Wheel of Fate Art by Kev Walker

How to Wheel in Drafts

Level up your Limited Magic drafting skill by learning about the key skill of wheeling, with Sierkovitz and the power of data.

One of the biggest level ups I have ever had in Limited was realising that my first pick can be worth more than one card. How, you might ask? The answer is wheeling. Wheeling a card happens when you open a pack, pick the card you find the most desirable and then when the pack comes back to you, you still find a card you are happy to pick from the whatever is left. If you pick smart, on occasion you can plan which card from the initial pool is most likely to wheel.

You pass it on, in Magic lingo – you float it, with hope of it coming back to you. While doing so, you pick a card that will work well with the card you are floating. If you are successful in doing so, you practically picked two cards from the pack you opened. This is a very cunning way of maximising the chance of ending up with a coherent deck full of useful playables. In some cases, the card you float can also generate an important signal in the draft that might inform your later picks.

By no means I came to this conclusion on my own. I listened to Ryan Saxe explaining how picking one card from a particular pack means you are extremely likely to wheel another one and how being almost certain you will wheel the second card makes your first pick not between card A and card B, but between card A and card B + card C. And this can change the equation by a lot. Imagine if card A is marginally better than card B – is it still better than cards B and C together? Is it still better than card B and a 50% chance of getting card C? Those choices are not easy and require some practice, but can improve your win rates and in general, make you a better drafter.

How to know what will wheel?

Wheel of Fate Art by Kev Walker
Wheel of Fate Art by Kev Walker

That all sounds nice, you might say, but how do I know which cards will wheel? This is where 17lands come in handy. Using the draft data I calculated how often each card wheels. The first thing you may see is the probability of wheeling a card decreases the later you see it. Let’s take a look at one of the best commons in the set, Blessed Hippogriff. It is taken relatively early in the draft. Its Average Last Seen At (ALSA) from 17Lands is at 3.37. This means that you will see it on average at pick 3-4 of the pack, but of course in some cases it will be slightly later.

The distribution of how late you see a card is pretty broad. Sometimes you will open a card in your pack, and it will not wheel. The ALSA statistic for that pack will be 1, but there is no way of knowing when was the card actually picked. This means that ALSA is underestimating when a card goes in the pack in most cases. Exceptions will be on the extreme end of the spectrum: cards with ALSA of ~1 will be first picked from the packs, so you will see those cards almost exclusively when you open them. Hippogriff is not one of those cards, and even if it is picked early, in some pods they might wheel. Most likely those are the pods with few people drafting white, and even then, there has to be an even better white card in the pack. In this case it will be either a strong rare or a powerful uncommon.

Still, if you open a Hippogriff, you will see it wheel ~10% of the time. Mind – this includes data from early format, I am pretty sure by now it wheels less frequently. If you see Hippogriff in your pick 2, that number goes down to 7%. And the number decreases as you see the Griff later in the pack. 10% of wheeling is not a zero, but 90% of the time it will not wheel, which means you should not count on it coming back to you, but be very happy if it does. Or, in most cases, just pick it early if you are in white.

Floating the signpost

The later you see the card on average, the higher the chance of wheeling it. Take Krydle of Baldur's Gate. Its ALSA is 5.36, meaning you see it 2 picks later than the Griff on average. And this changes the chances of wheeling dramatically. Forget 1 in 10 – you will wheel Krydle 1 in 3 times. Now that is already a number showing some promise. Krydle is also a special case. It is a multicolor card and is in a not so popular color combination. Cards like that are excellent targets for floating. In the first pack you can float them to get a valuable signal. Say you first pick a strong black card, like Cast Down and send Krydle for a journey around the table. If it wheels, you have a strong signal that there might be no other UB players in your pod. Strong signal is not a guarantee, but it is information you can act on. You also get a relatively strong card late, while if you picked Cast Down, you have almost zero chance of wheeling it. If it doesn’t wheel, you can think about other cards in that pack and decide – is it likely there is another Dimir drafter on the pod, or did someone pick it for vault progress or as a speculative pick.

The situation changes in later packs. Multicolor cards in regular sets go much later in pack 3 than they did in packs 1 and 2. In pack 3 many players will focus on getting effects they are short on. If you have few 1-2 drops, you may desperately picking them even if they are not the best cards in the pack. Same goes for mana fixing. Dual lands, for example, are picked much higher in pack 3 than in pack 1 in formats where they are present. On the other hand, multicolor cards go later in pack 3. Most people are already deep in their colors and they will not be tempted in an off-color multicolor card, especially if the card in question is not a bomb, which is the case with Krydle.

If you know this and end up in a draft where you are quite convinced there is no other UB drafter, you can pick a monocolored card first and have a reasonable chance of getting Krydle back. This is a valid strategy if the multicolor is not the decisively strongest card you can pick for your deck from the pack. It is a no-brainer to float Krydle-like cards when they are not the strongest card you can pick for your deck. The situation becomes slightly more complicated if Krydle is the best card to pick by a small margin. You have a decision to make then: pick or float. I would err on the side of floating in such scenarios – the upside of wheeling the card is a real one. But that is only a rough heuristic. You should take into account how many good cards are in the pack. The more there are, the more likely you can float successfully. Or how likely is it to have another player drafting the same color pair. In case of UB in HBG it is pretty unlikely, but if we were looking at Trelasarra, Moon Dancer in WG, there is a good chance there is another WG drafter at the pod, making the wheel less likely.

