It has been a long-held notion in rugby that weather impacts how well teams are able to perform as well as how they approach the game tactically. As of this article, there do not appear to be any publicly available quantitative studies on the impact of weather on team performance in rugby, although there do appear to be some general descriptive articles provided by various gambling web pages (Reference 1; Reference 2). The purpose of this article is to take a data-driven approach to evaluating how a specific aspect of weather – temperature - impacts team performance and tactics in English Premiership Rugby.
There are several perceptions as to how the weather will impact teams during a typical Premiership winter. The prevailing thought is that playing games in cold and wet environments (a.k.a. winter conditions) will result in slick conditions which adversely impact a team’s ability to play attractive “ball in hand” rugby compared to playing under warmer, dryer conditions. Based on this perception, we hypothesized the following:
Hypothesis 1: Teams will score fewer points in colder temperatures
Hypothesis 2: Teams will commit more turnovers in colder temperatures
Hypothesis 3: Teams will adjust their tactics by kicking more in colder temperatures
For this study, data from 10 seasons of Premiership rugby (2012-2021) comprising a total of 1274 games and 107,413 possessions were analyzed. Weather conditions for each game were obtained from a weather webpage by recording the temperature, humidity, weather category, and wind speed from the nearest weather station to the home stadium on the date of the game at the time closest to kickoff. One thing that became evident early in this study is that weather is complex and it is often a combination of factors which could potentially impact team performance. To control for potential confounding effects due to interactions between the various weather factors, we chose to isolate temperature as a variable to measure its effect on performance. In testing the 3 hypotheses, a simplifying assumption was made to analyze temperature setaris parabis. For those of you not fluent in latin, or without previous exposure to macroeconomics, this means that for any given temperature all other weather conditions were treated equally. Thus, for a game at a specific temperature (e.g. 0°C) all other coexisting weather conditions (e.g. wet/dry, windy/calm, etc.) were not differentiated and were aggregated into the same bin.
To test whether or not there is a relationship between temperature and the ability of teams to score points, a scatter plot of the effect of temperature on the total combined scores for each game was generated (FIGURE 1). A regression and correlational analysis of the graph revealed a significant positive relationship between temperature and total score (r(1272) = 0.15, p < 0.05). Looking at the slope of the graph (m = 0.3571) we can estimate that for roughly every 3°C change in temperature, the combined total score of the game will increase/decrease by 1 point.
FIGURE 1: The effect of temperature on total combined scores per game in English Premiership rugby (2012-2021)
To further test the proposed hypotheses, a scatter plot of the effect of temperature on the number of possessions that resulted in tries scored, kicks out of hand, and turnovers conceded as a percentage of total possessions was generated (FIGURE 2). A regression and correlational analysis of all 3 variables revealed:
There was a significant positive relationship between temperature and try scoring frequency (r(95) = 0.42, p < 0.05)
There was a non-significant positive relationship between temperature and turnover frequency (r(95) = 0.19, p> 0.05)
There was a significant negative relationship between temperature and kick frequency (r(95) = -0.37, p < 0.05)
FIGURE 2: The effect of temperature on the number of tries scored, out of hand kicks, and turnovers conceded as a percentage of total possessions in English Premiership rugby (2012-2021)
The first insight pertaining to try scoring frequency complements what was initially revealed from the analysis in Figure 1. We know that scoring has a positive relationship with temperature and now we can also add that this is tied to the number of tries scored. The data suggests that the number of tries scored decreases as temperature gets colder and this is likely one of the main drivers behind why fewer points are scored at colder temperatures. However, it is currently not clear how goal kicking contributes to the scoring relationship to temperature (e.g. Do teams attempt more penalty goals in lieu of kicking to the corner in cold weather? Is goal kicking accuracy affected by temperature?), however this is something that could be investigated in future studies.
