How to Improve Your MTG Draft Skills Using Data (In 4 Simple Steps) • MTG Arena Zone
HomeHome > Blog > How to Improve Your MTG Draft Skills Using Data (In 4 Simple Steps) • MTG Arena Zone

How to Improve Your MTG Draft Skills Using Data (In 4 Simple Steps) • MTG Arena Zone

Apr 26, 2023

Table of Contents

Limited is a challenging format. Navigating through a draft is difficult. It takes time and repetition to get really good at it. But not as much time as you may think, if you use the right learning tools. Today I will try to give you some tips on how to improve rapidly in draft with the use of 17Lands.com data. The greener you are in limited, the bigger the improvements you can make using the highlighted approaches, so if you have some friends you want to get into drafting – make sure to share this with them. If there is one thing better than drafting, it is drafting with a group of friends. There is no better way of opening a draft booster box, so it is worth it to help your friends to hook up on the, arguably, best way of playing Magic. And what better way to hook them up than to give them tools that help them draft better?

This article will propose 4 simple steps on how to improve your draft skills. The steps come with some data showing you the impacts of not taking them, the reasons why players make those mistakes, and what is most important, actions you can take and methods to use to avoid those mistakes. The article is for everyone interested in improving their limited skills, but players at the beginner to intermediate level, those starting to draft or just planning to start to draft will benefit the most. It will not make you play your games better, but just applying the knowledge you will find below to your drafts and sealed pools will make the decks better enough that your win rate should improve. And especially on Arena, those extra wins translate to gems and make your adventure in limited a much cheaper one.

The biggest difference between beginners and draft masters is not the knowledge of the good cards, but the understanding of the bad ones. Players with lower win rates are more likely to play underperforming cards, and this is part of the reason why their win rate is low. Being able to identify bad cards, and avoid picking them is in my opinion the single biggest level up you can make while learning to play limited. I am not alone in this opinion, this has been also a theme of many a pieces of content by, in my opinion, one of the best Limited theoreticians in the field, Chord_o_Calls. And since his thinking impacts my ideas, I thought it would be rude not to give him a shoutout. But even though the idea of not playing bad cards is well established, there was little data evidence that is is true. So in preparation for this episode, I looked at that: do players with lower win rates play bad cards more than the good players?

To answer we need to look at two things. Firstly, do lower win rate players pick the cards with lower win rate higher? Secondly, after they pick the cards, do lower win rate players put the cards with lower win rate into their decks more frequently. To answer the first question, you need to look at two pieces of data from 17Lands.com.

First one is the number of times a card has been seen by 17Lands.com users, and second one is the number of times the card was picked after it has been seen. In the screenshot above, I used the example, you can see Archangel Elspeth is a highly picked card, 80% of the time 17Lands.com users see it, they pick it. Other cards in the screen shot are less popular – Attentive Skywarden was seen ~290k times, picked ~30k times, so just over 10% of the time it is seen, it is picked by the 17Lands.com user. But the data in the screenshot I used is about all the users. To compare more successful and less successful players you can switch between the bottom, middle and top player tiers on 17Lands.com. The tiers indicate a long term win rates of the players in them – players from the top tier will have win rates of around 60%, players in the bottom tier – around 48%. I calculated and compared the pick rates for the top and bottom tiers and this in turn showed me which cards are picked more frequently by players with lower win rates.

Above you will find the uncommons that are picked by the bottom win rate tier players more frequently than by the top win rate tier players. You can see that a majority of those are white cards. But not only white cards, more specifically white cards that are not particularly good at being aggressive. Invasions, Tiller of Flesh are all cards that promote or benefit from long games. And this is the opposite of what white decks want to achieve in March of the Machine, as I showed before when analysing the Battle mechanic in the set. There are some other non-white cards on the list, including Renata, Called to the Hunt and Harried Artisan – both cards being underperformers.

But my personal evaluation that those cards, overpicked by players with lower win rates, are weaker is not enough in a stringent analysis. Yes, I am an informed limited player, but we all have our biases and blind spots. In order to test if low win rate players indeed tend to overpick bad cards, I looked at the pick rates of bottom 50% cards in the set. Lower half consists of bad cards and fillers. Based on all of them I calculated the average pick rate for those mediocre to bad cards for top and bottom tier on 17Lands.com.

