Moneyball Rugby
-
The recent discussion about the relative worth of players in comparison to their peers made me look up some stats on backrow players in Super Rugby. Unlike baseball, it appears to be more difficult looking for a specific skill that you want from a certain position. Apparently, it's not as easy as "make tackles".
This is what I had to work with:
- Tries
- Points
- Carries
- Metres
- Clean Breaks
- Defenders Beaten
- Offloads
- Try Assists
- Passes
Open play kicks- Lineout catches
- Lineout steals
- Tackles
- Missed Tackles
- Tackles Success
- Turnovers Won*
- Turnovers Conceded
- Yellow cards
- Red cards
*Frustratingly for flankers, I could not locate that stat for every player. Suffice to say that this year Pocock was clearly the best at it, followed by Savea.
I took every player who had played more than four games and 200 minutes to exclude the outliers, worked out their involvements against the above measures per minute played and ranked them.
Then I realised that if you play against weaker teams, your stats are more likely to look good, so I normalized the data depending on their final ladder position, i.e. if you play the Kings and Sunwolves, you're more likely to rack up big carry metres etc. than you are if you're playing against those who made the finals...
And I got sidetracked.
-
@antipodean So what you're trying to say (from my quick glance), is that the Blues have the best loosies in NZ?
-
Sidetracked:
Looking at the relative merits of the opposition that teams got to play through the season, quite a discrepancy became stark. If you add up the final ladder position of teams and ranked that numerical value you can see that some teams had a remarkably easy ride compared to others. For example; the Stormers came fourth on a combined ladder but only played one who made it into the top six. Their opposition's median position was ranked 13th. The Kings on the other hand only had five outside of the top 10.
Here's the ladder position based on the final standings of their opposition:
-
The big problem is the raw stats miss out the effectivness. On the above Kaino & Cane are mid table mediocrity. But the effectiveness of what they do is the key.
EG purely on tackle making. 10 tackles where you cling on as the guy makes 5 yards before falling over & setting the ball perfectly is better than making 6 tackles where you cream the guy on the gain line, drive him back & let your scavanger get hands on the ball.
The Wayne Smith take on this has been posted a stack of times. Its why the AB selectors have guys adding effectiveness stats to all those raw ones.
Baseball its infiniately easier as you have "hits" & to get "effectiveness" you have "gets on base". In rugby you could have a guy batting 1.000, but never once getting on base. In theory he's awesome, in reality he's fucking useless.
-
@Bones I'd say it's the strength of their team. Given they also have outside backs aplenty, it really demonstrates how much they require a decent 9/10 axis.
@gollum said in Moneyball Rugby:
The big problem is the raw stats miss out the effectivness. On the above Kaino & Cane are mid table mediocrity. But the effectiveness of what they do is the key.
Yeah, It's a point I've made countless times citing the Professor Wayne Smith.
What this little exercise did do for me was to highlight that Pocock is clearly the best in the SH at pinching ball. The problem is he does it in a team that does nothing with it. Almost as if they're not coached to take advantage of the attacking opportunities this provides. What he's poor at is running with the ball. Given he isn't a lineout option, it seems odd to have someone who adds nothing on the attacking front.
So when people say he'd make the All Blacks, the stats point out he wouldn't. There's no way the coaches would have someone who can't be an attacking threat as a flanker.
Just as an addendum on the effectiveness issue; if you could quantify the effectiveness of Sam Cane's tackling, he'd be much further up the ladder.
-
Quality/Effectiveness is obviously a key metric missing from the raw stats.
The other one is ruck and maul involvements. It's quite incomprehensible that this still isn't included by almost any stat provider. (Is Haka still the only one who does it?). And it quickly makes an exercise like this pretty redundant (sorry antipodean). At least for forwards it does.
The guy securing the ball often has just as an important role as his team mate carrying the ball.
It's like saying Ben and Owen Franks didn't do anything in those final few minutes in that buzzer beater in Ireland, according to the stats. When we all know that if they weren't hitting rucks like their lives depended on it, the All Blacks don't win.
-
I like this sort of theoretical exercise Antipodean and I commend you on your efforts.
As has been pointed out it the danger of using pure stats alone to judge rugby.