Double picking

The later card goes, the higher probability of wheeling, and at some point, you arrive at numbers where you can assume to wheel a floated card. One such card is the Earth-Cult Elemental. a 6 mana 6/6 creature is something that is definitely not a priority for a deck, but the card has good stats. It performs particularly well in RB decks that can ramp it out early with Treasure tokens.

It is a top 10 common card for those decks and wheels 70% of the time if you open it. Now that is an information you can act on. If you open a good card for the archetype, like a Skullport Merchant, you can reliably wheel the Elemental making a Skullport Merchant a double pick in many cases, getting both the merchant and wheeling Earth-Cult which is actively good in your potential RB deck.

This has another very important merit. Say we are not talking about the elemental, but about a niche card that is essential for a successful build in your estimate. You really want a copy of those, but frequently you don’t want to dedicate early picks to get it. In early picks you want to focus on cards you are unlikely to see later. Successfully floating your niche card gives you an opportunity to pick your niche important card late and therefore not conflicting with your powerful picks, while providing you with a welcome and important functionality.

Key wheelers

There are many cards that wheel, but as you can imagine, many of them have a good reason to wheel. They are not good. And you would be right, most cards that wheel reliably have low win rates. Limited players nowadays are pretty good in evaluating cards and even if a strong card was initially misevaluated, quickly the community learns to value them highly. But every format few cards fall through the cracks.

I collected 15 cards in data that wheels more than 30% of the time with the highest win rate. As you can see half of them are white, most likely related to white being underdrafted in the early days of the format. But even now you still get to wheel the Steadfast Unicorn reliably, despite it being one of the top commons. The cards you see on the list are not only undervalued white cards. Another large category is combat tricks. You hear something on watch, Arcane Archery and Valiant Farewell fall into this category, and to some extent Deadly Dispute can be treated as a trick, when you sacrifice a chump blocker to get extra value. Another small category is expensive large creatures, of which we see Hill Giant Herdgorger and the mentioned Earth-Cult Elemental.

Those two categories have something in common. Both tricks and large creatures are types of cards you usually don’t want to have too many in your deck. More 6 drops risks you having unplayable opening hands with 2-3 lands and 3-4 6-drops forcing you to mulligan more often. Having too many tricks leads to situation where you have 2 tricks in your hand and zero creatures to boost, a scenario where you are very likely to lose. But few copies of those are still valuable and knowing you can wheel them can make you prioritise more important elements of your deck without worrying too much about not getting them.

Top 15 cards that wheel >30% of the time with highest Game in Hand Win Rate from 17Lands

Wheeling beyond the current draft format

OK, you might say, I don’t have the time to carefully calculate odds of wheeling any time I want to use the knowledge gained through that in draft. Anyway Alchemy Horizons: Baldurs Gate (HBG) draft season is soon ending – memorizing a list of cards that wheel seems like right waste of my precious time and brain capacity. I hear you. So why not use a more general and universal way of learning those numbers? A method easy to apply across multiple formats. This is exactly why I looked at the link between ALSA and wheeling probability. Is there a link between those two values? Sure there is. What it means for you is you can skip all the painstaking calculations I made and simply look at the ALSA values of the cards you chose between and based on that make a decision to float something.

In the graph you have all the HBG cards, and their ALSA plotted against the probability of wheeling. You can see that there is some variance, but that is worth not going through the painstaking process of calculating wheeling probabilities in the future. For example, you can assume that you will wheel your ALSA 4 cards ~15% of the time. With ALSA 5 it goes up to ~30%, with ALSA 6, ~50% of the time. The rough estimate will be good enough to give you sufficient information.

And you can use this information to go deeper than my analysis. I mentioned some cards wheel less at this stage of the format than in the beginning. Using this graph you can estimate it. On 17Lands.com you just need to play with the time ranges for the data and look at last week only and see the current ALSA of a card like the Blessed Hippogriff. It was 3.37 at the time I did the analysis, but now 2.87. That means nowadays it is much less likely to wheel – more 3% than 10%. Even more importantly, you can use this graph for the future and past sets. Bored with HBG and playing VOW instead? You can easily apply this graph to VOW and estimate that you will wheel Heron of Hope (Last week’s ALSA 4.85) roughly 25% of the time.

The data above is only for P1/P9, so for wheeling cards you open. But the same can be done for all the other picks. You will see a universal decrease in the wheel probability but the results are consistent enough to still predict the wheel probabilities.


Hopefully the tools I provide here will help you play a bit with wheeling in your drafts. But be wary of a trap of overusing it, so common with level-ups. Keep in mind that some cards will have a slightly different distribution of wheeling. For example, due to Arena economy, rares that go late will not wheel in the same way as commons. Okay, this might be a completely unplayable rare but I am not going to pass on 20 free gems. In the future there may be distortions of those graphs if the booster pack anatomy changes from set to set (e.g. set has 15 cards in a pack rather than 14).

If you enjoyed this article, you might want to check out my seminar on the topic where aside of wheeling I also talk about how do high win rate players pick cards differently than lower win rate players and what lessons can we take home from that.

Card Kingdom - Double Masters 2022
Default image

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.

Articles: 4

Leave a Reply