And we have a clear 3%p. difference. It may seem little, but we are talking about a lot of cards and each of them (on average) is picked a bit more frequently by low win rate players, leading to larger differences in final pools. Over a draft you see 360 cards, and a majority of them will be from the bottom half of the win rate, as the great cards get picked up early during the draft. This means that the seemingly small difference can lead to 6-8 more mediocre cards in the final draft pool than a top win rate player would have, which means some of the mediocre cards will make it to a final deck, which will have impact on results.

Which brings us to the other issue lower win rate players have: picking a card in a draft doesn't mean you will play it. Because of that, on top of looking how frequently are the cards picked, I also looked at how frequently they are played when picked. To do so, I calculated how many games are played with a card per each time it is picked.

Wait – you may say – how can you compare those numbers for a group of players who win 48% of their games and 60%? Of course players with 60% win rate will play more each card because they win more, so play more games per each draft they play. Don't worry, in order to remove the impact of this discrepancy, I normalised my results, to make the results comparable. In simple terms, I calculated how many games each tier of players would play with each card if their win rates were identical rather than 12%p. different).

Now, I can convincingly show how many games are on average played with each card across the skill tiers. Since the win rates are approximately the same, the higher the number, the higher the percentage of the decks a card is played in. Due to the calculations I made to normalise the data, if you play a card 100% of the time it should play 5.5 games – so at 2.25 games, a card makes the deck 50% of the times it was drafted. Which cards made the decks of the lower win rate players disproportionally often?

We see some of the same cards we have seen on the previous graph. Seraph, Tiller, Renata, bunch of the weaker Invasions, Harried Artisan. The list is supplemented with Copper Host Crusher and Ravenous Sailback. What does it mean? It means that not only lower win rate players picks given cards more aggressively. They also play them more often. The measure is linked to how often are those cards played when picked. But they are not picked at the same rate, so for example Seraph of New Capenna is picked 2x more frequently by lower win rate players and then, despite being picked more frequently, it still is more often put in the decks of those players. This strongly suggests that low win rate players actively want the cards on those lists. Which in part is good news for them.

Why good news? There are two likely reasons why beginners put more bad cards in their decks. One is evaluation issues. The other is draft navigation problems. You can imagine a player not being able to find a good draft lane and because of that being forced to pick and play weaker cards. It is much harder to learn how to change your draft navigation and how to find open draft lanes. It is a skill that takes years to master. On the other hand, evaluation is relatively easy, especially if you have 17Lands.com data to guide you. As simple as learning which cards are traps and actively avoiding them.

You can learn cards that are traps actively for each format and it certainly is a good idea. But we can also look at the traps from MOM and maybe see some broader categories of cards for you to keep in mind. In order to do so I looked at the compounded impact of cards being picked more and then played more than picked. To come back to Seraph of New Capenna example: it was picked 2 times more frequently by lower win rate players. But it was also played 40% more often when picked by lower win rate players. 2 x 1.4 = 2.8. This means Seraph's low win rate impacted lower win rate platers 2.8 times more than it did top players, because they both picked and played it more. What were the worst offenders on that list? I calculated this compounded impact for all cards in the set, not only uncommons like on the earlier graphs.

The two most overplayed cards by a large margin were Jin-Gitaxias, Core Augur and Seizan, Perverter of Truth.

Both cards a rare and both truly awful. Jin is a 10 mana, so almost an uncasteable creature. Which is probably great news for people who put it in their decks, as if you actually manage to cast it, it kills you in 1-2 turns. Seizan is an expensive threat that will most likely generate loads of card advantage for the opponent in some games or be unplayable in others as your life total will be too low. And this is sort of the theme of a good part of the graph. We have a solid collecrion of borderline unplayable rares and mythics. Vorinclex, Voice of Hunger, Firesong and Sunspeaker, Teysa Karlov, Complete the Circuit, were a neat try by the design team to put something interesting into the set. But they missed. And invested players got the memo, the lower win rate players didn't, and still try to make those cards work or play them because trust the design team too much and think that a rare should be good enough for their deck. Other cards on the list are either too expensive (Copper Host Crusher) or too situational (Atraxa's Fall) or too weak (Placid Rottentail, Burning Sun's Fury). And experienced and successful players are much better in recognising that.

