Analysis: Going beyond rugby statistics

Adam Kyriacou

Our analyst Sam Larner is back this week and is looking outside the box at rugby statistics, with some interesting findings in his work.

What follows isn’t meant to be, and hopefully shouldn’t read like, a rugby version of a maths textbook. I am in no way a maths expert so everything complicated will be explained fully here.

I was also helped out by my a more statistically wise friend, James Gibbins, whose Whatsapp over the last couple of weeks has been filled with half baked ideas for new metrics which he has helped form into something coherent. Also a big thank you to Jac Larner, who usually wastes his time looking at unimportant matters like politics and voting tendencies but has been lured out to use his statistical skills on rugby. As you can guess from the name, we are related so you can imagine how much fun Christmas dinners are in our family.

You might ask what the problem is with the current stats; tackles, passes, carries etc. Well, there’s nothing wrong with looking at the number of tackles someone makes and getting an idea about work-rate, but you ignore a load of context focusing on that stat alone. There are three outcomes to a tackle; a dominant tackle, a passive tackle, or something in between which is neither good nor bad.

If players A & B both make 100 tackles you would think they were equally good, but if 40% of A’s are passive and just 10% of B’s are then it’s fairly clear that B is the better tackler. The numbers below are going to go beyond the stats you’re used to seeing and will hopefully do a better job of explaining who the better players actually are, and, if you end up winning a debate because of these stats, I do take payment in beer.

Index

If you already know about indexes then feel free to skip ahead. Lots of these stats are put on a scale to help you understand what they actually mean, if the stats are put on this scale then we will highlight this by mentioning an index. If we just gave you the raw stats you would be able to tell that person one was better than person two but you wouldn’t know where they fell against the average. To get around this, and to make it easy to understand the results quickly, we have placed the data on the scale with 100 as the average. Anyone with a score of 100 is bang average in that data set – Newcastle prop Jon Welsh fits this bill for the Defensive Impact metric – if you have a score of 110 then you’re 10% better than average, 80 and you’re 20% worse than average, 200 and you’re twice as good as average. There’s theoretically no top to the scale but 0 is as low as you can go.

Lifter Independent Hooker Rating (LIHR)

Who is the best hooker in the Premiership at throwing in? Is it just the hooker with the best line-out success? Well, it turns out that it is, but it’s not quite as simple as that. The success of a line-out might be based largely on what the hooker does, but the work of the rest of the forwards matters as well. You could look at every line-out and try and work out how much of the success or failure was down to the hooker or the lifters/jumper. That sounded like too much work, so I worked out how many throws each team have per lost throw – the same as success percentage really – and then I worked out that figure for the individual hookers. To give an example, Richard Hibbard throws 4.2 successful throws between each unsuccessful, that’s not good, and Gloucester’s average is 5.5, also not good, so Hibbard gets a score of -1.3 meaning he is 1.3 successful throws worse than the team average and therefore it follows that Hibbard is more responsible for that element of poor performance than his jumpers/lifters.

Jamie George has been incredible, he’s missed just four throws in 118 opportunities, he averages 28.5 successful throws in between each unsuccessful one. Schalk Brits is the third best hooker by throw success rate but he slips down this list because of how high the team average is, largely because of George. One other person of note on this list is Rob Buchanan, he’s only thrown 25 line-outs all season, so don’t read into this too much, but he is significantly ahead of a truly awful Quins hooking trio – the trio average 5.4 successful throws in between each bad one, just behind Gloucester who have 5.5, Saracens have 13.8.

Dylan Hartley comes from the fourth worst ranked Saints team but he still scores -0.86 compared to the team average, that’s 5.6 successful throws per missed one.

Exeter Chiefs are the next best team after Saracens, 10.5 successful throws in a row, but that score is heavily reliant on Luke Cowan-Dickie (22.2), they’re 15 successful throws worse when Jack Yeandle takes over.

Tom Dunn (Bath) is the seventh best hooker by throw success rate but he doesn’t make it onto this list because the third ranked Bath team (9) have an incredibly consistent trio of hookers.

Defenders Beaten Per Try

Some players have it easy and get handed tries on a plate, some players have to work hard for every single try – both get given 5 points for their efforts. We’re not happy with that though, so we’ve created this stat which takes the total number of defenders beaten and divides it by the number of tries scored – a high number of defenders beaten per try indicates a player working harder for his try.

It’s not a big surprise that Sinoti Sinoti is ranked highest in this stat, and comfortably so. The Newcastle wing has built a reputation for quick feet and making defenders miss and that is reflected in this statistic. One thing he won’t be happy with is only scoring four tries in the season, Telusa Veainu has scored 3 more tries and beaten 17 more defenders in total. Josh Strauss’ and Thomas Young’s performance is hugely impressive, the back -rows both rank significantly better than the average in this stat.

