Why do so many winning teams have a lower hit percentage than their opponents.

+1 vote
asked Jul 15, 2012 by garrickvanburen Kubblic ❚ (7,350 points)
edited Jul 16, 2012 by garrickvanburen

 

I just finished uploading all the championship game stats (save the 3rd place game) to the Games section of the Planet Kubb wiki - huge thanks to @chad-b and @thansenite for their help scoring the games. 
 
The most immediate question I see from the stats thus far is:
 
How can a team win with a lower hit ratio than their opponent?
 
Half a dozen examples from the US Nationals Championship bracket:
 

 

2 Answers

0 votes
answered Jul 15, 2012 by ChrisHodges Kubblic ❚ (7,300 points)
selected Jul 16, 2012 by garrickvanburen
 
Best answer

Hey Garrick!

 

This question really piqued my interest and my first thought was "It HAS to be the kubbs per baton ratio!", and I immediately went about researching the games you listed to find out if some teams were basically getting more done with fewer batons.

Unfortunately, the answer seems to be much simpler - Bad Data.

In each of the games you list the names of Team A and Team B are misrepresented in the game info box at the top - take a look at the player initials in the round scores and you can see that they were switched. Before I realized this I did a more in-depth analysis of the first game (game 2 of the finals) and discovered the following as well:

First, there appears to be an innaccuracy in the game record, as it shows the Kubbsicles open with only one base kubb, but the Knockerheads throw two kubbs in turn two, only topple two fields, and in turn three the Kubbsicles again throw 2 fields. My memory is junk, but I think this has to mean that the Kubbsicles opened with 2 base kubbs.

Secondly, there seems to be an error in the calculation of hit percentage, as the game stats show that the Kubbsicles threw 42 batons, but Mark slew the King with baton #5 of round 7 meaning only 41 batons were thrown. This updates the stats to:

Kubbsicles: 41 batons, 12 misses (70.7% hit rate), 40 kubbs toppled (0.976 kubbs/baton)
Knockerheads: 36 batons, 14 misses (61.6% hit rate), 32 kubbs toppled (0.889 kubbs/baton)

In this instance, the winning team had a higher hit rate AND KpB.
 

...and then I noticed that the names were switched and did a quick check of the other listed games to find the same error.

 

So, let's see about enacting the following:

  1. Correcting the hit percentage to disgregard batons not thrown during the final turn
  2. Adding a "Kubbs-per-Baton" stat (I just think this would be interesting is all...)
  3. Adding some error checking to the form (without a "rescue" a team can't throw more field kubbs than their opponent had hits in the prior round)

 

...what can I do to help?

 

 

 

commented Jul 16, 2012 by thingles Kubblic ❚ (6,110 points)
edited Jul 16, 2012 by thingles
Good bug catches Chris. And for everyone, remember this is very fresh code and I know there are a number of issues. This morning I refactored the entire loop for incrementing counts. Here is the refactor.

http://wiki.planetkubb.com/w/index.php?title=Template%3AGame_turn_score&diff=1679&oldid=1341

Before the loop wasn't honoring X as unthrown. I totally redid it using a big switch so it makes more sense. I'll tackle #2  above soon.

I fixed the team A/B thing as well and now this game

http://wiki.planetkubb.com/wiki/US_Nationals_2012,_Quarterfinals,_Kubb%27d_v._Kubbsicles,_July_15_(Game_2)

Shows Kubbsicles (winner) with 48% hit rate and Kubb'd with 28%.
commented Jul 16, 2012 by ChrisHodges Kubblic ❚ (7,300 points)
I know it's still new, and I'm sorry if I came off as harsh or judgmental. I want this data tracking system to be as good as possible, and while I'm not sure how many of the problems I can actually fix I can t least help FIND them - I'm a reporting analyst, so data analysis is kinda my thing...
commented Jul 16, 2012 by thingles Kubblic ❚ (6,110 points)
Not harsh at all Chris, no worries here. Just highlighting that before we think too hard about the numbers lets make sure the numbers are right. Just like you did! :-)

The team names were flopped on another game as well. We should probably validate those are all correct and it highlights that the form on the wiki needs some better UI so that doesn't happen.

