Round 17 of the Gallagher Premiership brought us some exciting predictions and results.
Our Rugby Outcome Score System (ROSS) started strong and has continued to improve as the season has progressed due to a combination of factors. Teams are finally getting players back from international commitments and with every round of games, more data gets loaded into the model meaning the relative strengths of each team are better estimated, and hence the predictions become more accurate.
In round 17, Northampton Saints’ win over Leicester Tigers went against the predictions of the bookmakers, but ROSS calculated the scoreline exactly but for a single conversion! Lifting the lid on the ROSS prediction reveals Saints were expected to score 3.3 tries on average for this fixture compared with Tigers’ 2.5. Saints were also expected to kick penalties (worth 3 points) more, with an expected average of 1.5 for this fixture compared with Tigers’ 0.6. Tigers have a slightly higher expected conversion rate than Saints (71% vs 65%), but of course, a team needs to score tries before they have the chance at a conversion. Taking all this information into account, it was hardly surprising to ROSS that Northampton Saints came out of the game victorious.
Another match that was hard to split before kickoff was Wasps v Bath, but not for ROSS. This match was predicted by ROSS to be a tussle between a superior try-scoring team (Wasps) and a superior penalty kicking team (Bath). Wasps were expected to score on average one more try than Bath (3.8 vs 2.8). Bath conversely was expected to get nearly one more penalty kick on average (2 vs 1.1). With a nearly identical conversion rate, the predicted winner of the game was effectively decided on account of the fact tries score more points in a game of rugby than kicks, a prediction that came to fruition!
Another pattern that has emerged since ROSS was created, is its tendency to predict draws. Since its first deployment for round 11, ROSS has predicted 5 draws, including two of the six fixtures in round 17. This may initially come as a surprise as draws in rugby are a rare match outcome, with only 2 draws occurring in 95 fixtures in the Gallagher Premiership so far this season. ROSS, however, does know that draws are rare outcomes in rugby and it’s predictions can be thought of as the average expected result. Therefore, most of the time ROSS predicts a draw, it is actually telling us that neither team is favoured to win the match rather than saying a draw is the most likely outcome.
There are many factors as to why draws are so rare in rugby. One of those is the nature of teams risking missing out on a potential draw in efforts to get the full reward of winning the match, which can swing a game either way. A good example of this being in the Bath v Leicester fixture in Round 16 where ROSS predicted a 28-28 draw and Bath won the game by a solitary point.
Similarly, something ROSS cannot account for is player availability and selection. ROSS predicted a draw in Round 17 between the two stand-out teams in the premiership this year Bristol and Exeter. However, Bristol was without some key players who have been vital in their success this season, with the likes of Semi Radradra, Nathan Hughes, and Callum Sheedy all absent. Players such as these have been in fantastic form for Bristol and played big parts in the wins and points scored that are fed into ROSS to predict the next weeks’ outcomes. However, for the most part, the brand of rugby teams set out to play and how successful they are in executing that don’t often change due to personnel. Therefore, ROSS knew the game wouldn’t be particularly high scoring and that neither team would be able to break down the other well enough to score 4 tries and secure a bonus point.
Finally, ROSS forecast a draw and told us it couldn’t split Gloucester and Newcastle, two teams in poor form. Newcastle raced out to a 7-21 lead at halftime but were struck down by injuries in the second half, including all 3 of their second rows, which allowed for Gloucester to turn the game around and capitalise late on and secure the win. The topsy turvy nature of this match and the in-game factors surrounding the Gloucester comeback just shows that, although on the face of it ROSS predicted a draw, it was actually telling us that either team was just as likely to win the match. Fortune seemed to favour Gloucester and the outcome went their way!