Saturday, July 27, 2013

Rankings & Results in the Champions Bracket

A fan of stats for paintball from Day 1 I haven't been quite as enthusiastic about some of the particulars (I find one or two of the stats to not only be not terribly useful but potentially misleading)--but that was before there was nearly two seasons of data collected. With the info currently available I wanted to do a rough analysis to see if (and how) player stats corresponded to team results. Yes, I realize it's not an apples to apples comparison but one still expects the teams with the (statistically?) best players to be among the best teams at least as a general sort of rule. So the question becomes do the stats bear that out? Not unequivocally as will be seen shortly.
In 2012 the top 7 pro teams were (in descending order); Heat, Damage, Infamous, Ironmen, Russian Legion, X-Factor & Dynasty.
So far in 2013 the top 7 pro teams are (in descending order); Dynasty, Damage, Impact, Heat, Ironmen, Infamous & X-Factor.
Six of the seven teams remain the same from one list to the other which is nearly ideal for this purpose. The obvious place to begin is with Dynasty. Seventh last season and first so far this season. And do the player stats inform that movement? They do.
In looking at the stats I divided the placements into separate tiers; top 10, top 25, top 40, top 50 & top 75 in order to see more precisely the relative placement of each team's players.
In 2012 Dynasty had 5 players in the top 40 but all were ranked between 11-40. In 2013 Dynasty has 7 ranked in the top 40 with 5 in the top 25 so it's plain to see that Dynasty's improved team results correspond to improved player stats too. Among the teams listed the Ironmen also seem to bear out the correlation though less convincingly as they had 5 top 40 players last season in finishing fourth but have only two top 40 players this season while holding fifth.
The other question that arises is do the player stats predict the results consistently? And the answer to that is no. Both seasons Damage held (holds) second place yet last season dominated the stats with 8 players in the top 40 while only having 4 this season. Russian Legion finished fifth last season with 6 top 40 players and still have 4 top 40 players even though they have been relegated once this season. Then there's Infamous. Finished third last season with three top 40 players and stand in sixth so far this season with three top 25 players.
At this point it's difficult to decide whether the Dynasty correspondence is an anomaly or if we need more information. When the numbers are extended out to the top 75 players (in 2013) 4 of the top 5 results make up the current top 4 in the rankings; Dynasty--9 players, Damage--7 players, Impact--9 players & Heat--9 players. The odd team out now is the Legion also with 7 top 75 players.
Running the numbers out to the top 75 gives us perhaps a better indication of potential roster depth but if it also provides superior corresponding accuracy in representing team results it opens up a whole new batch of questions about the relative meaning of the stats and the variations in placement.
More on that next time in--Stats & Status of Pro Rankings.


Anonymous said...

The problem you're experiencing is the stats try to normalize performance by weighting and averaging across games. In this sense they try to give you a clue about player performance in the abstract. If you want to see the best players line up with team performance simply take the raw points won and raw games won by the player/team and introduce that into the player metric.

What I'd like to see added and what I think is missed is who makes the moves/eliminations early on combined with whether or not their team won. Ya sure some guys clean up at the end of the game but in my book he more important guy is the one who gets the early eliminations in a winning game. I don't care so much trading out at the end of the game.

Nick Brockdorff said...
This comment has been removed by the author.
Nick Brockdorff said...

First of all, the biggest problem with the PBA stats, is that they do not separate prelims matches from finals matches, and obviously, it will be generally easier to make good stats in the prelims, than in the finals, because you (on average) face teams of lesser quality.

Because most the stats are averages (survivability, kills per points, etc.), it will be generally harder to maintain a high average in finals games, against tougher opponents.

Secondly, when you are using an odd number (7) of teams as your basis in the blog, it creates the statistical problem that some of the teams have faced more "top 7" teams than the others, and as such, those teams should (theoretically) have worse player rankings, because they have faced more tough opponents.

I'd say top 6 is a more valid basis, or alternatively stats from games between the top 7 teams only.

Baca Loco said...

The point here is whether or not there is anything of value in player stats when examining team results.

Nick Brockdorff said...

Yes, I know Baca, and I am saying the value of the stats will increase, if the stats can show against who the player got what stats.... or they are at least divided into prelims and finals stats.

Anonymous said...

One stat that skews results is the winfluence. If a top player is playing every point for a team that isn't winning quite as much can be pulled down when compared to a player that doesn't really play as much. IE Damian's overall rank compared to Nick Slowiak.

Anonymous said...

"One stat that skews results is the winfluence. If a top player is playing every point for a team that isn't winning quite as much can be pulled down when compared to a player that doesn't really play as much."

The would skew the results to support that winning teams have higher ranked players.