The weather forecast

It’s often said that hurling is a game for the summer; despite the majority of intercounty matches being played in the league, during winter and late spring, many don’t feel this is truly representative of the sport as it should be played. For the majority of fans, ‘real’ hurling doesn’t start until the summer, with harder, faster pitches and gentler breezes. While a large part of this view is simply due to the greater importance of the championship, in this article we’re going to investigate what impact the weather has on hurling scores.

Method

A database of matches was compiled, featuring all the games over the previous 20 years (1998-2017, inclusive) between the current Leinster and Munster teams (Clare, Cork, Dublin, Galway, Kilkenny, Limerick, Offaly, Tipperary, Waterford and Wexford). These teams were selected as they have all consistently played in the upper tiers of the league and championship over the years investigated; occasional matches against far weaker teams would likely have led to scoring outliers, which could fog the impact of the weather alone. The vast majority of these matches occurred in the aforementioned counties, with a small handful also occurring and Laois or Carlow. Using the Met Éireann website, weather data was downloaded for each date on which a match occurred. The different weather factors looked at were rainfall for the day, maximum temperature for the day, minimum temperature for the day, average wind speed, and average atmospheric pressure. All of these factors were selected as they could be easily related to how a team performs: High rainfall would lead to a muddier, slower pitch, very low or high temperatures could lead to the players being sluggish, high winds could increase the number of wides, and high pressure could lead to less cloud cover, and the sun getting in players’ eyes.

Different weather stations were used depending on where the match took place; as not every weather station records the same information, some were far closer to the pitch than others. Some stations were used for multiple counties, as they were the closest major station to multiple home grounds (for example, Shannon Airport was used for Clare, Limerick and Tipperary). The list of weather stations used is listed in the table below:

Stations.png

Some of the weather stations were not opened until after the match database began, so certain grounds had fewer data points than others. However over 800 matches in total were investigated for this article, which should be plenty to determine correlation.

Making a weather forecast

Each weather factor was compared with the total number of goals and points scored in each game, and the Pearson correlation coefficient was calculated. This number represents how strongly correlated two variables are: a value of 1 means that they are very strongly correlated, with a fixed proportional increase in one following an increase in the other. A value of -1 indicates the same, but in a negative direction; a decrease in one leads to a fixed proportional increase in the other. A value of 0 indicates that there is absolutely no relationship between the two variables. In other words: The higher the absolute value, the more strongly correlated the variables are, and the more appropriate it is to draw a linear relationship between the two. The chart below shows the results for each weather factor when compared with the number of points, goals and the overall scoring margin:

The results provide us with a few clear implications. Firstly, goals and scoring margin are far less susceptible to the weather than points. Even the strongest correlation has an absolute value of less than 0.1, so even with the large sample size, it’s difficult to argue that these are really significant. This makes sense; shots at goal are typically taken at close range, where wind isn’t as much of a factor, and anything which would slow the shooter down would also slow the goalkeeper. Also, the small number of goals in a given match means any change in number produces a large proportional swing, making their cause harder to pin down. There’s also no obvious reason why the weather should affect the scoring margin, unless a strong wind dies down after half time.

Secondly, we can see that the air pressure makes very little difference. It seems that the sun getting in players’ eyes is not a major risk factor when it comes to making their shots. Perhaps this correlation would be stronger if the dataset focused more on the days before helmets were mandatory.

Overall, none of the weather factors have very strong correlations. However, other than air pressure, it seems reasonable to believe that all of these effects are still real, given the large sample size of matches. They may not be strongly linked, as there is so much more that contributes to the final score than just the weather, but there is an effect.

The greatest effect is the maximum temperature; this correlation is, perhaps, much stronger than the minimum temperature, as the vast majority of matches occur in the early afternoon, when the temperature is close to the maximum. This makes the value more relevant to the actual conditions. The temperature is the only factor studied with a positive relationship to the scoring; the warmer it is, the more scores occur. Assuming a linear relationship between the points scored and the maximum temperature, each team would be expected to score an extra couple of points on a warm summer’s day in the mid-twenties than in a league match at a cool ten degrees. This result does tie in with the traditional wisdom that the game speeds up in the summertime.

The second biggest influence is the wind speed. It’s perhaps surprising that this doesn’t have a greater impact, as it has a very direct impact on a shot’s trajectory. However, there are limitations. We don’t know the wind direction, as a strong breeze towards either goal, rather than towards the sideline, could cause one team to go on a scoring run. We also only have the average wind speed over the course of the day; the air could be still during a match, with an overnight gale, which would disrupt the data. However, accounting for all of this, the wind speed still seems to have a real impact on the final score, with faster speeds resulting in lower scores. On a still day, the average team from the dataset might be expected to put up 18 points in a match. Even a moderate breeze (using the technical definition, according to the Beaufort scale) will reduce this by 2 or 3 points.

Rainfall is less impactful. Though it can slow down the pitch, ground hurling isn’t what it used to be, and so teams may just take more shots from range in these situations. However, there is still a slight negative relationship; the rainier the day, the lower the score.

Summer versus winter hurling

The average weather conditions for each day of the year over the last number of decades was determined, as recorded in Shannon, Cork and Dublin airports. Using the two variables with the strongest correlations, max temperature and wind speed, the expected total points scored was determined for every day of the year, assuming that these variables had a linear relationship with points scored. This was then compared with the actual number of points scored in matches on each date.

The above chart does highlight how weak the relationship is; there’s simply too much complexity in hurling for these two factors to have such a large impact. However, looking at the overall trendline for each group, the results do seem to line up. In each case, a curve is formed, where less points are expected in the winter, and more are expected in summer. Though there is a lot of noise in the graph, the actual points scored do seem to line up quite well with the expected points from wind speed, on average.

In summary, the old adage holds true: Hurling speeds up in the summer, and if you want the excitement of a high scoring game, then your best bet is to watch a championship match. However, it’s not as strong a factor as you might think; the league is still perfectly capable of delivering a surprising and exciting match, albeit slightly less often than the championship. Still, given the lower cost of league tickets, and the smaller crowds to wade through, it’s certainly something to consider.

Just remember to wear a coat.