The stats in this table only include players who have scored four or more tries, if we discount that then Jake Polledri (Gloucester) would rank in first place with a score of 28, Mike Haley (Sale) would be second with 26 and Joe Marchant (Harlequins), 25.5 would fill out top 3 – all have score two tries.

The first qualified front-row player (>3 tries) is Kyle Cooper (Newcastle) who scores 6.75 – good enough for 101.9 in the index, which includes all players.

Lastly, 29 wingers or full-backs rank below 100 in the index, these include the likes of Byron McGuigan (33.95 index score), Matt Banahan (42.25) and Vereniki Goneva (65).

Fun Fact 1 – Restart Success

Six players have failed to collect a restart four or more times this season, these consist of four locks and one wing. The five players who have failed to collect a restart four times are; Mitch Lees (Exeter, lock), Michael Paterson (Northampton, lock), George Kruis (Saracens, lock), Christian Wade (Wasps, wing), and Bryce Heem (Worcester, wing). The only player with five failed takes is Ben Glynn (Harlequins, lock) – thankfully, Glynn has taken more catches than he’s failed, Wade has been unsuccessful four times and successful just twice. Jackson Wray (Saracens, back-row) has the most successful takes, 35 but Sione Kalamafoni (Leicester, back-row) has the most successful takes without a single failed take.

Hardest Yardage Index (HYI)

The metres gained statistic is useful, but it tells us more about opportunity than ability; full-backs and wing tend to gain a lot of metres because they get the ball kicked to them in space and get to run for a while before they have to come face to face with defenders. Metres gained doesn’t tell us who the best ball carriers are, instead we have to look at their impact when they get the ball and for this we look at how many defenders they beat for every 100m gained.

The gap between Ellis Genge and Cooper might look small in the table above, but Genge has a score of 250 on the index compared to 220 for Cooper. As we discussed, the back three tend to gain metres with ease compared to the rest of the players and it’s no surprise that there is only one wing in the top 25 – Alofa Alofa. England will be pleased to see both Ben Te’o and Manu Tuilagi appear in the top 10, both offer a similar style of crash, bang, wallop hard running. Without ruining next week’s article, Te’o ranks dead last in passing ability. Finally, unsurprisingly for anyone following the Premiership this year, Polledri ranks pretty high on most of the stats and here he is the only Gloucester player in the top 30.

Matt Smith (Leicester) doesn’t manage to get an index ranking, he has gained 215 metres but is still yet to beat a defender. Everyone else on the list has beaten at least three defenders – Lewis Ludlow has the next lowest defenders beaten, 3, which puts him last in the index (12.9).

18 scrum-halves made the list and just three rank higher than 100; Sonatane Takulua (Newcastle, 118.1), Ben Youngs (Leicester, 120.4), and Willi Heinz (Gloucester, 139.4).

Five hookers made the list and Cowan-Dickie was unfortunate to miss out on the top 10, he was 11th with a score of 9.2 defenders beaten per 100m or 177.3 on the index.

Fun Fact 2 – Defensive Penalties

Of the 532 players included in the data – everyone who has played at least a minute of game time this season – 74% have conceded at least one defensive penalty. Of those 74%, just four have over 20 penalties; Maro Itoje (20), Kyle Sinckler (21), Welsh (22) and Ben Franks (27) – Franks also has the most scrum offences, his 25 are 11 more than the next worst – Welsh. Gloucester lead the way in offsides though, two players are joint worst with 7, Billy Twelvetrees and Lewis Ludlow. Finally, you might not have heard the name Cameron Holenstein, the Harlequins prop, but he has the highest rate of foul play per minute – he got pinged for foul play once every four minutes – he only played four minutes though.

Conclusion

These stats shouldn’t be too surprising in the most part. They’re not meant to completely redefine how you look at the sport of rugby, if they were doing that I would question the quality of the logic and calculations I used to design them. What they are meant to do is disprove some commonly held myths and enhance your understanding of the game. Too often we see the most recent result and think it means something more than the vast majority of results that have come before, this is true in life not just in rugby. Back when they made their debuts, I was not a fan at all of Scott Williams or Dan Biggar and I wasn’t really willing to step down from this viewpoint, my Dad has never really let me know forget this as both have turned into, at points, world class players. If I’d had some additional stats back then, I might never have held such an embarrassing position.

Next week, we’ll be looking at defence, tackling, goal-kicking and Don Armand.

by Sam Larner