In addition to "effectiveness" you'd have to consider "usefulness" of a given stat.i.e., did you give any weighting to the various stats? Is passing (number of "passes") as important as "carries" for example?
P.S. @gollum I thought batting 1000 was getting on base?
-
@Frye said in Moneyball Rugby:
Quality/Effectiveness is obviously a key metric missing from the raw stats.
The other one is ruck and maul involvements. It's quite incomprehensible that this still isn't included by almost any stat provider. (Is Haka still the only one who does it?). And it quickly makes an exercise like this pretty redundant (sorry antipodean). At least for forwards it does.
The guy securing the ball often has just as an important role as his team mate carrying the ball.
It's like saying Ben and Owen Franks didn't do anything in those final few minutes in that buzzer beater in Ireland, according to the stats. When we all know that if they weren't hitting rucks like their lives depended on it, the All Blacks don't win.
This is especially important when playing rugby on the PS4! Yes, I'm frustrated with my players making breaks and then the support players just standing by the ruck as the ball is turned over.
But yeah, those Franks brothers were just picking up daiseys against ireland.
-
this is the exact reason i've always been dubious on anyone putting too much faith in stats in rugby. it's a complex game, and things like effectiveness and usefulness come in varying degrees, change at times of the game, on parts of the field, or opposition player involved - and are subjective anyway - any method of statistically modelling that stuff is of necessity going to involve arbitrary subjective decisions, which makes it no more valid than non-statistical analysis.
-
@Frye said in Moneyball Rugby:
Quality/Effectiveness is obviously a key metric missing from the raw stats.
The other one is ruck and maul involvements. It's quite incomprehensible that this still isn't included by almost any stat provider. (Is Haka still the only one who does it?). And it quickly makes an exercise like this pretty redundant (sorry antipodean). At least for forwards it does.
Agreed. All this shows is the relative impact of loose forwards who secure possession from set piece and attack with the ball.
I've no doubt that the order would be very different if I could include breakdown work (the core role IMO). It seems that Tracey at haka.co.nz stopped doing them after the Wales series, I can't find them anywhere else.
-
I think theoretically you could devise a model of all rugby players and work out how valuable they are. Right now, we just don't have the stats to do something like that. If there was a Major League Rugby in the USA and it was a huge sport, then you would absolutely have more advanced statistical models.
-
Interesting discussion. The key to the "gets on base" principle in baseball is that most runs are scored after one or more batters have got on base i.e. after a series of discrete plays which advance the batting team closer to scoring.
Rugby is a more dynamic game than both baseball and American football - both those sports are a series of intense / quick, discrete "set plays", which then end quickly and a new set play begins with everybody back in position (the NFL has occasional exceptions to this). In rugby, a try is typically scored after a series of phases (yeah, yeah, I know the All Blacks often score "quick strike" tries, such as from turnovers), but the general nature of the game changes after each ruck / maul, as there are differing numbers of players involved in the game after each succeeding ruck. Rugby also has multiple modes of attacking e.g. from a scrum, a lineout, a ruck / maul, in comparison to baseball.
Rugby is probably more similar to basketball if anything due to its fluidity and the proportion of plays determined by players' "ad libbing".
I enjoy both baseball and American football, but I don't follow basketball so I can't talk to the nature of statistics that come out of the NBA which could be adapted to rugby.
Pretty sure I added fuck all to the debate.
-
@antipodean said in Moneyball Rugby:
@Frye said in Moneyball Rugby:
Quality/Effectiveness is obviously a key metric missing from the raw stats.
The other one is ruck and maul involvements. It's quite incomprehensible that this still isn't included by almost any stat provider. (Is Haka still the only one who does it?). And it quickly makes an exercise like this pretty redundant (sorry antipodean). At least for forwards it does.
Agreed. All this shows is the relative impact of loose forwards who secure possession from set piece and attack with the ball.
I've no doubt that the order would be very different if I could include breakdown work (the core role IMO). It seems that Tracey at haka.co.nz stopped doing them after the Wales series, I can't find them anywhere else.
A while back Ben Strang on Stuff presented an analysis of stats amongst Cane, Savea and Todd (height of the Savea vs Cane debate). Included numbers on dominant tackles, rucks attended and participated in etc in Super rugby. So someone is collecting those stats but I couldn't find a public source. Which makes this a pretty useless post.....
