data simply isn’t there for hurling. If we really want to pin this down, huge amounts of information would be required to tell what are the optimum situations to go for goal, showing things like shooter location, defender/goalkeeper location, wind direction, whether or not the shot was definitely meant to be a shot at goal or at a point, how many possessions you have that end in a shot versus your opponent’s, and many, many more factors. This website really doesn’t have the resources to look into all these things. However, there’s probably no harm in making a very crude attempt. Perhaps someday, when analytics has ruined the magic of every sport, we’ll look back and laugh and how people thought goals should work. Until then, we’ll work with what we have.

In a previous article, shot charts were made of five different league games. This is a terrible sample size, but we’ll press on and do what we can with it to try to determine where both goals and points should be taken from. Across these five games, a total of 366 shot attempts were counted. Of these, only 17 resulted in goals, while 182 resulted in points. This leaves 167 shots which were either saved, were deflected off the crossbar or posts or went wide.

Let’s try to break these down by location. While the shot chart doesn’t have precise player location, we can at least divide it by the pitch markings:

The first chart displays where shots were taken from. If you want, you can consider it an example of how easy it is to obtain a shot from this portion of the pitch. Following this, we can see where all the points and goals were scored from. This is followed by a chart showing points per shot in each location; how many points resulted from the average shot in each location, from a minimum of zero to a maximum of three. This can be considered how efficiently players tend to shoot in each location, or how easy it is to score from a shot in a given location. The next chart is the total score gained in each location; perhaps a more realistic metric of where to shoot from than efficiency, as it reflects how easy it is to get shots from different areas; obviously, it’d be nice to take every shot from right next to the goals, but that’s never going to happen. In order to try and determine what the best shooting locations are, we must combine volume and efficiency in some way. One such example is below, where we have subtracted the percentage of total shots taken in each location from the total score gained in each location:

A paler segment indicates that players were probably taking an appropriate number of shots from it. The more blue the segment, the more players should attempt to score from it. The more orange the segment, the less players should attempt to score from it. From this, we can get an idea of where players should be trying to get more or fewer shots. For example, the large rectangle is hard to get to, but so many of the shots that occur there result in goals. Therefore, it may not be a bad idea for players to drop the occasional shot in there, as the reward may outweigh the risk. Trying to get the ball into the small rectangle more often is less appealing, however, as it’s so difficult to get that close, without much additional reward compared to the large rectangle. In contrast, it’s easy to get shots from beyond the 65m line, but these are so difficult to score from that it may not be worthwhile; it’s very likely that fewer shots should be attempted from here. However, players have much greater success when between the 45m and 65m lines, without too much added difficulty in getting there, and so are probably taking a more appropriate number of shots from here.

Now, all of the above needs to be taken with a grain of salt, as not all shots are made equal. A player may be likely to attempt a shot from very close range, even with defenders right on them, whereas shots from very long distances only tend to happen with a good deal of space and time. The small sample size also impacts things, as we can see there was a goal scored from a long ball in the Wexford-Kilkenny match. Taken from between the 45 and 65m lines on the other side of the pitch, intended to be flicked in by a full forward, it instead dropped into the Kilkenny net. Due to the small number of shots taken from this segment of the pitch, the goal skews the diagrams tremendously. The takeaway from this shouldn’t be to let players take shots at goal as soon as they’ve successfully caught a short puck out from their own keeper, even though it looks like the charts suggest it. However, using some common sense to account for all of this, and assuming that players have some degree of sense about where and when they shoot, we can definitely pick out some trends to make some informed decisions about when to go for goal, and when to attempt a point.

Almost no goals are scored from outside the large rectangle. Though we lack a good sample size for goal attempts, it appears safe to say that if you get this close, it could be a good idea to go for goal. Once you get outside the 20m line, goals dry up almost entirely, though there are a handful scored from inside the arc, as this would have a reduced distance compared to other locations beyond the 20m line. So, in a very obvious conclusion: you probably shouldn’t attempt goals from far away, and the best location to try to get goals from, after factoring in the difficulty of getting a shot from there, is the large rectangle.

So, we’ve looked at the options for scoring goals, but what about points? Again, the findings seem to be mostly expected: 78% of all points happen between the 20m and 65m lines, despite only 63% of all shots occurring here. This covers a very large portion of the pitch, so saying this doesn’t really narrow things down too much, but: This is your ideal spot to take your points. Further than this point, and the odds of making your shot go down drastically, and closer than this point it begins to become more valuable to go for goal. However, there is one interesting, and perhaps unintuitive, finding: Between the 45m and 20m lines, there were 0.65 points gained from every shot taken in the five matches. Between the 45m and 65m lines, this only dropped to 0.62 points per shot; a surprisingly small difference. When you remove shots from inside the arc, which includes a couple of goals, the 20m to 45m range actually drops to 0.6 points per shot: slightly less than the longer range shots. In other words: up to a point, the range of the shot doesn’t matter too much. As players get closer to the goal, it’s expected that the defence will begin to tighten, or begin to push the shooter out to the edges of the pitch for a tougher shot. The further out, the more likely the player has some space to shoot. It could also be possible that this number is skewed by how many frees occur in this location, though the success rate may indicate that it’s still not a safe location in which to give away a free. It’s commonly said that a team’s half-back line is its backbone, and this lack of a drop-off with range helps to show one of the reasons why: if they fail to close down on their opposing forward, what should be a poor shot will instantly become a lot more valuable. You can’t rely on senior inter-county forwards to miss much more often, just because they’re standing a bit further away, and good defences will realise this.

Conclusions

Similar to the previous article on shot charts, we can see that there’s a lot of information that can be gained from looking into information about player position. However, we see again that, with only a limited amount of data, and limited context of what else is happening in the game, the insights to be gained are also limited; findings like ‘Go for goal when you’re near the goal, go for points when you have space within your opponent’s half’ aren’t exactly groundbreaking. However, we can see from the earlier section in this article how maybe this wasn’t always so obvious, as early teams tried to excessively force goals: a huge amount of what we know can be attributed to what we’ve learned through trial and error, not through intuition. With more work into analysis of the game, this process could be accelerated, and with more research, there could yet be some keys to the sport that are waiting to be discovered, and many accepted truths waiting to be debunked.

You never know: maybe that Wexford goal from the other half of the pitch wasn’t an outlier, and it’s the way to go after all.