There used to be a time when “Magic competitive play” meant one thing and one thing only: Swiss rounds followed by a Top 8 playoff. With the dawn of MTG Arena, we are seeing a multitude of events introduced that are undeniably very competitive, but with formats that don’t follow the usual formula. Arena is reaching new peaks of popularity, with events like the Fandom Legends tournament offering big cash prizes. One aspect that comes alongside this brave new landscape is something that Magic players aren’t very accustomed to, but is an industry standard in the esports arena: the Ranked Ladder.
Unfortunately, with the introduction of this shiny new technology, a wave of misinformation has disseminated through the Magic community. How does the system work? What exactly needs to be done to stay on top? Can the system be gamed?
Competitive players and grinders can’t afford to rely on false facts, so today we’ll go over some of the most widespread aspersions and see if we can shed some light.
First of all, how does the Ranked Ladder work?
Verdict: We currently don’t have full official information on that. What we do know is that it is based on a “Glicko style system”.
https://twitter.com/ChrisClayPlays/status/1091089344264204289
Glicko is a mathematical model invented by Dr. Mark Glickman, a Senior Lecturer at the Department of Statistics of Harvard University. A full description of Glicko is beyond the scope of this article, but it is beneficial to know at least approximately how it works. Glicko assigns you three hidden scores: rating, deviation, and volatility.
Rating measures your play strength, and is the score according to which the players are ranked: at any given time, the #1 ranked player has the highest rating on the server, the #2 ranked player has the second-highest rating, and so on. Deviation measures how “uncertain” the system is about your rating. A higher deviation signifies more uncertainty, a lower deviation means your rating is more stable. Volatility measures how consistent your results are, with higher scores denoting more variable results.
After each match, the system runs a bunch of math and updates your scores depending on the result. Like in all matchmaking systems, the one-line summary is that defeating high ranked players gives you a lot of points, while opponents with fewer points have lower bounties on their heads. The innovation of the Glicko system is that all those parameters produce a more precise statistical assessment that makes the ranking more resilient and less susceptible to being gamed.
You must play a lot to place high on the Ranked Ladder
Verdict: Partially true, but misleading.
The goal of matchmaking systems is to make it so that, in the long term, the hidden score of each player converges to a reliable measurement of that player’s play skill. Glicko is a well-designed system, and as such it fulfills this goal excellently, even in a game with as much randomness as Magic. In fact, it could be argued that Glicko is too efficient at that. Resetting ranks after each month of play has no scientific purpose, it is just a tool to encourage players to play more. If the system were run according to pure Glicko, your rating would eventually plateau, providing fewer and fewer incentives to continue playing. Playing more games does not inherently provide any advantage.
An unfortunate side effect of time-limited seasons, however, is the increased importance of the last results before the time runs out. Matchmaking systems suffer from a systematic recency bias: in other words, if you plateau to a 95% rating but you manage to go 10-0 on the 31st of the month, your score won’t have the time to converge back to its “real” value.
To summarize: playing 16 hours a day won’t do you any favors, but playing a lot towards the end of the season does in fact help.
When you queue for ranked, you should cancel the match and re-queue if you are not paired within the first few seconds.
Verdict: Unable to either confirm or disprove with the official information available, but anecdotally true.
Apart from the fact that the system tries to match you with players at or near your rating, there is no officially confirmed information about the inner working of the pairing algorithm. Anecdotally, the system tends to try and pair players very close in rating, increasing the difference in rating it deems acceptable in a way that appears to be correlated with the time spent waiting in the queue. Although it must be stressed that none of this is official information, this behavior is similar to the way the Ranked system in other competitive games works.
Assuming you are a relatively high ranked player, it is in your best interest to be matched with players whose rating is as close as possible to yours, making this seemingly useless procedure actually beneficial.
Your rating will decay if you don’t play.
Verdict: Pants on fire.
In a Glicko system, spending time without playing doesn’t decrease the rating (it does increase the deviation, but that is besides the point). During the final days of a season, your position in the standings will quickly deteriorate if you don’t play, but that is simply the result of other players “overtaking” your rating, and advancing past you in the standings.
