Some rugby data: This has been something I've been working on for the last couple of nights from an Opta data set of the Aviva/Gallagher from 2017-18. The plan was to identify what elements of the game correlate best with success, specifically looking at tackle types (1)
There's some statistics going on here but basically you only need to know about RSquared or R2(superscript 2 in the graphs). An R2 number of 1 means perfect correlation - so for example the age you are and the school year you are in, as one goes up the other follows perfectly (2)
An R2 number of -1, means perfect negative correlation - so the amount of time it takes you to complete the 400m and your overall position. As your time goes up your position goes down (down if first is the peak). An R2 of 0 means zero correlation.... (3)
Team performance and average hairline height of a coach for example. Anything between 0.1 and 0.99 indicate various levels of correlation with 1-0.5 being strong, 0.5-0.3 being moderate, and 0.3-0.1 being weak (4)
I initially wanted to see if there was any correlation between tackle types and success. Do teams who have more dominant tackles do better than those who have fewer dominant tackles, and is that relationship significant? (5)
I wanted to start off by investigating whether defence or attack was the more important element of the game. This graph shows all pts scored by league pos. Unsurprisingly it correlates pretty well with overall performance but only an R2 of 0.63 - which is lower end of strong (6) 
Next one is Pts conceded by league pos. The more pts you concede the lower your pos - obvious. But, look at the R2 - 0.82. That's a much stronger correlation for success than pts scored was. That implies that if you want to finish high in the league then work on the DFence (7)
Tries scored by league position is similarly a less good indicator of success than pts conceded - R2 of 0.46405 means that it's only a moderate indicator of success. (8)
Tries conceded by league position is as similarly useful as pts conceded, R2 of 0.80174. Tries conceded is very slightly less good at predicting success than total pts conceded but, if you want to work on one thing, work on stopping tries not scoring them (9)
Understandably elements of defence will never be as correlated to success as defence as a whole. However, what is interesting is that total missed tackles is a better indicator (0.23808) than missed tackle % is (0.15543). We should care more about missed tackle # not tackle %(10)
On the flip side, dominant tackle % (0.23218) is a better indicator of success than total dominant tackles (0.15793). I have no idea why this would be the case for dominant tackles but not for missed tackles - it could be a result of the relatively small sample possibly (11)
So, if you're still with me, the takeaways - at least for the Aviva Premiership from 2017-18 was that defence wins leagues and a good defence is a much better indicator of success than a good attack. I'll be doing the same analysis for a couple of other leagues and years (12)
to see if that holds true across the board or whether 17-18 in England was a fluke. I might also take a look at specific attacking stats and how they correlate to success. Any questions, let me know! (13)
