Is there any advantage for the team starting the game?

0 votes
asked May 17, 2013 by Albert62 Kubbnoob (670 points)
Starting the game is an advantage? And if so is there any statistics which could express this advantage in terms of winning percentage?

4 Answers

+1 vote
answered May 17, 2013 by Dobbie Kubblic ❚ (6,450 points)


A stab at the Hodges Brevity Theorem

0 votes
answered May 17, 2013 by garrickvanburen Kubblic ❚ (7,390 points)

Here's a chart I put together last November from the games in the wiki. It shows the number of turns per game, from this you can see the exact same number of odd-turn games as there are even-turn games. Meaning - there's no advantage to going first. 


commented May 17, 2013 by anonymous
reshown May 17, 2013 by garrickvanburen
Well... it's all a question of statistics and the level of the players! Once the players are strong and hit the baseline kubbs by a higher than 50% probability, there is for sure a correlation between opening the set and and winning it. The same goes for a best of three game. The more sets played in a game, the higher the likelihood for a break! Writing this, I think merely of the very top players in Switzerland...
commented May 17, 2013 by garrickvanburen Kubblic ❚ (7,390 points)
The games included in the stats above are from top tier, experienced, competitive players in tournament play. It's reasonable to imagine a strong team clearing all the baseline, and just as reasonable (based on game stats) for a strong opponent to answer with a 5F. Equally matched teams play equally well. If starting was a significant advantage, I think we'd see more best 2/3 going to 3 games. Instead, it's far more likely for the match to end after 2 games.
commented May 17, 2013 by Dobbie Kubblic ❚ (6,450 points)
There has to be a TON more data now that when this report was generated last year. Also, 6 person stats vs. less than would be interesting as well. Or America (generally 2-3 player teams vs. Sweden (mostly all 6 person teams). Where is the csv download link for all planetkubb data anyway?
0 votes
answered May 17, 2013 by ChrisHodges Kubblic ❚ (7,300 points)

I think the graph in Garrick’s answer tells a much larger story than the overall results being 50/50. I don’t see a Y-axis so I can’t really break down these numbers, but look at the distribution of games won by the opening team for games finished in 9 turns or less (the opening team’s 5th turn). It looks like around 2-to-1 in the first teams favor. In games that last 10 turns or more it flips the other way. In fact, if you threw first and haven’t won by your 6th turn then it seems the odds are you aren’t going to.  Bernard’s Law of Infinity states that equally matched teams will play indefinitely, but I disagree. The Hodges Brevity Theorem states that the length of a game is inversely related to the performance of the two teams: high performing teams play short games.


Playing well means you’re making progress on the baseline (or killing the king for the win). It doesn’t matter if the field is empty or you’re throwing ten, you need to advance  or end the game. If you’re failing to do that then you’re passing the control of the game to your opponent. So, the only reason that any game should last longer than the 11th turn (Team A’s 6th) is that Team A failed to advance the game in at least one turn. Couple this with the average open being about two base kubbs on turn one and really Team A should be finishing by turn 9 – right at the 2-to-1 winning ratio seen in the graph.


I realize that these are all tournament games from top-tier teams, but in my opinion all of the games that lasted more than 12 turns are examples of good teams playing badly (I don’t mean to sound haughty or like a dig against anybody  – I’m sure I am well represented in those games!) Our anonymous Swiss friend commenting on Garrick’s answer is right though; when teams are playing well there is an advantage to throwing first – and in my opinion they better they do the bigger the advantage becomes.

commented May 17, 2013 by Rekubblikanen Ironkubb ✭ (1,550 points)
I agree with Chris Hodges. Talking to Terry Ekelof, World Champion 11 times, i'm convinced that when it comes to really good teams it's always an advantage to start the game. Always. A good example is one of the games in the Swedish Championships the other year, after winning the king toss a perfect set followed, 5 kubbs and the king with the first 6 batons. Thats what I would call an advantage ;-)
commented May 17, 2013 by Dobbie Kubblic ❚ (6,450 points)
If you go first and can win in under 8 turns, then there is an advantage of going first.

Good teams playing well win in under 8 turns.
0 votes
answered May 17, 2013 by garrickvanburen Kubblic ❚ (7,390 points)
edited May 17, 2013 by garrickvanburen

I grabbed all the games in the database indicated as the 2nd game of a match. It seems to me that if the starting team has an obvious advantage, it would be most evident here. 


While I was in there, I took a look at whether or not games were won on a advantage line. 82% of the answering teams' wins came on an advantage line. Seems to say that if opening teams have a greater chance of anything, it's losing the game.

commented May 18, 2013 by Eric A. Kubblic ❚ (7,810 points)
Interesting. Very interesting. Also, it really amazed me in the US quarters, semifinals, 3rd place, and Final how few three game matches there were. I think two. Look at Rockford this year and last, how many three games, and there were some of our best teams. how about DM last year. To me the top eight were all good to really good. How many three game matches. The only thing I can think is that there is great separation, still, between a lot of teams at our top level AND, for most, but not all, it is very difficult mentally to come back from 0-1.
commented May 18, 2013 by garrickvanburen Kubblic ❚ (7,390 points)
yes, Eric, I concur. I think the data is suggesting that what appears to be an advantage in throwing first may instead be the difference in ability. Or put another way, in the words of Ekelöf, "Get Better."
commented May 18, 2013 by thingles Kubblic ❚ (6,110 points)
Since we have a mathematical representation of skill using the TrueSkill ranking someone with enough math capability should be able to normalize skill out. I store the skill ratings before the match and after to show deltas, but the pre-match scores would be a way to normalize. I also know over 2,500 matches what the likelihood of winning based on skill ranking is, but my math is far too basic to deal with this type of problem.
commented May 18, 2013 by Dobbie Kubblic ❚ (6,450 points)
The data on turn 2s above continues to support Hodges theory. Amazing!