News Shop
Events Forums

[Tournament] Lum's Long Odds - Ongoing!


#362

The main idea in this format is to complete the historical matchup chart by making us play the balanced matchups.
Usually, the unbalanced matchups tend to be by far the most played (because the best counterpick against a character is the most unbalanced matchup against him).

This format just tells the opposite : you are incited to play the matchup where the historical matchup chart is the wrongest in your opinion, or simply a matchup you think fun.

Adding an additionnal ammount of points just does the opposite : it incitates to play a match-up where you have an advantage on your opponent.

I’m really happy we contributed the historical match-up chart to be a bit more accurate. We also now have a statistic about which character do people prefer (independently to their actual performances)


#363

I realise that it does add strategy. I probably shouldn’t have said that it didn’t add anything. However, the main idea of the tournament was to make interesting choices based on higher risk leading to higher reward. However, if the set is tied 2 - 2 the bonus points allow you to take less risk for an equivalent reward, and I don’t think that particular aspect gels with the ‘long odds’ style of gameplay.

But, it isn’t a huge change and it’s not like it’s going to ruin the format or anything.


#364

Okay, this here is my final analysis of the tournament. Not only am I including the W/L in comparison to points gained and lost, I will also be noting significant matchups and matchup discrepancies/changes from the chart that the whole tournament began with. Buckle up!

Character Analysis

Points won/lost

:chibiargagarg: 42/37.5
Record: 10-7
Rating: 1.12
Argagarg was not used all that much this tournament. He ended up with a positive win record and a positive point spread. His most significant contribution was going 3-1 over Menelker, which reflects the matchup chart’s 6-4 number.

:chibibbb: 66/36
Record: 13-6
Rating: 1.83
BBB was used a little bit more than Argagarg, but not by much, though his showing was far more impressive. Ending with over double wins vs. losses and almost double points, he did work. His most significant contribution was going 3-0 over Persephone. While BBB does indeed have the advantage in this matchup as indicated by the historical chart (and most conventional MU charts as well), 3-0 is still pretty good.

:chibidegrey: 94/131.5
Record: 22-24
Rating: .71
Degrey ended up with only a slightly losing record, but a terrible points deficit. I am slightly to blame for this for forfeiting a 7 point matchup, but even without that he would still be in a bad position. He went 6-9 vs. Menelker, which on the chart was listed as 6-4 in his favor. That right there accounts for 24/54 score. In fact it was enough to shift the total matchup chart to (rounded off) 5.5-4.5 his favor. On a good note for him, he went 4-2 vs. Midori, which is listed as a 5-5 matchup.

:chibigeiger: 10.5/18.5
Record: 2-3
Rating: .57
Used the least out of any character, there is little appreciable data for analysis on the man of science. The reason being, of course, that he only went 2-3, but is at an 8 points deficit because he lost to Jaina, the most lopsided matchup in the historical books. Everyone was scared of this exact thing, so he didn’t see much action.

:chibigloria: 84/74.5
Record: 18-17
Rating: 1.13
Gloria ends up slightly up in both wins and points. This doesn’t really make any strong statements about her either way. She went 6-3 over Menelker which is listed as a 5-5 and went 3-0 vs. Grave and Setsuki, so that’s where she got her points. On the other hand, she went 1-5 vs. Midori (listed as 3-7, so predictable).

:chibigrave: 192.5/164.5
Record: 37-32
Rating: 1.17
Grave finished the tournament with a decent win accrual and respectable points surplus. Most of his matchups actually followed the chart precisely (7-8 vs. Menelker, 3-4 vs. Persephone) with the sole notable exception being going 3-1 over DeGrey, which is listed as 4.5-5.5 in DeGrey’s favor. He also went even with Quince which is listed as 4-6 Grave disadvantage, but they only played 6 games, so it’s not a huge statement.

:chibigwen: 116.5/131
Record: 25-26
Rating: .89
Gwen finishes off the tournament on a low note. She started off down, went up, but ended on a decline. Notably, she went 3-4 vs. Persephone, which was listed as 6-4 in her favor. She also went 3-4 vs. Quince which has the same score. In her favor, she went 4-0 vs. Valerie, a 5-5 matchup. She went 0-4 vs. Lum, but that’s already listed as a terrible matchup (3.5 for her) so it wasn’t a huge source of points loss.

:chibijaina: 121/122.5
Record: 22-27
Rating: .99
In the end, Jaina - the poster child for this tournament - ended up going even in points, but with a 5 game deficit. This not only shows that she is low tier, but that the numbers are pretty accurate for her. In general, all of her notable matchups went the way the charts say. Stay frosty, Jaina.

:chibilum: 90/83
Record: 20-16
Rating: 1.08
Lum ends up with a sizeable lead on wins but pretty close to even on points spread. His major points loss came from going 1-5 vs. Vendetta, which is a match listed in his favor at 4.5-5.5. He went 5-2 over Menelker which is slightly Menelker favor historically though.

:chibimenelker: 282.5/244
Record: 48-45
Rating: 1.16
Menelker ends up being the most popular character in the tournament! It’s not hard to see why, with his slight w/l advantage showing up as a good-sized point advantage. He also had by far the most mirror matches with 5. His biggest point gainer was already mentioned up in DeGrey’s writeup, but he also went 7-2 over Valerie, a 5-5 on the historical chart.

:chibimidori: 118.5/100
Record: 24-17
Rating: 1.19
Midori ends up with a great w/l record but only a mediocre points advantage. The result of a lot of wins coming from his best matchup, Gloria. He did manage to go 3-0 vs. Setsuki though, a losing matchup (historically).

:chibionimaru: 70/84.5
Record: 15-18
Rating: .83
Ends up with a point deficit. Most of his matchups went by the numbers, he just happened to lose a bit more than he won.

