SuperLadle Discussion - cont. from time change thread

[PokeMaster]

Maybe a way to get around it is to add Rounds Bye'd * (Round Wins / Rounds Total) (aka adding the number of wins a team would have had based on their round win percentage). Obviously it won't be exact, since matches are far from predictable, but it does seem like the most likely representation of how teams would perform if they had played matches in those bye slots.

[sinewav]

Maybe a way to get around it is to add Rounds Bye'd * (Round Wins / Rounds Total)

I would like to get away from anything that requires me to collect more data. It's hard enough to do this by hand, and even the helper script I wrote is not easily modified. Also, let's not make the formula for deciding seeds look like a recipe for super-symmetry. That's just ridiculous. We are dealing in minutia right now. There is plenty of data available to make a good evaluation of how teams performed. Besides, anyone who makes the Bowl cutoff is in for a wicked-hard tournament regardless of how they are seeded.

I think adding weight to the Ladle wins is probably the simplest and most efficient thing to do. Or we can lower the weight of the round wins.

[Concord]

I had a think on this. Here's my conclusion (for the moment, anyway).

When the bowl match ups are set, the team with more byes gets to choose from among the available servers. The team with the most byes gets to pick first, and that one is used for semis and finals. Second most picks second, and that's used for other semi.

doesn't effect seeding or the numerical score, but creates some homefield advantage

[Soul]

I had a think on this. Here's my conclusion (for the moment, anyway).

When the bowl match ups are set, the team with more byes gets to choose from among the available servers. The team with the most byes gets to pick first, and that one is used for semis and finals. Second most picks second, and that's used for other semi.

doesn't effect seeding or the numerical score, but creates some homefield advantage

This.

[sinewav]

I'm really hesitant to include byes in the calculation at all. Byes are a random element since they arise from how many teams sign up in any given month. There is nothing a team can do to influence it. We might as well flip a coin. It would be better to revise the formula we use, possibly using a ratio of round_wins / total_rounds? Why don't we just say:

Code: Select all

Ladle_Wins + (Round_Wins / Total_Rounds)

Surely that's enough to sort the teams effectively. How about this: why don't we practice using the stats from the table I've labeled "2012 Pre-Season" to test some formulas out? There is 7 months worth of data and we can see if the 8 teams that meet Bowl requirements feel right.

Keep this in mind: The top teams are getting in no matter what. Byes are not an issue for them. Where they get seeded will have little consequence. It is the teams in danger of getting cut that have the most to loose from these changes.

[Concord]

I think that's way way too heavily weighted towards Ladle wins. And since the second half of the sum can't be greater than one, teams would be tiered by ladle wins, with no way to assess consistency across the entire season.

what we have now is fine, but this might be a bit better

Code: Select all

(rounds reached + ladle wins) * (match wins / matches played)

the new thing is rounds reached. the finals are the 4th round, semis 3rd, etc. If you reach the finals, regardless of how you got there or how you perform there, you get 4 points. this is good because a team it should matter which rounds you win. Imagine Team Alice has a bye, and wins the quarters and losses in the semis. Compare this to Team Bob, who Alice beat in the semis. Both have 1 round won and 1 loss. Ignoring match win percentage, these teams would have identical scores. One team lost in the semis, the other in the quarters. In fact, Alice beat Bob and yet they have the same score. The later in a ladle the match is, the harder it is to win. Either we could weight rounds increasing, or, simpler, just count what round the team gets to.

[owned]

I'd like to add in my two cents. I think using (rounds won/total rounds) makes more sense than (matches won/total rounds).

Here are some examples why:

1.Team A wins its first 2 rounds 2-1 and loses 2-0 in the semis. (matches won/total matches)=.5
Team B wins its first round 2-0 and loses its second 2-0. (matches won/total matches)=.5
A=B

2. Team A wins its first 3 rounds 2-1 and loses 2-0 in the finals. (matches won/total matches)=.55
Team B wins its first 2 rounds 2-0 and loses 2-0 in the semis. (match won/total matches)=.67
B>A

I'm of the opinion that a team winning more rounds should always get more points, and I think most people here would agree with me. Of course winning 2-0 is better than winning 2-1, so perhaps (match wins)/(total matches) could be used as a tiebreaker if two teams are tied.

[sinewav]

Ok you guys can go ahead and drive yourselves crazy with over-complicated ideas for what is really a simple thing. I personally don't care how heavy Ladle wins are because really, if a team win 4 Ladles in a season and the next contender wins 2, well damn, that first team is definitely better.