The apparent rating decay is mainly caused by two factors: the first is the inherent recency bias of the matchmaking system that we discussed before, the second is the distribution of the ratings. At the time of writing, the most relevant reward for doing well in Ranked is the invitation to a MTG Arena Mythic Championship Qualifier Weekend, which is awarded to the top 1,000 players at the end of the ranked season. Because of the way the invites are allocated, players have little incentive to continue playing if they are “safe”, creating a huge blob of players with very similar ratings and the feeling of a decaying rating. Players who hold very high standings do not in fact experience decay at all.
Constructed decks that are good for tournaments aren’t necessarily good for ranked, and vice-versa.
Verdict: The difference isn’t that marked, but the gist of it is true.
One of the first steps towards becoming a competitive player is learning that there is no such thing as a “good deck” or a “bad deck” in a vacuum – decks exist within context. The format is the most obvious piece of context, but other factors, such as the expected metagame, impact the evaluation just as well. The tournament format is one of those pieces of context.
The overwhelming majority of the decks that have ever been played in Magic were originally designed with the goal of winning the Pro Tour. With the top 8 cutoff at 37 matchpoints, we are looking at a bare minimum performance of 6-3 over 9 constructed rounds, and realistically aiming for 7-2 or better. Not even world-class players can average .777 against competent opposition. In other words, you need to spike. And if you need to spike, you need to play a deck that puts you in the position of being able to spike: less linearity, less reliance on the sideboard to correct bad matchups, trying hard not to have unwinnable matchups.
The opposite is true for a system like a ranked ladder where you are allowed to play as many games as you want: you will never average .700 the entire month, therefore the optimal strategy is to maximize your expected win rate and let the die be cast.
This is not to say that decks played in high-level tournaments such as Mythic Championships are unsuitable for ranked on MTG Arena, the difference is minimal in modern Magic. But the optimization processes are indeed different, and might lead to different choices for non-core cards.
If you are high on the Ranked Ladder, it is more beneficial to play best-of-one.
Verdict: Depends on your definition of “beneficial”, but most likely false.
In order to give a precise answer to this question, we should define what exactly is your goal: realistically, finishing #1 might earn you some followers on Twitter, but the marginal utility of finishing #2 over #3 is effectively zero. From a strictly cynical point of view, if you are very high ranked you should stop playing altogether.
Disregarding extrinsic rewards and assuming your goal is to maximize your rating, your problem isn’t speed, it’s the fact that a victory earns you way fewer points than a defeat loses. There isn’t much you can do to fix that problem, and even world-class players can’t win more than about two thirds of their matches against competent opposition in the long run. The only realistic way to give yourself the best shot at advancing is maximizing your expected win rate and hoping to get lucky. Best-of-one is actively harmful in that respect due to having higher inherent variance.
Ranked Ladder heavily favors aggressive or otherwise fast decks.
Verdict: Only true in a very limited sense, and otherwise very false.
The argument in support of theses of this kind generally cites as evidence the fact that aggressive decks can play more games per unit of time, and are therefore “more efficient”, even with a worse win rate.
As we discussed above, our findings tend to contradict this notion: Glicko isn’t based on wins or win rate, playing more games isn’t that strong of an advantage, and we cannot game the system with speed. This is sufficient to dismiss this take in the general case, but there are two cases where playing a fast deck yields a competitive advantage.
The first is when your rating is much lower than your actual play skill could earn, usually at the beginning of a season or after an unusually long streak of losses. In that case, you can converge to the rating you “deserve” much quicker by playing more games in the same amount of time.
The second is if the clock is running out and you need to score some points very fast: football players call it a “Hail Mary pass”; we call it “jamming mono red”.
In Conclusion
For better or for worse, it seems clear that the ranked ladder is here to stay. It’s no mystery that I’m a big fan, but I sympathize with those who point out the problems of the system, chief among which is the inexcusable lack of transparency regarding the details of the ranking and pairing algorithms. The ladder doesn’t need to be the final form of competitive Magic, but it is an important part of a functional competitive ecosystem, and as such it is vital for competitive players to approach it with the correct mindset.
Hopefully today’s article has shed some light on what we know about the system – good luck with your climb to the top!