:chibipersephone: 163/202
Record: 30-42
Rating: .81
Persephone ends up with the worst w/l record of the cast. Fitting for a character who frequently comes up in discussions for least powerful character. Funnily enough, her rating is not the worst though, because her matchup chart is generally unkind to her in the first place. She ended up doing well against Gwen, but tanking against Quince (4-8). Also hurting her was her 2-4 record vs. Rook, a 5.5 advantaged match for her.

:chibiquince: 169/150.5
Record: 33-29
Rating: 1.12
Quince ends up in the positive, with a decent w/l surplus. His 3-7 performance vs. Setsuki did him no favors though, being a 5.5 advantaged match for him.

:chibirook: 63/100
Record: 10-19
Rating: .63
Rook ends up with almost double losses to his wins and a pretty bad points deficit. Persephone had a higher loss spread, but Rook’s w/l ratio is worse by far. He didn’t win a single matchup that more than one game was played in.

:chibisetsuki: 110/141.5
Record: 19-24
Rating: .78
Setsuki began this tournament as the worst character for many weeks, only being slightly vindicated the last few rounds. She still ends up with a pretty good loss margin and a large point deficit though. She went 0-3 vs. Zane AND Gloria, but as mentioned earlier did pretty well for herself in the Quince matchup. In fact, well enough that if we ran the tournament again right now it would be listed as 5-5.

:chibitroq: 20/19.5
Record: 5-4
Rating: 1.03
Not used much in fear of his excellent matchup numbers, but still ends up even.

:chibivalerie: 54/99
Record: 11-21
Rating: .55
Valerie ends up with the lowest rating of the entire tournament. Between going 0-4 vs. Gwen and 2-7 vs. Menelker, she really dug herself a hole. the only matchup she “won” was going 1-0 vs. BBB.

:chibivendetta: 112/55.5
Record: 20-11
Rating: 2.02
And in a complete turnaround, Vendetta comes in with the best w/l record and best point spread in the cast. I personally went 14-6 with him, meaning the remainder of the player base went 6-5 so I’m going to go ahead and take full credit here. In that vein, I’m not sure what these numbers say about Vendetta at all.

:chibizane: 38.5/37
Record: 11-7
Rating: 1.04
Zane ends up with a 4 win surplus but nearly even points. The converse of what my analysis was for Jaina, this means that he has great numbers but deserves them (though the data is somewhat more scant for Zane in this tournament).

Ranking of characters by points ratio:

Vendetta 2.02
BBB 1.83
Midori 1.19
Grave 1.17
Menelker 1.16
Gloria 1.13
Quince 1.12
Argagarg 1.12
Lum 1.08
Zane 1.04
Troq 1.03
Jaina .99
Gwen .89
Onimaru .83
Persephone .81
Setsuki .78
DeGrey .71
Rook .63
Geiger .57
Valerie .55

Ranking of characters by absolute points difference

Vendetta +56.5
Menelker +38.5
BBB +30
Grave +28
Quince +18.5
Midori +18.5
Gloria +9.5
Lum +7
Argagarg +4.5
Zane +1.5
Troq +.5
Jaina -1.5
Geiger -8
Gwen -14.5
Onimaru -14.5
Setsuki -31.5
Rook -37
Degrey -37.5
Persephone -39
Valerie -45

Ranking of characters by Historical Matchup Chart Accuracy (in comparison to results of this tournament)

Jaina 99%
Troq 97.1%
Zane 96.2%
Lum 92.6%
Argagarg 89.3%
Quince 89.3%
Gwen 89%
Gloria 88.5%
Menelker 86,2%
Grave 85.5%
Midori 84%
Onimaru 83%
Persephone 81%
Setsuki 78%
DeGrey 71%
Rook 63%
Geiger 57%
Valerie 55%
BBB 54.6%
Vendetta 49.5%


#365

Praise be to the increased accuracy of the MU chart.

I’d say it means you play a really mean Vendetta, and that the ceiling on the character is much higher than his average results may imply.


#366

So, in some of your analysis, such as with Jaina and Zane, you note that their closeness to an even ratio and predictable win/loss ratio signify that the matchup chart is accurate. Can we get a statistic for this? Perhaps even just a ranking of absolute distance from even ratio, smallest first. Whatever metric Jaina is number one at.


#367

Yes, yes I can. I will update the analysis post with the data.

EDIT: Actually, I changed my mind. That can be easily extrapolated from looking at the “rating” category I have and just going from the middle toward the edges. I tried to make a correlation based on number of historical games charted vs. relative “accuracy” found in the tournament, but there was no pattern to be drawn and would just be viewed through my subjective analysis. So unless you’re asking for something that I’m not understanding, I don’t think it is info that can’t be already seen from the rating chart and just taking the closest to 1 being the “most accurate”.


#368

That is all I’m asking for. Just make a list of 1-20 of that for convenient viewing.

I mean, I could do it myself, but then it wouldn’t be in your post.


#369

Okay, done.


#370

My suggestion for the next time you run this is to not round off for the matchup numbers and just use the straight values from the MU chart. Yes, the chart will appear a bit “messier”, but it will be more “accurate”, and also go a loooooooooong way to preventing any ties.

EDIT: I came to this conclusion after data mining to create a list of matchup numbers that are now different from what you have on the original chart. There were a LOT, but almost all of them were between .7-.8 or .2-.3, right at the roundoff margins. So a very tiny change actually would change the point handouts quite significantly.

And yes, I meant just use the truncated numbers supplied by the MU data, which all round to .1 by default.


#371

Probably even just changing the rounding from 0.5 to 0.1 would be sufficient (and wouldn’t make the numbers awful to work with).


#372