Staff's Ideas for ISML 2017 Format

[maglor]

Option 1 was tried in this year. Wasn't too great.
Option 2 and 3 have many rounds that just feel pointless with the fact how many characters advanced to next round. Respectively 70% and 80%.

In this case option 4 and 5 are most interesting for me.
But certain thing about option 4 bothers me:

Round of 100: 10 groups of 10 characters, round robin; Aquamarine, Topaz and Amethyst Period. Each group 4th to 6th will be thrown into an arena where only top 10 will advance.Top 3 will advance from each group.

One arena with 30 characters? Too many... Easy to miss some characters.
Would prefer something like 5 arenas of 6 characters where top 2 from each arena advance.

Round of 40: 4 groups of 10 characters, round robin; Sapphire, Emerald and Ruby Period. The rest will be thrown into an arena where only top 4 will advance.Top 3 will advance from each group.

The same case. One arena with 30 characters... But in this case limiting to 4th to 6th places in groups would solve the problem.

If we do Single Transferable Voting for those spots, it will become hard to miss any character. This is true of any wildcard voting schemes in any of the options.

[Euphony]

Calculation of total number of 1v1 matches in the tournament for…Show

I'll add my own thoughts once I've figured them out.

Summary of total number of matches:

• 367 for Option 1 (2016 format)
• 960 for Option 2
• 1115 for Option 3
• 661 for Option 4
• 535 for Option 5
• 415 for Chibasa's Hax Option

For comparison,

• 2228 for 2015 format, Preliminary through Elimination

I'm sure the burden on the staff depends on more than just the number of matches, but from the numbers, the burden doesn't seem as great as I had initially suspected. So, never mind.

I like round-robin tournaments, because it is very interesting to see each character fight against each other. The truth is I liked the 2015 and previous editions much more than the 2016, that I found quite boring until the top 16. Also, it was very complicated to follow.

I agree about 1v1s being interesting, though I don't like that most tournaments only care about wins and losses. To me, there's a big difference between a 51%-to-49% result and a 99%-to-1% result. In determining standings, I'd like to see the margins of victory take precedence over the Win-Loss-Tie record.

2016 was quite boring to me as well. In 2015, I liked seeing the cumulative standings change from day to day. In 2016, the groups were so numerous and the size of the groups so small that the standings for each group hardly meant anything to me. It felt as though the whole 2016 season was nothing but successive seeding matches, where no leader could be observed throughout the tournament. I much prefer to see how everyone stacks up against each other (as in the 2015 statistics: http://2015.internationalsaimoe.com/statistics/" onclick="window.open(this.href);return false;) and seeing the standings change over time, like a race.

I was also disappointed with how the Necklace Periods were structured in 2016. They were more like cut-off points this time, whereas in 2015 they were more like checkpoints in the race for the Tiara. An initially weak contestant in 2015 could make a comeback, which is exciting. But in 2016, the contestant would just be cut, which is anticlimactic. So, rather than having the field of contestants get smaller and smaller as the rounds progress, I'd much prefer to see all of the contestants remain for the duration of the regular season, as in 2015.

As for the given options, I like Option 4 best, if for no other reason than those promising wildcard rounds. Also, I do think the round-robins with 10 contestants would be more interesting than ones with 4 or 6 contestants. So, Option 4 has my vote. I'm still working out my own ideas and will post them later.

I think it can be useful, so I put this here.

For wildcard round in option #4, in the final 10 from round of 100 to advance to round of 40, I am heavily leaning towards simply letting the voters choose up to 10 characters to let the top 10 character advance. I thought about using Single Transferrable Vote scheme, but that will require most of the voters to rank all 30 which probably would be tougher task than simply choosing the 10 they like. Similarly for final 4 from round of 40 to advance to round of 16, I would like the voters to pick up to 4 characters so we can simply go with top 4 in number of votes received.

I very much support this instead of STV. For 100 to 40, I'd propose allowing voters to choose up to 30 and letting the top 10 advance, and similarly for 40 to 16, allowing voters to choose up to 28 and letting the top 4 advance.

[maglor]

Calculation of total number of 1v1 matches in the tournament for…Show

I'll add my own thoughts once I've figured them out.

Summary of total number of matches:

• 367 for Option 1 (2016 format)
• 960 for Option 2
• 1115 for Option 3
• 661 for Option 4
• 535 for Option 5
• 415 for Chibasa's Hax Option

For comparison,

• 2228 for 2015 format, Preliminary through Elimination

I'm sure the burden on the staff depends on more than just the number of matches, but from the numbers, the burden doesn't seem as great as I had initially suspected. So, never mind.

