Friday, March 01, 2013

Stat of the Nation Address

Who ate all the pie charts? Maybe it was @RugbyKino as he crunched the numbers on Irish coaches since 1993…

Born out of idle curiosity, and a combination of recent wild accusations and gross generalisations in regards to our national team’s performance, I decided to have a peek at Ireland’s international record going back as far as 1993.

All the graphs that follow plot Ireland’s matches won as a percentage of matches played over time, with draws being excluded from the win total. Each manager over that period is graphed separately with the percentages calculated purely on their own results, and an average for the full 20 year period is shown in each case. I’ll try my best to keep the comments to the hard data without bias, and they’ll be minimal. Consider it an invitation to draw your own conclusions from them.

Win over Mng

Initially we can see here that the transition to professionalism was not an easy one. However, the groundwork that seems to have occurred under Warren Gatland’s reign, with a rather healthy and relatively consistent upward trend looks to have been capitalised upon by Eddie O’Sullivan after a slightly shakey start. A slow decline in performance was almost instantly improved upon by Declan Kidney, which has more recently shown a drop towards the average.

Home Win over Mng

Separating out the home results, we can see that it was a rather mixed bag under Gatland, but certainly was improving in the latter two thirds of his tenure. O’Sullivan started very strongly with a 9 match winning run, and with a low that was not much less than the best results achieved since. Again, Kidney improved the results initially, but has overseen a substantial dip in form since, with a current level below that at which Gatland finished his tenure.

Away Win over Mng

With only away results into account we can see certain familiar characteristics again. Gatland’s side, after an early dip in form, were steadily improving to the end. O’Sullivan peaked early, and was relatively constant thereafter, though dipped below 50% at the end. Kidney’s results are visually similar to O’ Sullivan’s, with a resurgence followed by a slow decline in results, both graph sections ending at a similar percentage.

SANZAR over Mng

Taking all results against the SANZAR nations, and 5 Nations countries to allow for historical consistency, we yet again see some familiar traits. Gatland’s side were constantly improving. O’Sullivan arguably delivered our most consistent results over his tenure, but with a small drop off towards the end. Kidney presided over an initial surge, but the results have since waned to below the average over the 20 year period in question.

Win over 20

I didn’t feel like I’d gained any stunning insights yet, and went in search of that most elusive metric: consistency. I graphed the percentage of wins over sets of 20 games, roughly a period of a year and a half. This would remove the ‘noise’ in the data caused by injuries, blooding new players, the transition between coaches and so forth. The result is consistent with previous characteristics: a consistent rise in performance under Gatland, a peak during a relatively steady reign with a slight fall-off at the end with O’Sullivan at the helm, and a resurgence followed by a dip during Kidney’s era.

However under Kidney we can see something which is not hugely apparent under the previous graphs. When the figures are primed to look at consistency, the Irish team have not performed above the average of a 53% win rate over 20 games since the point at which his performance graph crosses the Rubicon of the 20 year average: the 6th of August, 2011, when they were beaten in an away match in Murrayfield. So when people say that Kidney’s cards were marked by a loss away to Scotland, they may not realised how right they are.

*Note for Stats Nerds: In the final graph I’ve included a 6-order polynomial trend-line which correlates to the filtered dataset with an R² value (a measure of how closely the trend line matches the data on a scale between 0 and 1) of 0.82. A value above 0.75 indicates a generally sound correlation to the data available, but the figure is not an indication of how accurate any trend that is forecast from it may be. I’ll let you draw your own conclusions from the trend-line itself.

@RugbyKino is a constant supporter and sometimes scribbler of all things rugby, and can be found lurking on twitter and the terraces of the RDS.


Taken by JLP from RDS press box on Nov 16, 2019