So, how to use 17Lands.com to start avoiding bad cards? Firstly, look at the Game in Hand Win Rate column in the "Card Performance" tab on 17Lands website. Look at the cards you like playing. Do you see any cards you play frequently that have low win rate? Do you find any results that surprise you? Cards that you would have thought are good that have low win rate? Well that is a strong signal that your evaluation might be off. Unless you have a good reason for playing those cards, think about lowering priority on them. Also think why they would have a low win rate, ask around, check if top 17Lands.com win rate players pick the card high or low. Answering those questions would help you reaching solid conclusions.

And once you realise you are playing some bad cards, ask yourself another question: Do you play the bad cards because you chose to play them or do you play them because after the draft portion you have no other choice. If you know you are playing bad cards but you navigate your drafts in such a way that you frequently are left with no choice but to play them, you need to work on draft navigation and finding your lane. This is a harder problem to solve, but there is plenty of good content available on finding the lane in draft and I recommend you to use it, like my article on how to wheel cards in draft, a strategy that will let you get more playables and at the same time test if your draft lane is more or less open.

If you play bad cards because you like playing them – you have two options. One is to reevaluate them and stop drafting them highly. This will increase your win rate, but will stop you from playing the cards you enjoy. Alternatively, you can continue playing them – you will lose more, but you will do so while playing cards you like to play. Consciously reducing your win rate to have more fun is not a bad strategy, as long as you are aware of what you are doing.

Even though it seems like playing good cards seems like an important skill to have, it is much less important for your growth than not playing bad cards. For a very simple reason – we are much better at recognising good cards than at identifying bad ones. Lower win rate and beginner players are quire OK in seeing which cards are powerful. Still, here are whole categories of cards that are harder to evaluate. Their quality will often escape the attention of lower win rate and beginner players. Especially since those cards are frequently harder to play, so beginner players may not be able to maximise their utility.

So which cards top players pick much more often than the beginners? Is there a pattern in what they do?

First thing to notice is, 14 out of 15 cards on the list have some blue in them, Lurrus being the only exception. More experienced players will put themselves in a position to draft the top color in the given set more frequently. This is not to say they force it, but they will be more likely to abandon other colors, more likely to stick with the best color, more likely to pick a solid blue uncommon over a medium rare in other colors. All those mean that they will be slightly more likely to be in blue: 61% of two-colored decks of top players have blue in them, that fraction is only 46% for bottom tier players.

Another thing you can notice on that list – almost every card on it is a 2-for-1. Either making multiple bodies, drawing cards while providing a body, being a removal on a body – apart from Ephara's Dispersal which is just a very cheap interaction and Zephyr Singer which is just a rate monster with game-finishing potential. It is not that low win rate players don't rate those cards highly – they do, but they don't rate them highly enough. The cards on the list are even better than low win rate players think, while top win rate players will go out of their way to play them.

But as in the previous step – picking cards is only half of the draft portion. It is also important to know, which cards are making the deck more frequently in the hands of the top players.

And unlike last time, this time most of the list doesn't overlap. Yes, Eyes of Gitaxias and Atris, Oracle of Half-Truths are there in both cases, but this means that in case of good cards, some cards are picked less and played less mainly because they are not drafted, in case of other cards, they are not picked particularly low, but often are left out of the deck. Lots of the cards on the graph above are build-arounds or mini-build-arounds. Omen Hawker is a good example of a card that is really good but only if you make the deck revolve around it. And one of the best cards that go well with it are the Eyes of Gitaxias, a card lower win rate players value much lower than the top players.

