Orbs of Discord – Will OWL Deliver the Best Playoffs it Can?

June 14th
written by Eden
Twitter

episode 2 playoff seeding

Playoff Qualification and Seeding Rules

Yiska: Do you think that bye should exist as a price for getting first in your division?

Eden: We’ve covered that I think it’s an abomination
I will feel somewhat better about it if they make it a best of 3 of best of fives like the final for the semis
But right now, the chance of a team making the finals off a single match is so easily broken that it is inexcusable. I like the division level admittance to playoffs, because you’ll want representatives from every region being involved especially as the league expands.
This brings up a topic though which we might differ on. What do you think about the equal weighting of matches across the season to playoff seeding?

Yiska: I think if we had had one patch with only minor balance adjustments that wouldn’t impact the meta fundamentally, then having each regular season match mean equally as much is a fine concept. The problem is that an unchanging game doesn’t seem to be desirable for players and viewers alike and so the system needs to be adjusted accordingly. To me it’s extremely tedious that we won’t see Shock or Fuel in the regular season playoffs, while with a slightly less garbage performance this stage, Seoul Dynasty could’ve made it in. This leads to bad playoff quality and nobody wants that.

Eden: So I agree with the sentiment of good playoff games. I do not think that you can design a rule based system off match history to effectively detect what team is the best going into playoffs in fundamentals or especially for the patch it’d be played on. I think that the good thing about using the entire season equally is that it gives sample size to what matters which is your average performance across metas, which is a better representation than your performance on a single patch over like effectively 20 games worth of sample size for instance
If you could somehow filter out the meta aspect, then I’d be more for it from good playoff angle, but I’m not convinced that team most recently strong is better predictor of next period performance over average
There are other reasons why equal weighting is nice
encourages good management decisions, increases stakes early
The area of major improvement we should talk about is stage playoffs effecting playoff qualification

Yiska: I agree that finding a system that perfectly predicts the teams which would do best in the playoffs is not effectively doable, but the weighting system would do a much better job than no weighting. Metas have been the single most determining factor of a teams strength bar none my books. We had teams with the blackest of swan scenarios who did very well in the stage, we had roster changes and coaching staff changes and yet nothing hammered or made a team as hard as meta switches did. That should be considered in the system. I’m fine if they answer is to patch less frequently. If NYXL had played the way they did all the way up to stage 4 and then dropped off to Shanghai levels because of a drastically new meta, I’d still not want them in playoffs and I’d still say that Blizzard’s patch philosophy was problematic. The sample size difference does nothing for me. It’s all about short bursts of performance, not an endurance race. The qualities asked for are very different and yes, stage playoffs are the best simulation of that. I think they should actually adopt the best of 3 of best of 5 system there and take more time to get these stage finals done in, so we have a deeper appreciation for the winner and it runs a better simulation for the eventual season finals. The winner of the stage should receive weighted points towards season playoff participation in my book, like Montecristo has suggested before.

Eden: I am correct in assuming that playoffs are on a different patch right?

Yiska: Most likely, but they haven’t announced that yet.

Eden: I’m assuming they will… so the problem with taking recent performance is that you don’t filter out actual randomness from specific results. You can win one random game and likely would be enough to get into playoffs if you were a midtable team for instance. it’s very off. now the one that I definitely am behind is to somehow bring up to current level teams when there is a major change (although in some way is pretty unfair to teams that had their shit together the entire time)
so for instance if Gladiators was on the very edge and slightly below, if I were god of the league, I’d choose to bring them in because they were actually a different team stage 1
but the rulebook has no basis to know that. and if you look at teams bump up and down from stage to stage and month to month in stage standings, I think that’s evidence that taking recent results to apply to playoffs would result in alot of really shitty teams. The teams with consistent performance at their lows can do something at least and have potential to pop up to a high and be a contender
maybe they aren’t now in this meta, but in playoffs they’ll shine

Yiska: The last stage meta is most likely to be the most like the playoff meta and therefore a win in the last stage showcases competence more than any other stage. Moreover because teams switch around during the midseason, you are assigning too much value to a team which the organization itself doesn’t think it should continue to go on in that incarnation. Stage 1 should be carried of the excitement of having a new Overwatch League season pop up and it doesn’t need much more than that. Which evidence are you talking about when you say that a lot of shitty teams would make it to playoffs under a weighted system? Which team that is currently doing well would be a shitty playoff contender?

