There is no "I" in Team
Over the past few weeks we have published some of our player ratings on various forums, and these have initiated some hot debate, and to be honest some questions about our sanity!
For example how can we have James Tedesco at number 32! Are there really 31 better players?
So this week we are going to delve into the detail and hopefully answer the question.
The approach to player ratings
First and foremost we are a tipping and betting site and not a player ratings one – but our approach to predicting the margins is heavily influenced by the players and their ratings. There are two key factors on the player side:
Games played rating – simple put a statistical analysis of historic games has shown that players with more experience have a greater impact on the result. We use regression analysis to determine what that impact is and effectively fit a curve. So a 50 game player statistically has more impact than a rookie, but less than a 100 game player. So two players with the same games experience have the same “experience impact”. At this stage we do not factor in where they play, or how good they are, just how many games they have played. To get to 100 games say, you must be pretty good!
Player impact rating– The second element is purely how much impact a player has on a game, and ignores experience. The player impact is largely driven by super coach scores, both current season (last round, last 3, 5 and 10) as well as prior seasons. Again statistical analysis has shown that this has a material impact on the result of a game. Now this factor does indirectly take into account things like position (as certain positions score higher), time on ground (can’t really influence a game from the bench) and general influence on result (eg tries scored). So in simple terms a player with a higher super coach score will have a higher player impact rating (it is weighted by other factors but you get the idea).
An example:
James Tedesco is widely considered to be amongst the most influential and best players in the game. But he has only played 58 (including round 6) games, so our model gives him a lower experience rating than say a 100 game player. So let’s compare him to a team mate with very similar games played, Kevin “Kevvy” Naiqama (noting hair does not come into the model but if it did…..):
So Tedesco’s ranking position is higher, with the impact on margin estimated to be 1.73 points – so if the game was a one player shoot em’ up, we’d predict Tedesco would win by nearly 2 points on average each week!
What about all players with similar experience? And then compare it to say the 100-120 game category.
So there are far less players with 100-120 games under their belt, and as you can see on average they have a bigger influence on the margin. You need good eye sight but Tedesco is higher than the average for the 100-120 group and is only just behind Shaun Johnson.
Now as Tedesco has less games than Johnson his experience ranking is lower but he nearly catches up with player rating advantage – to put it another way, if he had 100 games, Tedesco would be rated higher.
Best players by games played
To see the overall impact of experience we have grouped all players by games played and highlight the best player in each category (we know Sam Burgess isn’t there! His last two games (albeit one injury interrupted) have lowered his player rating – he will bounce back!).
Whilst this is all very informative (hopefully) and has a big impact on the prediction, there are 17 players in each team.
It’s all very well having the best player but there is no ‘’i” in TEAM!
Let’s look at the Tigers compared to the comp average and this week’s opponents, the Knights.
So Tedesco is actually marginally behind Aaron Woods in terms of influence on the Wests side and both are above Jarrod Mullen, the Knights highest rated player.
So what does all this mean? Well we combine the player ratings with team based factors (ladder position, home ground advantage, form etc) to get an overall prediction.
So this week we have the Wests winning by 9, mainly due to players like Tedesco. The Knights have a slightly better team based rating just because they are at home.
Now we know our system isn’t perfect and like all statistical based models the outliers are the ones you can’t model well. Tedesco is a gun, and clearly the best of his cohort. Should we increase his influence in the model? Potentially, but how, without invalidating the statistical analysis? Maybe there’s another factor?
As always we welcome your feedback. Good luck with the tips and bets (or if you want to take luck out of it have a look at our site)!
GUEST WRITERS WANTED – Do you have a view on anything NRL related? Maybe a response to this article or something else. We are happy to post here* and make you famous (well sort of).
To see our betting recommendations, including our Bet of The Week, and to read more about our approach, and to compare our performance with your own, please visit www.tipbetpro.com or follow us (abuse us) on Twitter @tipbetpronrl.
*Subject to editors discretion
