Snapalytics: How to Play More Like the Best
Table of Contents
This week, I dug into game stats at the player level. I wondered — what kind of lessons could be extracted from the numbers if we studied those who are putting up the best numbers? Should I be snapping earlier? Maybe retreating more often? Or is it as simple as needing to up my win rate?
Some details upfront to keep the rest of the article neater.
The Sample Group
The data set referenced throughout the article contains matches meeting the following criteria:
- During the past two weeks
- Player rank 50–99
- Player collection level 2100+ (roughly halfway through series 3)
Any player who didn’t record at least 100 qualifying matches during the past two weeks was excluded.
I checked to see if this group was representative of other groups. The numbers are similar for players with lower ranks or lower collection levels, but they’re quite different once players hit Infinite rank (100+).
I believe this is because the stakes are so different in Infinite. Players who reach Infinite no longer need to hustle for higher rewards, and their rank cannot fall below 100 for the rest of the season.
I’ll be referencing the average number of cubes won per match frequently throughout this article. From here forward, I will call that stat cube rate.
I started with the top 200 players by cube rate. What I learned was, well… kinda nothing. There were no insights that were readily apparent. Instead, I found a shocking amount of variety. For demonstration, here are the kind of extremes found within the top 20 (yes, twenty, not a 200 typo):
|Average snap turn||2.96||5.66|
Those are some huge ranges, right? Like, the #6 player by cube rate is snapping almost 5 times as often as the #12 player!
That wasn’t a total bust. I think it’s super interesting that there are so many different ways to play and still maintain one of the best cube rates in the game. You might need to find a deck that suits your playstyle, but you can climb whether you prefer to play super safe or extremely aggressive.
But I’d be out of any insight beyond that (and it wouldn’t be much of an article). I figured there had to be something useful there. Maybe even something juicy! I wasn’t ready to throw in the towel and change topics, so I dug deeper.
I expanded the search in two directions. First, was to include all qualifying matches. That boosted the sample size by an order of magnitude — there were now over 2,000 players and 500,000 matches!
Second, was to explore more intricate metrics. Instead of just “snap rate”, I broke down what percentage of snaps happened on each turn, from confident first-turn snaps to final-turn “boomer snaps”. Instead of just “retreat rate”, I looked at how often a player retreated after snapping, on the same turn their opponent snapped, or on a turn after their opponent snapped. I had 42 metrics by the time I was done.
The next step involves a bit of math nerdery. I’ll keep it basic and relevant. r² (pronounced “r-squared”) is a measure of correlation. It describes how much the value of one variable can be explained by the value of another variable. I calculated r² between each metric and cube rate, and then again between each metric and win rate.
Metric: Win Rate
SafetyBlade posted a great article discussing cube rate vs. win rate today (a happy coincidence). I looked at the r² between those, too. Unsurprisingly, it was the highest of all the r² values — 51.6%. For simple two-variable correlation, r² works both ways. In other words, you could read that as either “The win rate explains 51.6% of the cube rate” or “the cube rate explains 51.6% of the win rate”.
That makes win rate a massive factor in cube rate, but I’m maintaining focus on cube rate for this article. It’s ultimately your cube rate that determines how fast and how high you climb the ranks. Besides, you can’t strictly try to maximize your win rate. Maximizing your win rate would involve never retreating, which is one of the most important ways to minimize cube loss!
SafetyBlade came to the conclusion that both are important, but win rate is more reliable and also more important when it comes to looking for a new deck. I agree. Win rate only explains 51.6% of cube rate. There are many other factors, including the matchup with your opponent’s deck, which locations show up, and the luck of your draws. Oh, and skill. Knowing when to snap and when to retreat is a huge part of cube rate. That involves both a generic skill level and deck-specific skill that comes from getting lots of experience with your deck.
Metric: Unique Decks Played
That leads into the strongest factor I found after win rate: how many decks did the player play with over the past 2 weeks? This had an r² of 5.3%. A player is expected to lose 0.008 cubes for each additional deck they played with. I’m glossing over the details, but it’s an easy logical leap that most players who have played lots of decks in a short timeframe (especially below Infinite rank) end up as jacks of all decks and masters of none. Taking the time to master a deck is a critical factor in maximizing your cube rate.
A quick break for those of you already familiar with r². 5.3% might sound like a very low r² to you. And, it is — for the sake of prediction. If all I knew was how many decks you played in the last two weeks, it’d be incredibly hard to guess your cube rate. A low r² doesn’t mean it’s worthless, though. A variable can still be significant even if it has a small effect size. And the sample size of our data is more than big enough to ensure that the effect of the metrics I’m highlighting today are statistically significant — not just results of random chance.
A small bump in your cube rate can really add up over time! If you were to raise your cube rate from 0.5 to 0.525 (just 5%), you’d climb 10 ranks over 190 games instead of 200.
I’m gonna go through some of the other metrics with small-but-significant effects at a faster pace.
Metric: Snap Rate
Snap rate has an r² of 1.7% with cube rate.
This one seems straightforward: snapping opens up bigger prizes (but also risks bigger losses).
Metric: % of Snaps You Retreat
The percent of your snaps that end in your own retreat has an r² of 1.3% with cube rate.