I like round-robin tournaments, because it is very interesting to see each character fight against each other. The truth is I liked the 2015 and previous editions much more than the 2016, that I found quite boring until the top 16. Also, it was very complicated to follow.

I agree about 1v1s being interesting, though I don't like that most tournaments only care about wins and losses. To me, there's a big difference between a 51%-to-49% result and a 99%-to-1% result. In determining standings, I'd like to see the margins of victory take precedence over the Win-Loss-Tie record.

2016 was quite boring to me as well. In 2015, I liked seeing the cumulative standings change from day to day. In 2016, the groups were so numerous and the size of the groups so small that the standings for each group hardly meant anything to me. It felt as though the whole 2016 season was nothing but successive seeding matches, where no leader could be observed throughout the tournament. I much prefer to see how everyone stacks up against each other (as in the 2015 statistics: http://2015.internationalsaimoe.com/statistics/" onclick="window.open(this.href);return false;) and seeing the standings change over time, like a race.

I was also disappointed with how the Necklace Periods were structured in 2016. They were more like cut-off points this time, whereas in 2015 they were more like checkpoints in the race for the Tiara. An initially weak contestant in 2015 could make a comeback, which is exciting. But in 2016, the contestant would just be cut, which is anticlimactic. So, rather than having the field of contestants get smaller and smaller as the rounds progress, I'd much prefer to see all of the contestants remain for the duration of the regular season, as in 2015.

As for the given options, I like Option 4 best, if for no other reason than those promising wildcard rounds. Also, I do think the round-robins with 10 contestants would be more interesting than ones with 4 or 6 contestants. So, Option 4 has my vote. I'm still working out my own ideas and will post them later.

I think it can be useful, so I put this here.

For wildcard round in option #4, in the final 10 from round of 100 to advance to round of 40, I am heavily leaning towards simply letting the voters choose up to 10 characters to let the top 10 character advance. I thought about using Single Transferrable Vote scheme, but that will require most of the voters to rank all 30 which probably would be tougher task than simply choosing the 10 they like. Similarly for final 4 from round of 40 to advance to round of 16, I would like the voters to pick up to 4 characters so we can simply go with top 4 in number of votes received.

I very much support this instead of STV. For 100 to 40, I'd propose allowing voters to choose up to 30 and letting the top 10 advance, and similarly for 40 to 16, allowing voters to choose up to 28 and letting the top 4 advance.

While approval vote works well for more casual occasion, I don't think it is the best way to choose the best N our of X when X/N is greater than 2

[Reverend]

Is the combo of both system (even) possible?

[maglor]

Is the combo of both system (even) possible?

It can be.

[Reverend]

Is the combo of both system (even) possible?

It can be.

Oh wow that should be an interesting concept for voters and also an interesting playground for those MVs and fanbase voters to play at.
Though the amount of coding jobs required as well as its... "mathematical security" are beyond my limited knowledge.

[Shmion84]

Thanks for voting and posting comments. Staff will discuss your input.

[maglor]

Here is a question I am desperate for comments : What are the steps you think is needed to give sufficient challenge to, and then find and award the necklace to the deserving candidate? Be specific about how many and what king of matches you think we should throw at each contestant in order to choose the most deserving characters to be placed into the necklace group arena.

[superunature]

This year's format is far too complicated with too many seemingly random awards.
If you know what's going on and been following all year, I'm sure you know what's going on.
but as occasional voter, I honestly have no idea of the tournament format to the very end.

especially exhibition matches.
There needs to be a place under statistic that displays results of all the exhibition matches separate from the periods

[Euphony]

Here is a question I am desperate for comments : What are the steps you think is needed to give sufficient challenge to, and then find and award the necklace to the deserving candidate? Be specific about how many and what king of matches you think we should throw at each contestant in order to choose the most deserving characters to be placed into the necklace group arena.

The contestants with the highest VF in the period deserve to be placed in the necklace group arena.
Not VD because anti-votes should be irrelevant.
Not Wins because a contestant with 99%-to-1% wins should be more deserving than a contestant with 51%-49% wins.

[Chocola]

The main problem with VF is that match counts fluctuate. Imagine someone winning 60% with 2000 votes, and then 60% with 2500. It's the same ratio, but one counts for more VF. If anything, we should be using total VF%.