Eden: So whether the meta from stage to stage is closer is up for debate. I assume that you are referring to this specific instance, because from patch to patch it’s been very bumpy with no major sense of continuity of what teams are looking really hot from the patch. So I think weighting only makes since if you build in the concept that it is a literally different team at the end than the beginning. From patching, I disagree though that somehow the final estimate is a better than the first one of how a team would do in the next (word salad). I don’t know which ones they would be. Valiant, Dallas and Shock are as a group probably doing better this stage than they would if we reran a patch simulation 1000 times.
on average
I think that maybe you can say that one or more of them has seen a fundamental change unrelated to patching, but it’s hard for a rule system to detect the difference between the two
and I don’t think you should gamble that maybe this change is fundamental to then discount the excellence of teams in a meta perhaps as representative of a teams ability to play well under a new patch as these results.
I think you have to assume it was the same team playing the whole time, with just patch variability

Yiska: But it virtually isn’t for any of the Overwatch League teams. The only team which didn’t change in essential conceptualization over the season was NYXL. All the others have undergone fundamental changes and the ship of theseus is flying high mast. It’s much less likely to see a team bring back what was good about them in stage 1, because for all we know they traded the reason they were good off and the window has closed or they had their player banned or removed and he isn’t coming back.

Eden: yeah so that’s an interesting thing and that’s what I’m somewhat sympathetic to. if a teams base without the patch wind at your back or in your face has changed, then playoffs in an idealized world should recognize that. But I don’t know at what point that should be a systemic influencer. If a single team had a major change in form from a significant shakeup? If two? I mean if you know for sure that an entire new OWL started with all new players at stage 3, then playoffs should only consider 2 stages of results lol

Yiska: And for some teams that might as well have been the case, right? Think of the shock with Architect, Sinatraa and Super as well as Crusty. Think of LAG. I have to concede that the negative example would be that London wouldn’t be able to play in those weighted playoffs either and they look to be the only kingslayers around.
It remains to be seen if through better coaching structure and more delicate handling of balance patches in the league as well as more forgiving schedules, we will see more consistency across the league.

Eden: So I think that this season is the best case scenario for discounting early results with such huge turbulance. I the trend will be towards more stable results as players become more known quantities and organizations settle on approaches a bit. If we did adjust for this factor to give some leeway early on to figure your shit out, then it shouldn’t balance to this season in calibration
But if we move to a world where 3/4 of teams are pretty stable through the season other than patch variability, then straight average or close is probably optimal. otherwise you might discount a teams high early in the season which could have happened later just as easily and is just as relevant to playoff potential

Yiska: Yeah, it’s a possibility and it’s unfortunate that the leagues environment won’t stabilize for several more season, so finding the perfect formula will take a long time. Next season we have more teams around and I imagine that will do something to the schedule. The season after we start travelling. The formula keeps changing and we have to align the system around it by induction.

Eden: I do think we should take high points of performance as extra weight for playoff qualification. If you think about what it will take to win the playoffs, the team will be one that hits a high form, not that a mediocre team just randomly wins 3 best of 5s in a row. It can happen but that’s a pretty sad world anyway. I think that if a team demonstrates a decent average AND has demonstrated pushing up to highs at some point that is a much better resume. That’s my vision where stage playoff performance factors in much more. I want the commentary around every team to be “will they reach the form to win this event” not well they have been pretty decent this season. That’s why London is second most likely to win if they get to the finals. We know they have those highs. Fusion as well
That also helps with your recency issue to some extent because it gives a more viable path which also doesn’t hurt teams that didn’t have early failings
Oh another change in a similar spirit, I think weighting stages equally is fine but would be more bearable with a more peak focused scoring metric per stage. I’d have points for real achievements and not incremental by match. So maybe a few points for not being SHD but then have points only really emerge for a few teams that carry through. That also would have the impact to punish temporary slumps as hard
A team goes 0-10 then 10-0 that team is more interesting in playoffs than a team that goes 6-4 twice for me
And more likely to win the bracket

Yiska: Yeah, that could be a fine solution as well.

Stage Playoff Prediction

Yiska: Yeah, if the idea of recency bias was not the be used, the second best alternative would easily be stage playoff wins as that signifies peaks and the ability to peak. Speaking of which, who will peak in this stages finals? Is the flood over and we can get rid of the sandbags?

Eden: I think gladiators is the dark horse of this stage playoffs for sure where they could easily make the finals and on a really good day and NYXL not looking top form could take it.

Yiska: I still think NYXL can whip it out and will just cash in that 125k with relative ease and unless there is a good reason to hold back in the finals, I think they will do it dominantly.