Rugby Pass stats nerd stuff



  • I haven't totally gone deep into it yet, Ardie the world best player atm 😉



  • @machpants He's the tallest, not the best.



  • Same thing, surely?



  • Doesn’t pass the BBBR litmus test.



  • @machpants not if you are vertically challenged.



  • I've had a good look at it and there are many mistakes.

    Look at the ranking of first five-eighths in Super Rugby and check out who they've ranked as the 15th first five-eighth. The poor lad can only dream ... JGB is also ranked way too high. And you won't find Dmac in that ranking, because - despite having played at 10 all season - they've ranked him as a fullback. 🤦🏾♂

    Oh, and try finding Matt Todd among the openside flankers ...



  • I thought it looked okay for a minute, then I saw they rated Maro Itoje as the second best player in the world. 👎



  • Ahahaha at their algorithm:

    Bono Cockwomble rated fourth best number 8 in Super Rugby:
    0_1541148281167_f8a48b61-d3b9-405c-96bd-a4eeea7ec0b7-image.png

    Contrasted with the 13th.
    0_1541148322448_361439c9-c2b9-401c-a33d-4006ed5bd7e4-image.png



  • They have Ollie Woodburn (Exeter Chiefs) as 8th best player in the Gallagher Premiership. Odd that seeing as he had his first game this season last week playing for the B side - and got skinned for a try by his opposite winger.



  • How are they working out "Influence"? Those ratings are subjective horseshit hidden behind spurious use of data.



  • @sparky said in Rugby Pass stats nerd stuff:

    How are they working out "Influence"? Those ratings are subjective horseshit hidden behind spurious use of data.

    Influence
    The 'Influence' score is a way RPI measures the effect a number 8 has on their team's likelihood of winning. If they are involved in a play 2 minutes prior to a winning moment, their player ranking climbs. Conversely, if it's a losing moment, RPI will punish them for it. This means that the special features of a player's game is not lost.

    Pass Complete
    In unison with the half back, RPI number 8s use their expert distribution to create game-winning moments. The 'Pass Complete' score measures successful passes. Each time the number 8 completes a successful pass, their player ranking increases.

    Scrum Score
    Managing direction and distribution at the helm of the scrum, number 8s are a crucial element to any successful forward pack. RPI uses 'Scrum Score' to measure a number 8's scrum ability. If a forward pack turns the ball over from a scrum, the number 8's player rankings grow. Conversely, if a forward pack loses possession from a scrum, the hooker's player rankings suffers.

    Attacking
    RPI uses a group of measurements to assess a number 8's attacking ability. Each line break, defender beaten, and successful offload will improve a number 8's player ranking.

    Standard Factors
    In their role as the bridge between forwards and backs, RPI number 8s are measured on diverse criteria:

    Carry Metres
    Line Tackle
    Lineout Score
    Lineout Take
    Other Tackle
    Missed Tackle



  • neat idea, lacklustre execution.



  • @antipodean said in Rugby Pass stats nerd stuff:

    Influence
    The 'Influence' score is a way RPI measures the effect a number 8 has on their team's likelihood of winning. If they are involved in a play 2 minutes prior to a winning moment, their player ranking climbs. Conversely, if it's a losing moment, RPI will punish them for it. This means that the special features of a player's game is not lost.

    What on earth is a winning/losing moment? Does this mean they'd get punished for being busy on defence before the opposition score?

    If a forward pack turns the ball over from a scrum, the number 8's player rankings grow. Conversely, if a forward pack loses possession from a scrum, the hooker's player rankings suffers.

    What?



  • @antipodean What on earth do they mean by a "winning" or "losing" moment? This looks like a concept they have exported from another sport.

    Every single second in a Rugby match matters.



  • From the website:

    How the RPI scoring works

    We have built a revolutionary rugby rating system based upon individual skill executed in real time in partnership with leading academics and sports professionals from around the world.
    
    Our approach is all about winning, and the contribution to winning. We’re dealing with professional sports people – it’s not about participation and making up the numbers. Our research shows that a decent team rating system beats popular opinion. More importantly, a team rating combining individuals beats the best team rating systems. Winning a game is about winning moments – quantifying this impact provides an even richer perspective of behaviour.
    
    This approach points us in a unique position as we can say that V is the most valuable player now, with an individual rating of W, because they rate high on X and Y which improves their team’s chances of winning by Z%.
    
    We now know that to win games in an elite rugby environment, key moments in a game need to be won. Using a combination of demand forecasting, survival analysis and concepts from statistical process control, we have been able to quantify the impact of moments on an overall outcome. Importantly, our approach is focused on winning. That is, what does it take to win?
    
    The emphasis on winning and predictivity ensures we obtain parsimonious models that are aligned with perception and are more readily interpretable and as a consequence, defendable. For instance, when we compared a team rating system for teams combining individual ratings with a team rating based only on team performances we obtained a 13% improvement in predictivity.
    
    As we usher rugby audiences into a new era of appreciating the game through informed analytic insight, we need to build trust. Our framework, geared around moments, in real time lends itself to a transparent framework. Importantly, as we expand on team ratings using individuals we are in a strong position to understand and compare competitions, due to relative performance and depth of talent. We can also expand this to comparing generations, positions, partnerships, coaches to name a few issues that consumers would find interesting.
    
    We are constructing an expert system. That is using a combination of machine learning, shaped by winning outcomes and guided by human observers, we are creating a dynamic system that will output meaningful, rugby orientated output that will stimulate, engage and challenge thinking of those interested in rugby across all levels.
    

    "informed analytic insight"

    {insert Tui ad here }



  • Behind the sophistry, what they define as a winning moment is essentially subjective.

    Their system looks nice but it is no more scientific than a MOTM poll on the Fern.



  • @sparky Probably less so given we ferners are all rugby experts!



  • It seems they have misunderstood the meaning of "transparent".



  • @bones said in Rugby Pass stats nerd stuff:

    It seems they have misunderstood the meaning of "transparent".

    You don't think the meaning is clear?



  • @nzzp said in Rugby Pass stats nerd stuff:

    @bones said in Rugby Pass stats nerd stuff:

    It seems they have misunderstood the meaning of "transparent".

    You don't think the meaning is clear?

    alt text


Log in to reply