Last month, Wallabeebeatle made an insightful video on how to build ranking systems, whether they be numeric or based upon personal judgment. I highly recommend watching the video in order to understand his reasoning and appreciate the ideas being put forth. Many statements made in the video were worthy of expansion, and some questions were raised but left open. Given how Wallabee has already framed the discussion, I will diverge and build upon his claims, providing an alternative position to his.

1. Should results be the metric used to determine rankings?

“...results aren’t going to be perfect but results obviously are going to be the best measuring device of actual skill but the whole thing is that results aren’t skill; they are an approximation are those approximations aren’t going to be perfect.”

The recognition that results are simply used since they are assumed to represent a team’s skill is an astute one observation. Moreover, the idea that these numbers are imperfect representations is very interesting, but before we gauge whether or not results are an adequate measure of a team’s skill, we must determine exactly why they is an imperfection in the approximation.

It is assumed that results are a representation of skill because a team’s skill here is quite literally just their ability to win games. The word “skill” is misleading in this context, as it is normally used to refer to an individual player or team’s overall aiming capability. In this case, firepower is simply one input among many that contribute to a team’s skill. To avoid this confusion, when referring to a "team’s skill" I will use the term ‘ability’.

While a team’s ability is a synthesis of all they can do to influence their results - map pool, firepower, adaptability, tactical prowess, anti-strating, etc…- it is not the only source of inputs that determines how they place in tournaments. Wallabee mentions randomness, which is most certainly a factor in Counter-Strike. Much like in hockey, where a puck that missed the net can bounce off of a defenceman’s helmet and go in, CS:GO has a randomness to it. Whether it be killing someone with the last bullet in a shoddy spray or hitting a long-range no scope, RNG can decide whether you win or lose a potentially pivotal round.

Yet there is a more prominent factor that has been undervalued when determining placings: the opponent. Which teams you have defeated and which you've been defeated by add much-needed context to placings. To use a recent example, using simply placements for a ranking system, both Virtus Pro and SK Gaming had a very similar placing in Star Series. Sometimes victories can be the difference -as Team Liquid beating CLG to reach top 4 at MLG Columbus -but at Star Series, it was the defeats. Virtus Pro got absolutely bashed by middling teams like Fnatic, NiP, and SK, getting collectively 9 rounds in three defeats. On the other hand, SK lost to a much better trio of teams in G2, Astralis, and a powered up FaZe. In addition, some of their games were somewhat competitive, particularly the one against Astralis.

For all intents and purposes, 9-11th and 12-16th are the same ranking. Even if VP had beaten another team like Heroic, SK’s result would still be significantly better. Nevertheless, the rankings would reward/punish them equally if we limit ourselves exclusively to results. The opponents SK faced were significantly more challenging than those Virtus Pro faced off, and an ideal ranking system would take this into account. G2 losing to FaZe in the Quarter-Finals is not equivalent to North losing to Hellraisers.

It’s important to remember why we use placings in the first place; it is an assumption that ability is manifested in results. We are trying to emulate ability through placings, but these examples show that every single tournament, results, used without context, do not probably represent ability. Though they didn’t quite do so, G2 could’ve performed exceptionally and looked as if they would have dominated both North and Hellraisers, while being rewarded less than Hellraisers and as much as North.

Another issue pertains to the relationship between a team’s quality and their results. It may well be that analysts prioritize certain inputs that a successful team is lacking when discerning a conclusion. A great example of this is the early 2016 Fnatic squad who won six tournaments in a row. While they were unquestionably the best team in the world, experts remain tentative in naming them one of the best teams of all time, despite their excellent results. As a matter of fact, I would postulate that Astralis is already a better team than that Fnatic roster was/is, despite having fewer titles.

Why is this? Simply put, Astralis have the input qualities of an all-time team: great map pool, systematic play, and just damn good, proper CS. Fnatic, on the other hand, had a boisterous play style with essentially no strategy. They had only three of the inputs that lead to success: legendary team play, excellent skill, and an incomparable swagger. Despite these qualities and the superior results, I would still argue that in an all-time ranking Astralis would place higher since they have a more well-rounded, reliable approach to the game.

To me, it seems that rankings must take into account the opponent, as placings are not a manifestation of a single team’s ability, but a relationship between that of the team in question and that of their opponents.

Now, this may be irrational, and it may be that the proof must be in the pudding, but if Astralis fails to win any other tournaments, they will have had a weaker set of rankings than Fnatic, despite winning the major. Competition for the first half of the former team’s run was weaker as they only faced VP as a true foe, while Fnatic had the likes of Na’vi and Luminosity to conquer.

Nevertheless, I consider these criteria of strategic depth to be important, but it may be simply because they hint at a higher level of sustainability. If that is that case, it may be that Astralis should have to prove that their run of form is more sustainable than Fnatic’s before their well-roundness are considered in their rankings.

2. Should we count online results?

This is a never ending debate. I am of the same position as Wallabee here, but those who disagree rarely get a proper answer as to why that is. Online results can be an indication of a team’s ability. The problem is that too often they aren’t. Whether it be cheaters or onliners or bad ping or lack of focus from established teams is irrelevant, these results translate too poorly to LAN to be considered truly indicative of ability. However, though it can be a fluke, it can also be a meaningful result. An example of a proper online result would be Hellraisers, who a squad looked good online before heading into Star Series, where they reached Top 4.

The problem with online results is that we didn’t know if those results were legitimate until they were confirmed on LAN. Due to this issue, online play ought to be considered more of a hint to help one evaluate ability, but by no means a proof of it. I do think they should be weighed in somewhat, especially in LAN droughts like the one we saw before the E-LEAGUE Major.

3. Should the proximity of the score have an effect on the ranking?

Wallabee argues that since what matters is the victory, whether 2 or 14, the result is that same, but I actually disagree with him here. Going back to the assessment that rankings are an attempt to grasp a team’s ability, I would argue that a team that reached 14 rounds demonstrated a higher level of prowess in that map than if they had gotten merely 2. A great example of this occurred at Star Series, in which Astralis got edged out on Inferno to lose the title to FaZe.

Now, due to the previous record of Astralis, they remain the top-ranked team in the world, but both moses and Thorin echoed the sentiment that Astralis looked like the better team on that map. In other words, they demonstrated the greater ability. In less extreme cases, it may be better to say that they didn’t “75% win”, but rather they demonstrated certain skills that input into ability: firepower, map prowess, good calling, or whatever may apply in that context.

What matters here is that they have demonstrated something that the 2-16 team did not and that something, whatever it may be, is necessarily an input for ability. If we recall that ability is described as the ability to win, they have demonstrated that they possess a certain portion of the inputs required to success, which makes them marginally or perhaps even significantly more worthy of a high ranking.

Conclusion

Image credit: Valve, ESL