Another mini-build-around is Tetsuko Umezawa, Fugitive, a card that plays extremely well in some versions of UW or UR decks, while being relatively low impact in others. Top players seem to pick the card at a similar rate as the lower win rate players but manage to navigate in such way that the card fits in their deck more often. Same goes for Transcendent Message– a card close to unplayable in some decks but very powerful in others. Top players seem to be more persistent to maximise its potential during the draft portion and reap the rewards, by being able to include it in their decks more frequently once picked. Other cards top players seem to be able to fit in their decks more are three color cards, like Zimone and Dina or Borborygmos and Fblthp. This is another important skill to learn – when to splash powerful multicolor bombs. And how to navigate the draft in a manner that enables doing so.

So how do picks and putting the card in a deck combine?

As you can see, there are no clear outliers. But there are several cards played ~2x more by the top players. The mentioned Omen Hawker and Eyes of Gitaxias are on the list. Marauding Dreadship has proven to be a card that overperformed in aggressive UR or WR builds but the card can look less appealing to less experienced players. Those types of cards are the ones you need to look for actively as the format matures if you suspect you might be missing out on some cards. Several cards on the list, like Seed of Hope, Flywheel Racer, Imoti, Celebrant of Bounty or Zimone and Dina are staples of multicolor decks – those are tricky to play for beginner players but looking at the data can help you visualising how to navigate those. And if you need extra help, this article should be useful.

As you can see with the good cards it is not as much not recognising the raw power but recognising the potential and circumstances of the card that make it good. But is there a trend in cards that are under- and over-used by the top players? I looked at the Game in Hand Win Rate of cards that were played than 50% less and more frequently by the top players based on the combined impact of pick and play rates and compared those.

As you can see, the cards overplayed by the top players are very high win rate ones, while the ones underplayed by them (or overplayed by low win rate players) are much lower win rate in comparison. So evaluation and card selection play a large role in the difference between top and lower win rate players. And unlike game play skills, the gaps in card evaluation are much easier to bridge. So if you want fast improvement – that is where you should invest your efforts. Look at the cards played by the best, if you don't prioritise them highly, think why they do. Maybe speculate on those cards more and try to see if they work in your builds. Identifying why there is a gap in your evaluation and the evaluation of more successful players will always be a level up in your limited game.

As you might have noticed in the last step, the difference between high win rate and lower win rate players is frequently having a good use for the cards. A clear vision of what can be achieved with it. Once you learn how to evaluate the good cards and avoid the bad ones, this is the aspect of the game you might need to focus on.

For each archetype you play you want to know essential cards. This requires diving a little deeper into the format. First thing to learn is the best color combinations. You shouldn't be forcing the best color combination, but should be aware of those. If you do and realise that playing, say WU, is worth it you will skew a bit more towards it because you know that the reward is worth it.

Same the other way – if you are starting your draft in one of the weaker combinations, but the cards don't seem to flow, it may be wise to keep a lookout for other options. Knowing what is worth fighting for and what is not will put you in favourable position more often than not. And you see it in data from top vs. lower win rate players when it comes to color pair selection.

You can see that top players draft UB and UW much more frequently than the lower win rate players. Guess, which were the two most winning archetypes across the format? This increase comes at drafting all the other color pairs just slightly less, but those little adjustments in preference convert to large differences in play rate of the top decks.

But aside from the positioning yourself favourably towards the metagame and having a plan for how to approach different colors and prioritise them, you will be well advise to know at least some color pairs inside out. Know what does your color pair want to achieve and which cards serve it best. I looked at the pick and play rates of cards but this time not in the whole format, but only in one color pair, WU.

On the graph above, you have 15 most played WU cards in the format. Most of them are picked roughly equally by top and lower win rate players. But 6 cards have a clear difference in pick rate. Preening Champion is valued by all players, but top players pick it higher than most uncommons and rares. As it should, given its win rate. Same applies to Ephara's Dispersal and again top players realise this and pick the card accordingly. But other cards overpicked by top players are not particularly great in a typical WU deck.

Eyes of Gitaxias, Saiba Cryptomancer or Meeting of Minds can be played in a WU Knights, but it is much better in a spell centric tempo shell with cards like Raff, Weatherlight Stalwart as enablers. And another thing those cards do well is play in other blue decks: UR or UB for example can have a great use of each of those cards. You are also actively cutting blue, increasing your chances of cutting someone off the color (although this is a minor effect). So how do the play rates look like?

