- We don’t have enough time. We don’t have the luxury to have a two day event like NJ Open or week-long event like USGC.
- We don’t have enough players. We have limited number of players with a not-so-evenly rank distribution. This is partially a result from (1) because if this is one-day event, our players would be mostly come from local area.
Given (1), we have to choose an one day event and squeeze 4 rounds into that day. So this our constraint. Under this constraint, what can we do to address (2)?
Possible solutions and their problems
1. Run a handicap tournament.
One solution would be to mix all the players, and use the handicap system to equalize their strength, such that we can have enough players of the same (handicapped) strength to pair evenly.
This solution has two problems:
- The handicap system is non-linear.
This has two meanings:
- A could give B 2 handicaps + B could give C 2 handicaps != A could give C 4(2+2) handicaps.
- A is 6d, B is 4d, it’s likely that A can give B 2 handicaps. A is 6d, B is 1d, it’s not so likely that A can give B 5 handicaps. This accuracy of this number drops quickly when the difference goes larger. It will be very difficult for a 3d player to give 8 or 9-handicaps to a 6k player. (I say “8 or 9” because sometimes in the calculation, people may forget about these’s no “rank zero” and just use minus when calculating between d and k.)
- We may end up with multiple all-wins at the end of the day, and it might be that the lower rank player’s “sum-of-opponent-wins” will be higher than the higher rank players, so we will see a 4-win 10k being the champion, with a 4-win 4d being the second place under the condition that they haven’t even played at all. I have seen this happened many times.This is the nature of having more-than-optimal players in a single band for a less(4)-round tournament. You will see this problem in the second solution too.
Adding on top of these is the fact that all handicap games, including high handicap games, are AGA-rated. Your rating is not affected by your play, but by the implicit flaw of the system.
There are solutions:
- To use (rank difference – 2) handicap stones, such as what Pandanet City League is doing. I don’t think this solves the problem, but it mitigates in most of the conditions.
- Handicapped games should not be rated. A much larger topic that I don’t want to start.
People might have different judgement about if these “problems” are really problems. However as an organizer of the Spring Open, I tried to avoid these problems by choosing a different system.
2. Banding with equal initial scores
In this solution, we could give same initial scores to a 7d and a 5d, and pair them in a purely even manner. This would resolve situations like: If A beats B and with more winning counts, why B’s ranking is higher than A?
However this solution has two problems too:
- If the band size is large enough, say, 10 or 12 people as we have in this Spring Open, under this pairing rule, we might end up with two people in the same band, with same win-lose games, but they have not played. If these two people are both in the reward zone, it might be like: “A ends with 3-1 and got 1st, B ends with 3-1 and got 2nd, but A and B has not played”. An example of this would be each rank band in US Open (or Chang-Qi cup in last September) which exhibits the same problem. Where there are many players (like 15+) in a same band with only 6 games to play (4 in Chang-Qi cup), it’s not a rare scene to see top two players never played against each other. In a smaller tournament like the Boston Open, if you end up with band of 10-12 people with 4 rounds, this is likely to happen too.
- If we shrink the band size to 8. The above scene would be less likely to happen. However, it’s more difficult to band (i.e. had to separate within a rank), and it’s already a loss because more top players in their band will miss the opportunities to play higher level players. Also, it doesn’t guarantee that the problem in (1) won’t appear.
Issues in this Spring Open (2016)
- Dr. Yun, a very strong player from Korea, who did not know his ranking in US, entered as 5d, beating Jie Liang 7d, and gave Andy Liu 1p a hard time. He ended up (3-1) with the 3rd place, and Jie Liang (2-2) being the second place.This may sound unfair, but the initial ranking ensured that Dr. Yun to not play stronger players in the first two rounds. Unfortunately this is not like the ladder system, where you take the place of whom you beat. It’s very likely that Dr. Yun will still end up with 3-1 only losing to Andy Liu if he entered as 7d, but he would have faced more stronger players in that case.I explained this to Dr. Yun before the final ceremony. I got his understandings. This will not happen again as Dr. Yun will enter future tournaments with 6d or 7d rank.
- Jake Game(4k) and Katya Popovich(7k) are all new players, they all ended up with 4 wins but no prize position. I didn’t watch Jake’s game but I took a quick look at Katya’s game, my judgement is she is clearly much stronger than 7k. I explained to both of them about our banding rules, and suggested they enter with higher rank to get chance to win higher rank players to boost up their AGA rating faster.
There might be other issues but I think these are the most outstanding ones.
What can be done better next time
- Have a very clear rules page before the registration.
- Have an option to ask the players “do you want to band up” by entering with rank higher than the AGA rating.
- Use other more universal systems (KGS/Tygem) for new players if they don’t know what their ranks are, and make suggestions.
An ideal system would be to have N players with the same strength and play N-1 rounds, for every and each rank. If we can’t have that, we have to make compromises. The spring open is only one option. If you think there are others, feel free to talk to me.
I must limit this post to be pure technical so I’m stopping here.