Tournament Tier System - PGRU Season 2 Fall 2019
Note: Everything in this document is accurate as of July 15, 2019.
Hello and welcome to a text document hosted on the internet. I know I’ve probably already lost 80% of you from the realization that this isn’t a video. Yes, you need to actually read the information presented to you, instead of just waiting for a pleasant-sounding voice to pass it through your ear canal, up your auditory nerve, and into your temporal lobe. Unfortunately my voice does not sound particularly pleasant, so trust me when I s̶a̶y̶ write that this way is better for everyone involved.
Now for the actual reason why you clicked on this link (barring any successful clickbait headlines): you actually want to know what the heck is up with the so-called Panda Global Rankings (PGR). Well, you’re in luck. I’m Andrew “PracticalTAS” Nestico, and I’m the architect behind the PGR algorithm (aka the thing that actually spits out our list of the top 50 Super Smash Bros Ultimate players in the world). So if you have any burning questions on your mind, you’ve come to the right place.
What is Panda Global?
Ah, I see we’re starting very high-level. That’s fine, everyone’s gotta start somewhere. Panda Global, or PG, likes to call themselves the World's Premier 1v1 Esports Team, caps and all. Their roster of players includes specialists in Smash Bros, other fighting games, and some curious online card games.
And what is the Panda Global Rankings?
For reasons which still elude me, PG also has their own proverbial island of misfit toys called PGStats, a department dedicated to obsessively chronicling and publishing lists of different games’ best players for public consumption (read: public outrage). These lists are called Panda Global Rankings, and currently include twice-yearly releases for Super Smash Bros Ultimate and Super Smash Bros Melee. The first of the calendar-year-2019 releases start on July 22 and 23 for Ultimate and Melee, respectively. As previously mentioned, the Smash Ultimate list is calculated directly from tournament results from an algorithm written by yours truly.
So...it’s just a list of players?
Oh no no no, dear reader, it’s not just any list. It’s the list. Thanks to Nintendo’s helpful lack of any level of developer support for Smash in the form of a pro circuit, the competitive Smash community fuels its “who is the best player?” speculation with hobbyist-made lists. Of these, the PGR is the one people care the most about because it’s the most official-looking.
Wait...the most “official-looking”?
Aha, that was a test to see whether you were still paying attention or just skimming at this point. You’ve passed. The PGR is indeed the most official-looking Smash ranking, but it’s also the longest-running and most sustainable (given our backing with a s̶u̶g̶a̶r̶ ̶d̶a̶d̶d̶y̶ major esports team). Critically, we also haven’t yet gotten ourselves canceled by publishing a list so bad that the community refuses to acknowledge it...but there’s still time for that.
So how does the PGR actually work?
The nitty-gritty mechanics of the ranking algorithm are not public, partially because I have “narcissistic tendencies” and “an inflated sense of self-worth”, but also because publishing it would mean that people could calculate the list as soon as the ranking season ends, ruining our reveal.
That...didn’t really answer my question at all.
Fine, fine. At a high level, the rankings take into account the tournaments you went to, the players you beat and lost to, and the players you outplaced or were outplaced by (it doesn’t care directly about the numerical placing, only who’s ahead of and behind you and by how much). It then performs an iterative calculation to determine the relative strength of each competitor, which directly corresponds to their ranking.
In simpler terms, you gain points by beating good players or placing higher than them at tournaments you both attend, and you lose points by losing or getting outplaced. How do we figure out who the good players are? We look at the players they beat and outplaced, and compare that to the players they lost to and were outplaced by. If this is starting to sound like a chicken-and-egg problem, that’s because it is, and that’s why we have math to solve it. If this was easy, I would have had useful sources to help me when I got started instead of basically building this thing from scratch via trial and error over the past 4+ years. No, I’m not still bitter, why do you ask?
Okay...Anyway, do you count every tournament?
Oh dear god no. Even if we only include in-person Smash Ultimate events with 100 or more entrants in 2019, we’re already at roughly 600 tournaments halfway through they year. That’s too many, so we have a Tournament Tier System (TTS) which defines which tournaments qualify. For the Spring 2019 season, events held in the contiguous US (not to be confused with “continental US”, a mistake I’ve been making for years and only fixed yesterday) with 200+ general entrants, and events held anywhere else with 160+ general entrants, count towards the rankings as long as they are:
- not locals or weeklies, biweeklies, or triweeklies,
- not “Arcadians” or amateur brackets, and
- not occurring in the same metro area within a few days of a much larger event.
In the 24 weeks of the PGRU Spring 2019 season, 84 tournaments from 10 countries met these criteria and qualified for PGR status.
For future PGRs, we will also rate each tournament with a “top player score”, calculated based on how many of the players who were in the most recent top 50 are competing at the event. The larger of a tournament’s two scores (general entrant and top player) is used as the actual score of the event.
Scores? I don’t care about tournament scores. Give me something simple and easy to digest so I can quickly tell which tournaments are important.
Don’t worry, my lazy friend, we have you covered there too. Tournament scores are translated into four tiers; S Tier events are the largest and most prestigious, followed by A Tier, B Tier, and C Tier. You’ll find that tournaments actually use their tier qualification as a selling point when marketing their event.