First thing that is apparent is the large difference in playing Aerial Boost by top players. That is the card that clearly escaped the attention of lower win rates players. The card was picked at similar levels by both groups but ended up in top players decks disproportionately more. Aerial Boost was a key card for the archetype in this format and ended up winning many games, but lower win rate players passed that opportunity – something that could have been easily remediated by looking at win rates of cards in the WU archetype on 17Lands.com.

There are other cards that top players play more. Temporal Cleansing is an amazing piece of interaction that lets you create turns with lethal damage at low cost, or turn races in your favour. Tarkir Duneshaper opens different avenues of WU when you combine few of those with Halo Hopper for an aggressive, convoke based version of the deck that is not too reliant on knight synergies. But more importantly, almost every card on the list is played a bit more frequently by the top players. How can that be?

Well, that just means that cards from outside the 15 most played are played more by the lower win rate players. Cards that are fringe playables, cards that don't fit the WU plan very well. Weaker cards. Lower win rate players, either for lack of good card evaluation or because of shortcomings in draft navigation end up playing those less desirable cards in their decks, and that will surely hurt their win rate.

Top players’ decks are streamlined – they contain only the best cards for a given archetype. Lower win rate players decks tap into more cards. I looked at the number of cards that cross 1% play rate according to the above metric and only 24 commons cross that threshold in top player tier. But in the lower tier – 31 do so. This means many more low impact cards make it to the lower win rate players’ decks and having a large chunk of your deck being fringe playable will hurt the win rate independent of your play skill.

To avoid it, think of the game plan you want to impose on your opponent. WU in MOM is an aggressive archetype. And aggressive archetypes should be streamlined towards aggression. It is tempting to put a Cut Short in your deck, to have a response to the situation when you are on the back foot, but that is a bad idea in most of the cases. Your deck will rarely have a staying power in a game. Most of the times if you don't seal the deal in 7-8 turns your chances of winning diminish. It is a much better idea to put cards in your deck that forward the aggressive plan, if you fall back far enough to be casting Cut Short, you will probably not dig out of that hole. Aerial Boost, on the other hand, can do just that – if you brought your opponent low enough on life total to enable that one desperate flying attack. What is more important, Aerial Boost earlier in the game might put you in a position where you gain enough board advantage to win the game, while Cut Short will be a dead card in your hand for a large part of the game.

So in order to understand the plan – look at archetype specific win rates on 17Lands.com. Envision your version of the deck. Think about which cards you definitely want in and what cards you would want to avoid if possible. Make picks in the draft that fill the gaps in your strategy. Sometimes you need to pick a 2-drop over a better card, because you are short on creatures. Sometimes you need to pick a Temporal Cleansing over a good creature because interaction slot is where you feel a bit weak. Make sure you, at least while you are learning to do so, fill the gaps to build the most complete version of the deck. Time for improvising will come later when this element has been mastered.

There is no need to reinvent the wheel. Hundreds of players draft every day and their data contributes to the 17Lands.com database. And is a tool you can tap into. Early in the format I look through the trophy decks and try to learn which cards are over-performing. Just 15 mins of screening through the trophy decks collection of 17Lands will greatly increase your understanding of what works. If you don't understand a card, but see from the data that it is good – ask questions.

There are many Discords that share ideas about limited, including the mtgazone.com one – tag me there and I am sure you will get some answers to the questions you are looking for. The Gathering is usually thought of as a social aspect of the game but we also gather thoughts, interpretations of the game and the know-how associated with it – you will improve quickly if you tap in to this resource

If you want to learn a bit more about this topic, why not check the video version of the seminar on my YouTube channel?

Enjoy our content? Wish to support our work? Join our Premium community, get access to exclusive content, remove all advertisements, and more!

I am a limited player, who mainly skips playing in order to analyse the limited data using 17Lands.com. I run a podcast: Magic Numbers, where I try to use data to let you improve your limited game play, find out which heuristics work out and which common ideas are not well supported by data.

You must be logged in to post a comment.

No ads: Exclusive Content: all Support: Discord: Special offer: L95WR9JOWV