Note that r² just describes the strength of the correlation, not whether the relationship is positive or negative. In this case, like for unique decks played, the relationship is negative. That is, the more often you retreat after snapping, the lower your cube rate will be.
Metrics: % of Snaps on Certain Turn
The percent of your snaps that happen on a certain turn proved kind of unusual. Only turns 3, 4, and the final turn had correlations high enough to be worth noting (1, 2, and 5 have a combined total below 0.4%).
For the final turn, I am considering whether or not Limbo was present.
I’m joking a bit with “boomer snaps” being bad. It’s clear we have plenty of players who often (or even exclusively) snap on the final turn while keeping cube rates that are positive or even high (above 0.5)!
Metric: Retreat Rate
Retreat rate has an r² of 0.7% with cube rate.
The correlation is weak (but statistically significant) and the relationship is just slightly negative. Note that there are a couple players who retreat more often than not but still maintain a positive cube rate!
Metrics: When You Retreat If Opponent Snaps After T1
These next two are kind of complicated, but stick with me. Take it as a given that your opponent has snapped on turn 2 or later (I excluded turn 1 because you have almost no information to base a retreat on at that point). In that situation:
The percent of times you retreat on the same turn that your opponent snapped has an r² of 0.6% with cube rate.
The percent of times you retreat on a later turn than your opponent snapped has an r² of 1.6% with cube rate.
See how retreating on the same turn your opponent snaps has a slightly positive relationship while retreating later has a slightly negative relationship? That seems obvious (it’s not until the turn following the snap that the stakes actually double), but I really think it’s worth highlighting. If you’re gonna wind up retreating anyway, it’s so much better to do it before your opponent’s snap takes effect.
These are the metrics we covered today, sorted by how influential they are upon cube rate:
- Win Rate: 51.6%
- Unique Decks Played: 5.3%
- Snap Rate: 1.7%
- % of Opponent Snaps After T1 You Retreat on a Later Turn: 1.6%
- % of Snaps You Retreat:
- % of Snaps on T3: 1.3%
- % of Snaps on T4: 1.2%
- % of Snaps on TFinal: 0.8%
- Retreat Rate: 0.7%
- % of Opponent Snaps After T1 You Retreat on the Same Turn: 0.6%
It was really cool (and kinda relieving) to see so much variety in key metrics among high-performing players. It seems that all sorts of playstyles can deliver high cube rates. That’s one of the best features of the cube system — it adds more nuance in what it means to perform well. That leaves the game open for a wider variety of viable decks and playstyles. Win rate is super important, but it isn’t everything (just check out how many players have positive cube rates but have win rates below 50%).
In fact, win rate explains only 51.6% of cube rate. That leaves room for lots of other influences. Some of which are based in luck (like which locations appear and what order you draw your cards in), but there are many that are in your control.
The most important of which seems to be to pick a deck or two and stick with them for a while. Take the time to learn their ins and outs, strengths and weaknesses. The time invested to master a deck will inevitably lead to improving both its win rate and cube rate while you’re piloting it.
Ultimately, finding a playstyle that suits you is critical to both the fun and success you’ll find when competing on the ladder. Discover the playstyle you like, find a deck that suits it, master that deck, and then there’s a list of several small factors you could work on to maximize your mastery — and to squeeze the highest cube rate possible from your favorite deck.
Of course, staying knowledgeable about what’s common and what’s enjoying success in the metagame is a huge part of having competitive success. The knowledge you bring to the game is really hard to measure quantitatively, but I’m sure it’s another important influence on your cube rate. Keep coming back to stay on top of the meta and pick up more expert insights!
This analysis was powered by the data from our Marvel Snap Tracker. If you play on PC or Android, consider giving it a download! It has many features including a live overlay, the ability to sync your collection to our site, tracking your match data for your own reference, and powering useful tools like the Meta Snapshot.
Enjoy our content? You can Support Marvel Snap Zone and your favorite content creators by subscribing to our Premium community! Get the most of your Marvel Snap experience with the following perks for paid membership:
- No ads: Browse the entire website ad-free, both display and video.
- Meta Reports: Exclusive daily meta reports, such as the Top 10 Decks of the Day and Trending 30 Cards tailored for you!
- Early Access Tier List: Exclusive early access to the weekly Tier List update (decklists, rankings, and article) 24 hours before the article goes live! Get an early start on your competition!
- Premium Dashboard: Get full instant access to the member-only dashboard, the all-in-one page for all your benefits.
- Support: All your contributions get directly reinvested into the website to increase your viewing experience! You get also get a Premium badge and border on your profile.
- Discord: Join our Discord server, claim your Premium role and gain access to exclusive channels where you can learn and discuss in real time!
- Special offer: For a limited time, use coupon code SBYREX4RL1 to get 50% off the Annual plan!
“During the past two weeks
Player rank 50–99”
This is your problem.
Very good players are already in infinite.
Be more interesting to do this with ranks 80-100 in the 10 days after new season starts.
You’re totally right, I’ll revisit it then.
It’s an interesting analysis though and it may actually be no different with the higher quality player base – which would be a result worth finding. Or maybe there IS a discernible difference between the two. I look forward to seeing the follow up!
Great Article, compliments!
Thanks so much! ☺️
Love me some good analysis. 🙂