Data and statistics are incredibly powerful tools to maximize your performance in any complex and competitive game. However, not all data is created equally, and the ability to correctly interpret or scrutinize data is not equally distributed. Without the proper interpretation of statistics, they aren't particularly useful. Because of the massive update to our Snap.Fan Marvel Snap Deck Tracker, we are expecting a flood of data, and today, I'm going to give you a primer on how to turn that torrent of data into a useful pool of information.
On Snap.Fan, you can find data that is populated from our tracker in a variety of places. In the Match Stats section, there is a player leaderboard, the individual card stats page, and patch stats change page. Under the Decks section, there is the submitted decks page and the deck meta page. The leaderboard is useful for understanding how much top players play and their win rates. It can be a fun incentive for top players to track their all-time stats. Also of interest: it takes most players somewhere between 100 and 200 games to reach infinite. That's a day for some people and a couple of weeks for others. The patch stats change page is not current, but keep an eye on that section as we make improvements in the future. The deck meta page is an experimental work in progress, and the archetypes are due for a refresh. This guide will focus on helping you navigate and get the most out of the individual card stats and submitted decks pages. This guide is meant for those new to card game statistics and those who wish to know where to access and how to use our statistics best.
Individual Card Stats
When you open the card stats page, you will see a ranked listing of all cards in the game.
The first column is Games Seen, which indicates how much play a given card is seeing. In the example above Shang Chi has been seen in 145,115 games. This counts the number of games where someone with our tracker was playing Shang Chi in their deck or saw it in an opponent's deck. The seen % is 36.75, which means Shang Chi was seen in 36.75% of games. This information about play rate is useful for a variety of reasons. This means you can expect your opponent to have Shang Chi in their deck 36.75% of the time. You can use this information to understand what is commonly played and plan accordingly. It is also an important indicator of the popularity of a card. A card's popularity has strong implications about the power level of a card and its place in a particular meta game, but further context is required to figure out why a card is played as much as it is. Cards are commonly considered underplayed or overplayed when their win or cube rates don't correlate with their played rate.
Next, we have three sortable categories of game win rates and cube win rates. The first is the in-deck rates or the stats anytime that card is in a deck. Next are the drawn win rates, which only count when you draw that card in a game. Scroll to the right and find the last category, played win rates. This includes only times when the card was actually played.
If you're unfamiliar with in-deck win rates, drawn win rates, and played win rates, participate in this small thought experiment. Try to guess which of these three categories is generally considered the most useful and most likely to be used by both game developers for balance and top-level competitive players. The correct answer is drawn win rates. The statistics for a card when it is drawn are widely considered the most valuable. The in-deck win rate is less valuable because it includes games when the card was never seen— meaning there was never an opportunity to play the card. This category tends to be pretty close to drawn win rate and can be valuable in a few unconventional situations when trying to assess whether a package of cards is good or not, but for most purposes, drawn win rates are most useful. If you answered played win rate, know that you're not alone. Before explaining the reasons, many players think this might be the most important. After all, there is something intuitive about it. This is when you played the card, so it should indicate a card's power. The reason this category is less reliable, though, is that cards play different roles and perform different tasks.
Consider the cards in the table above. Hela has been on top of the played win rates for most of the game's history, not just recently. Think about why: you're only playing Hela if you're pretty sure she's going to win you the game. Occasionally, you may get disrupted, but most of the time, you're playing Hela when you've already achieved what you need to do to set her ability up. Similarly, you're not going to skip turn 5 to play Infinaut if you don't think it's a winning play. Most of the cards above are strong finishers that are highly dependent on having already done something specific. Spectrum is a perfect example. I don't think many people would argue that Spectrum is a top 10 card in the game, but when she is being played, it is only in situations where she is likely to have a major impact and win. Her drawn and in-deck rates (54% and 53%, respectively, at the time of that screenshot) are far more indicative of her current power level. Played win rate can still be somewhat valuable in showing how effective a card is when it “does its thing.” However, because there are so many statistical biases and contextual noise in played win rates, drawn win rate is more generally useful.
Another underrated factor to consider when looking at these rates is to apply some contextual knowledge about deck construction and synergy. Take the example of Sauron. Sauron tends to have pretty good stats because he can essentially only be played in one deck, and that deck is pretty good. This means his statistics are only being populated by this deck. Unsurprisingly, the High Evolutionary set of cards and Miles Morales have good stats for similar reasons.