Also, the more you worry about this, the more work it is for you because I'm not collecting any more data points other that what we currently have. Make it work with the current setup or take over responsibility. Your choice.

I don't see anything wrong with:

Code: Select all

( Round wins + ( Ladle wins * 2 )) * ( Match win / Match total )

That compensates for random byes and doesn't involve adding arbitrary points for each final, which is ridiculous and complete overkill.

Did any of you even think to try out the Pre-Season data or are you just mathtrubating?

[Concord]

yeah on the second page of the thread…


I like the original best actually

Code: Select all

( round wins + finals match wins ) * match winning percentage

note that this is identical to your

Code: Select all

( Round wins + ( Ladle wins * 2 )) * ( Match win / Match total )

except that it credits second place teams that force a 3rd match in the final.

[sinewav]

Code: Select all

( round wins + finals match wins ) * match winning percentage

...it credits second place teams that force a 3rd match in the final.

Hmmm... That is really cool, yes. But maybe we can compromise and use your modifier idea instead? Again, I don't think any of these formulas will drastically affect the top teams, but will be more important for the ones near the cutoff.

Code: Select all

( Round wins + ( Ladle wins * 2 )) * ( Match win / Match total )

This ^^

I have a feeling this debate will erupt again in March, haha.

[PokeMaster]

I haven't yet read through all the different suggestions above, but we should just pick one and stick to it. I think it's already a bit improper (for lack of better word) to be trying to perfect the system once we've already started a season. The best thing to do would be to play with what we have, and then make adjustments at the end of this season, and before the next. There's no way to guarantee perfection when we haven't even seen how it'll play out, so let's just see what happens with our already seemingly pretty-darn-good system.

But if you guys are just hashing out ideas for next season, das coo.

[sinewav]

I ran the preseason data through the formula with the Wins*2 modifier:

Code: Select all

...Table 1...
Team            Ladle Wins    Win*2
Revolver             2        9.63
Crazy Tronners A     1        8.59
Speeders             1        7.88
TeamBaylife*         1        7.00
MeetYourMaker 1      1        6.43
Rogue Tronners                6.15
Team Unknown                  5.11
Twi¦×¦ted ¦×¦ats              4.54
-------------------cutoff---------
Speedhax             1        3.92

...Table 2...
Team               FMT     Score
Crazy Tronners A    8      12.27
Speeders            5       9.84
Revolver            4       9.63
MeetYourMaker 1     6       9.29
Rogue Tronners      3       8.00
TeamBaylife         2       7.00
Team Unknown        3       6.81
Speedhax            3       4.58
-------------------cutoff--------
Twi¦×¦ted ¦×¦ats            4.54

The first table with the Win*2 modifier shows Ladle wins do not overshadow performance. Team Baylife played one perfect Ladle and wound up on the top of mYm's 3 shaky visits, Speedhax's single Ladle win didn't count for enough to make the cutoff. When I look at this readout, I think it feels about right.

The second table shows "finals match total" (FMT) and the distribution is really sketchy. I don't think it feels very good at all. FMT credits 2nd place team -- and unfairly boosts the winner too. I'm voting against it.

This is 7 months worth of data. I imagine a 9 month season will look even better with the Win*2 method.

*Note: The current rules disallow a team from entering the Bowl with a single Ladle visit, so Baylife would not play and Speedhax would. Baylife is there for illustrative purposes only.

[sinewav]

I think it's already a bit improper (for lack of better word) to be trying to perfect the system once we've already started a season.

Concord's point about byes is an important one though. I think the Win*2 modifier, which was part of the original discussion from last April, addresses the problem well. Adding it now doesn't change the season's outcome much since overall performance is still what gets you into the Bowl.

Just keep rockin' it little bro.

[PokeMaster]

Yup, I don't disagree that it's important. But you wouldn't make a change to the ladle scoring system, even a tiny change for the better, halfway through the opening rounds. You'd just wait until the end of the ladle and fix it for the next one.

[owned]

Ok you guys can go ahead and drive yourselves crazy with over-complicated ideas for what is really a simple thing.

I fail to see how any of what I said was "overcomplicated." The statement is basic: I think using (rounds won/total rounds) makes more sense than (matches won/total rounds). I could've not used any math to keep it "simple" but then I would've just been making a baseless comment with no evidence.