Please share any and all other data issues that you see. The brains of the scoring all happen in one template in the Wiki called "Game turn score". If you look at the code it's not pretty but it is fairly readable. Use "View source" link to see the actual code, the template renders it's (currently undocumented) documentation on the initial view. The code is in the source.

http://wiki.planetkubb.com/w/index.php?title=Template:Game_turn_score
commented Jul 16, 2012 by THansenite Ironkubb ✭ (2,660 points)
The rescue kubb is a great catch Chris.  The addition of a rescue kubb could throw my KITE system off.  Not sure what it does to this system.

Jamie, I really like the addition of [inkast], [rethrow], [penalty].  One question I have is if 7 kubbs are thrown, then three are rethrown and one of those three knocks a previously legal kubb out of bounds forcing that one to be rethrown.  Would that be marked as "7, 4" or "7, 3+1" or some other nomenclature?

Another suggestion might be a glossary of what all the stats are.  Some are fairly straight-forward, but others are a little confusing about what they are exactly.  A breakdown explaining them may help.

I'll keep pondering the system and can't wait to get some games from around here in the system.  It will be great once we get a lot of data coming in since an individual team or player's stats can vary wildly game to game.  

It would be fun to eventually have a leaderboard where once a player has been entered in the system a minimum number of times (say 10 games minimum) to see who has the best Baton to Kubb ratio, or the least (or most) number of rethrows, or the most penalty kubbs just for fun to see how people stack up.
commented Jul 16, 2012 by garrickvanburen Kubblic ❚ (7,350 points)
Chris thanks for checking the data. The A/B stuff has been fixed across all the offenders.
commented Jul 16, 2012 by thingles Kubblic ❚ (6,110 points)
Regarding a rethrown Kubb that knocks an already thrown Kubb out and requires another rethrow - I think simple is key. Using your example I would put "7,4". Now, in reality on the sheet I've probably already written "7,3" if I'm scoring a game. So I would scratch out 3 and put 4 above. But in the wiki the field should just have a number, so "4".
commented Jul 16, 2012 by thingles Kubblic ❚ (6,110 points)
I agree on the glossary as well. I sort of expect to create this new, very readable and understandable set of derived statistics. However, I fully plain to dump all the raw counters at the bottom for whatever people want to reference them as. I'll likely just create a wiki page that will have all of those on it.
commented Jul 16, 2012 by thingles Kubblic ❚ (6,110 points)
I just stubbed out a page to define the raw stat counters on.

http://wiki.planetkubb.com/wiki/Statistics

Nothing filled in but the game stats now link to that page as well.
commented Jul 16, 2012 by thingles Kubblic ❚ (6,110 points)
I just modified the form on the wiki to allow for the player throwing the drill and the rethrow and penalty kubbs. You can see it in operation on these two games:

http://wiki.planetkubb.com/wiki/US_Nationals_2012,_Semifinals,_Leinenkubbels_v._Kubbsicles,_July_15_(Game_1)

http://wiki.planetkubb.com/wiki/US_Nationals_2012,_Finals,_Knockerheads_v._Kubbsicles,_July_15_(Game_1)

I think it works pretty good. Plus it's awesome to be able to know how many penalty kubbs drillers get! :-)
0 votes
answered Jul 15, 2012 by C=mxcXKubbSection Ironkubb ✭ (1,330 points)
Hi Garrick, I think because it's about a game, not about stats ;) Meaning... You don't score goals or points. There's a dimension about the relations 'between' the kubbs that is important, and stats can seldom capture relations. My opponents can hit everything, all my baseline kubbs only can be the ones in play for a couple of rounds, and suddenly their last toss misses and gives me the crucial penalty line to kill the game in the next round. I do envy your stat effort, and I will come back with some thoughts about it soon.
commented Jul 15, 2012 by garrickvanburen Kubblic ❚ (7,350 points)
Yes, I understand. What's interesting is that based on these handful of games, it could be argued that teams that miss significantly more win more. I suspect that's a sign of something else. Additionally, only 2 of the games I listed end with an advantage line. One game only lasts 18 batons.
...