The results around turnovers conceded bring forth some interesting discussion points. A visual inspection of the scatter plot suggests that there is a slight positive relationship between temperature and the rate of turnovers conceded since the slope is positive (m = 0.0009). However, this relationship was determined to be statistically random as the correlation coefficient (r = 0.1889) was not significantly different from zero. This means that temperature does not appear to have a meaningful effect on turnover rates which runs counter to proposed Hypothesis 2. What is interesting about this finding is that even if the relationship between temperature and turnovers conceded was found to be statistically significant, Hypothesis 2 would still not be supported as the trend in the data suggests that as temperatures get colder, the frequency of turnovers conceded decreases.
Finally, when it comes to kicking out of hand, this analysis supports Hypothesis 3 which proposes that as temperatures become colder, teams will elect to kick the ball more. As another general rule of thumb, for every 4°C that the temperature drops, teams will increase their frequency of kicks by ~1%. It is currently not known how the frequency of contestable kicks (e.g. box kicks, territorial kicks, bombs, etc.) versus attacking kicks (e.g. chips, grubbers, cross pitch, etc.) will change in colder weather but again this is something that can be investigated in future studies.
It’s interesting to note that temperature has the greatest effect on kicking (compared to scoring and turnovers) because it has the largest slope and the highest correlation. This is interesting since kicking is a tactical choice by the attacking team to give up possession - and therefore a chance to score – in order to gain territory. At the same time, teams will have less control over their turnover and scoring rates which is consistent as to why they appear to be less sensitive to temperature changes. The conventional reasoning behind kicking in bad weather is that there is a greater probability of turning over possession if teams try to play too much with the ball. Thus, the closer you are to your own goal-line, it is better to give up possession and gain territory rather than continuing to play and turning over the ball in your own half of the field. Nonetheless the findings of this study suggest that teams are more likely to kick the ball away in colder weather even when there doesn’t appear to be a greater probability of turning over the ball. Thus, is this an opportunity to challenge conventional thinking around game tactics in the winter? Are teams playing too conservatively in response to the perceived risk of a greater probability of turning the ball over in colder temperatures when in fact the risk is no different from playing under warm and dry conditions?
Another thing to consider is that even though teams may choose to play a less than optimal style in cold weather, are there enough cold weather games played in a season to actually make a difference in the standings? Figure 3 shows the distribution of the number of games played at various temperatures from 2012 - 2021. The 353 games played at temperatures less than 5°C represented 27.8% of all games played, which translates to ~6 games on average in a 22 game schedule. Given how competitive it is to make the playoffs in the current Gallagher Premiership competition, if you’re a team in contention to make the playoffs (or even battling relegation) wouldn’t you want to optimize your team’s strategy and tactics during these 6 games?
FIGURE 3: Number of games played at various temperatures in English Premiership Rugby (2012 – 2021)
One of the main limitations of this study was that temperature was analyzed in isolation with the assumption that it functioned independently of all other weather factors. In reality though, cold winter weather in England is typically accompanied by wet conditions (rain or snow) so it is likely they are covariates and therefore it will be difficult to statistically isolate their effects independently of each other (e.g. think of temperature and wet conditions as being like Siamese twins). Whether or not temperature and other weather factors behave as covariates still has to be determined. Data for the other weather factors (humidity, weather condition, and wind speed) was collected but not analyzed for this study. Once we have determined the impact of each of these other factors in isolation, then we can start combining them systematically to check for any significant interactions. Another limitation to consider is that this work is preliminary and based on correlations that demonstrate that there are relationships between weather and various aspects of performance in rugby. However, since correlation doesn’t necessarily mean causation, further work investigating the details around these relationships will be necessary before these insights should be put into practice.
While there is a lot of work left to do, these findings around temperature are a promising start towards investigating the impact of weather on performance in English Premiership rugby. For your convenience, here are the key findings of this study:
CONCLUSIONS:
Scoring decreases when temperature decreases
The decrease in overall scoring in colder temperatures can be attributed to fewer tries being scored
Rates of turnovers conceded are not impacted by temperature
Teams kick more when temperature decreases
27.8% of all games played in English Premiership rugby are played at temperatures less than 5°C
These results suggest that teams tend may tend to play too conservatively by kicking more in response to cold temperatures. However, further work is required to better understand the optimal strategies to implement in response to winter conditions in English Premiership rugby.
Comments