A card like Jeff the Baby Land Shark would be at the opposite end of the spectrum. I don't think many people would argue that Sauron is a better card than Jeff, who is likely one of the five best cards in the game, but his statistics look better— because a card that can be played in more decks can also be played in bad decks. This is an important context to remember and a powerful tool to discover underplayed gems. Good examples are cards like Ghost Spider and Mirage, who mostly see play in decks that drag their statistics down, or Silk and Titania, who are flexible and see play in enough decks that the data obscures their power somewhat.
One peculiarity of Marvel Snap is that you can climb with decks with negative win rates as long as they are cube-positive. Therefore, the first stat I always check is the when-drawn average cube rate. Drawn win percentage gives a good overall snapshot of how consistently good a card is, but the cube rate is ultimately more important in Marvel Snap. However, this is when it becomes especially important to filter our data by rank. After all, snapping and retreating are skills that not every player has mastered to the same degree. If you feel (or better yet, have data to indicate) that you are an especially good snapper, you can apply that as a factor when looking at cube rates vs raw win rates. Otherwise, the cube rate is probably more important, but you should check both. Again, context matters a lot. Some decks, like Mr. Negative, have wildly higher cube rates than win rates, while other decks will rank similarly in both metrics. There is no quick and easy way to decide when game win rates are valuable as opposed to cube win rates. Instead, it takes experience and contextual meta-knowledge. Your individual habits, play preferences, and strengths will also play a role. There are other cases when you'll also have to apply intuition and context. Namely, when looking at tech cards, ensure you use the appropriate time period and pay attention to the sample sizes. A good tech card today will be useless if the metagame shifts tomorrow. Unfortunately, we have to think about the data we are seeing!
There are a variety of filters, but I usually filter to top 100+ only to ensure I'm excluding bots, and I limit data to the last 7 days. Occasionally, there are reasons to use other filters, especially if the sample sizes for what I'm trying to look at are too small. In that case, I'll want to widen the filters to see what changes when I do.
Submitted Decks Page
We've looked at individual cards and said we needed to apply more context in many situations. The largest chunk of context we can apply is their place within entire decks. When you open the decks page, you will see a list of decks filtered and sorted based on which options you've selected.
You can see the name given to the deck by the person who submitted it, the game win and cube win rates, the name of the person who submitted the deck, when it was created, information about the deck's curve and power averages, and how many times someone has viewed the deck on our site.
There are quite a few filtering options. The filters can be quite powerful, and it is worth investing some time into playing around with them. Generally, I exclude bot data but filter to rank 100+ and often sift through for card-specific information by including and excluding particular cards. There are also ways to sort for qualitative data, such as deck views (how often people check the deck out on our site), last-created decks, and which decks have videos made about them by community members.
One of the most valuable filters is the cards filter, which lists the best uses for a particular card. For an additional example of how this data can be used, check out this extensive article on Red Guardian by TerminalVertex.
For example, here is a list of decks from the last seven days that include Nico Minoru while excluding Deadpool and Venom. It includes data from infinite only and decks with 50+ games played. The list is sorted by cube rate, and duplicates are hidden. This is an excellent way to get ideas, find hidden gems, or try an interesting way to play with a cool new variant.
Below is another example. These submitted decks include White Widow but exclude Annihilus, with whom she is usually paired. This shows that few people are experimenting with her outside of Annihilus decks, where I think she has some potential. I can then check in on these same filters in a few days to see how the card’s use is evolving. The filters give us various ways to evaluate new card releases and assess their importance and value. While White Widow is not being experimented with heavily outside of Annihilus packages currently, it doesn't mean no one is playing her in other ways. There could be several good rogue decks that no one has submitted to the site yet.
It is important to note that only decks someone decided to create and submit on our website will appear here. It does not scrape tracker data and populate it with every single deck that is being played. As long as a particular set of 12 cards has been submitted, though, players playing that deck with our tracker will contribute data for the deck. This means two things: first, some rogue decks will escape notice and fall through the cracks. You can help by submitting decks you are playing with and against to ensure maximum coverage. Second, it doesn't matter who submits a deck. Statistics for that particular set of 12 are aggregated together. You can cut down the number of decks you see on the page by checking the “hide duplicates” box since every identical submission will have identical stats.
Conclusion
It is a brave new world for data-driven analysis at SnapDotFan. The popularity of the new tracker will lead to larger sample sizes, more accurate data, and increased deck submissions. Get ready to see more of this data in our articles, as we writers will also have access to this improved data. SnapDotFan is going to be the premier destination for FREE high-quality Snap data. Download the tracker to get the most out of our data and up your game.