By the way, the tier labels themselves have absolutely no effect on the actual algorithm. They’re 100% just for show, but I do have to admit that “S Tier” does roll off the tongue better than “2400-point PGR-qualified event”. Also, if you want more granularity than just 4 letters encompassing the full range of events from supermajors to glorified locals, the TTS lists a modifier which splits each tier into “Low”, “Mid”, and “High”.
Hold on a sec. What happens to tournaments between now and the ranking release?
Good eye, reader. I didn’t think you’d catch that. As we give events a “top player” score based on rankings that haven’t come out yet, what happens with the events that occur after a season ends but before its rankings are released?
We don’t have an offseason; events that occur now will count towards the next rankings. We keep track of events that qualify for the rankings and retroactively reveal the ones that do after the release. Unfortunately this does mean that some people will compete in events which qualify for the rankings, without knowing until a few weeks later that the event qualified. On the bright side, you can make educated guesses on whether an event will qualify by top player score, and entrant score is already public.
You haven’t mentioned invitationals yet. I’m starting to get suspicious.
Alright, alright, you got me. For season 2, we will be trial running the inclusion of invitationals on the PGR.
Invitationals will not contribute to pgr qualification, but they will count towards your score. In other words, players must attend a minimum number of open-bracket events to qualify for the pgr, whether or not they have good performances at invitationals. This means that a good performance at an invitational alone will not be enough to make it on the rankings.
This change means that the upcoming Thunder Smash 2 Invitational is likely to qualify as a C Tier event, and Smash Ultimate Summit 2 is likely to qualify as an A Tier event.
I still don’t know. That didn’t go so well last time.
The PGR algorithm has had some hiccups in the past when it comes to invitationals. We’re confident that those issues have been solved; in additional to the qualification change listed above, the algorithm has been tweaked to be less rewarding to players who “peak” at an invitational without having similar results at open-bracket events.
Why does the system make it easier for international tournaments to qualify?
The US is traditionally the epicenter for competitive Smash events; most of the top players live there and most of the largest events are hosted there. It’s far easier to have a large Smash tournament in the US than elsewhere, so making that happen elsewhere should be rewarded. We also want to avoid causing a feedback loop where most of the top players are in the US, so most of the big events are in the US by top player score, so players who can’t come to the US don’t make the top 50, so even more of the top players are in the US, and so on.
Of note: even with the international multiplier in effect, roughly 60% of tournament points allotted for the Spring 2019 season were awarded to the contiguous US. The other 40% was split between Japan, Canada, Mexico, Hawaii, Puerto Rico, and six European countries.
Why isn’t prize money a factor in how a tournament is weighed?
Several reasons, all of which stem from the following scientific theory, which I have just named PracticalTAS’s Second Law: “After taking into account the players in attendance, prize money has no additional predictive power over how difficult a tournament will be to win”.
First off, Japanese Smash tournaments are traditionally held without prize pools due to Japanese gambling laws, so any prize pool factor would nerf Japan (the weebs in this community would never stand for that). Second, invitationals tend to come with large prize pools, so we don’t want to “double-dip” and count exclusive events as even more valuable than they already are. Third, back in the d̶a̶r̶k̶ ̶a̶g̶e̶s̶ early Smash for WiiU days, when we did take prize money into account, we found that events would chip just enough money in as a pot bonus to get to the next TTS tier; not only is that a very poor use of funds, but it’s also unfair to events without the resources to do that and events held in places that have poor exchange rates against the US dollar. Yes, other countries use currencies that aren’t the US dollar, and yes, that needs to be taken into account.
Are you making any other changes to tournament qualification that we should know about?
New this season, we are lowering the B Tier qualification from 800 points to 600 (300 US/240 international entrants, previously 400/320). This stems from the fact that C Tier was very very large in the Spring 2019 TTS. The minimum C Tier requirement will remain at 400 points (200/160), as will A Tier at 1200 (600/480) and S Tier at 2400 (1200/960).
Other than that and the previously listed changes (invitationals counting and the reintroduction of top player score), I don’t have anything else. Thank you for making it this far.
But I didn’t even do anything. I’m just a rhetorical device meant to provide structure for the reader and a foil for the author to play off of.
Shhhh. Let me have some fun. I only get to communicate with the outside world once every six months.
Can you at least put a summary at the end for the people with short attention spans and no appreciation for text-based humour?
Fine, but it’s being listed without context, so they might as well go back and read the important bits anyway. Tl;Dr:
- We are trialling the inclusion of invitationals in the next PGR season. Scroll up to get important context about this decision.
- As with prior seasons, the upcoming season will see tournaments’ tier classifications evaluated based on top player attendance. Scroll up to get important context about this decision.
- There is no offseason between ranking seasons. Tournaments which occur between now and the release of the Spring 2019 rankings will be retroactively rated via top player score in addition to their general entrant score. Scroll up to get important context about this decision.
- For the upcoming season, the B Tier qualification threshold will be 600 points, down from 800. You don’t need to scroll up for this one, that’s pretty much it.
You can find the list of events which qualify for the Fall 2019 PGR (aka our Tournament Tier System) here.