CubeCon 2023: A Cube Owner’s Retrospective
CubeCon has been in the books for a month, but it wasn’t until recently that cube owners received a wealth of data in the form of draft pools:
As part of the security process of the main event, immediately after a draft, each drafter had a picture taken of their pool, uploaded to the webapp Hedron Network, and associated with their name/draft seat. After 3 rounds, the drafter had another picture taken of their pool and uploaded to the “Check Out” section, cards turned back in, and was free to go. The coolest part of this was the organizers being able to anonymize the photos, associate the record of the player with the pool, and share the photos with the community. In this way, the Cube owners could then begin to loosely tie their cards with results.
Now as Dan Schneider has rightly pointed out, this data is difficult to rely on heavily for actually making changes to a cube. @mtg_ds also highlights a number of difficulties that we just can’t account for. I see the pool, I often don’t see the decklist. I don’t know the skill of the player or their opponents. Magic is already a game of variance, how many of the wins that I am shown come from the opponent keeping a bad hand and losing to mana issues as much as to the other deck? All this to say, I am taking these pools, analyzing them, and trying to take very broad conclusions from them where possible. I still think that the analysis itself may be worthwhile, or at the very least, interesting.
I also had the almost-unique honor of having 3 cubes in the main event. For each, I will approach data analysis differently: the Vehicle Cube is already a silly environment primed for change with the release of more Vehicles each set, and the Museum of Modern can only be changed so much and remain true to its time-capsule goals. The Amonkar Desert actually has the most usable data and the greatest need for change, which I will elaborate on in its section. For now, let’s get started with the least patient of the cu-
VEHICLE CUBE: EIGANJO DRIFT
THIS CUBE IS ABOUT VEHICLES! SO given that, we’re not going to focus on the draft pools, deck lists, or archetypes so much as how individual Vehicles performed. We’ll do a little deck-color analysis, but we know who the stars of this car show are. Everyone starting with a Mech Hangar means the vast majority of Vehicles can be played in any deck, so comparing them directly is mostly fair.
Some context for the data I’m about to cover: Eiganjo Drift was drafted four times in the main event by a total of 30 players. I’ve associated the match points earned each draft (3 for each win, 1 for each draw) with each of the 65 unique Vehicles, then combined and averaged across the number of drafts they were played in. In cases where the player uploaded a decklist, I removed their points from the Vehicles they drafted but didn’t play, in the hope of making the data as accurate as possible. At the time of these drafts, I broke singleton for 9 Vehicles. With the release of Lost Caverns of Ixalan (and its many wonderful pirate ships), that number will be 6, and before long there should be enough Vehicles I can have just 1 copy of each.
So which Vehicle came in first place? There are many ways to view success here, but High-Speed Hoverbike made a pretty compelling argument for the gold medal. It is one of the 2x copy vehicles mentioned above, and it managed to appear with both copies in TWO 3-0 draft pools, and never appeared in an 0-3 pool. The silver medal goes to Flywheel Racer, as every pool with this manabike went 2-1 or better. Bronze, and the best-performing rare, was the mashup Vehicle/Equipment Unicycle (thanks again Deinonychus for the recommendation.) An honorable mention, and 2nd place in raw points thanks to its 2nd copy, was Bomat Bazaar Barge. At least one of the decklists I received cut High-Speed Hoverbike, making the Barge the best-performing card with 8 copies in pools across the weekend.
Bringing up the rear, we have some worst-in-show worth mentioning. Daredevil Dragster had the worst performance of a 2x card, no pool with it did better than 1-2. Peacewalker Colossus had the worst performance of a rare, just barely doing better than Goliath Truck, which sported the worst record overall, appearing in the pools of two 1-2s, and two 0-3s. An additional note on Aradara Express, it was one of the cards which players with a decklist cut, and so it was played only two times maximum and still had a poor showing in records. You can only have so much top-end, and there are now many good options for us at that higher mana value.
And finally, a little color commentary. With only 4 colors available, each achieved something unique: Decks which played White had the lowest winrate, Blue was the least played overall, Black had the best winrate, and Red was the most played overall. There was only one monocolor deck (which went 1-2) and only one 4-color deck (which went 2-1). Of the 2–3 color decks, Rakdos was the most successful, and tied for most popular with Boros, and Grixis had the worst showing as a color combination. The colors were reasonably balanced in play and win rate, especially at the very top and bottom among 3–0s and 0–3s.
I won’t be making any immediate changes to this cube as a result of the CubeCon draft data. A few player impressions I’ve heard included that Prowl and Surgehacker Mech are power outliers, and while Prowl is indeed in the top 10, Surgehacker garnered only 3.25 average points across these 4 drafts. My current belief is that only Esika’s Chariot and Shorikai are too strong for the cube (not to mention the fact Green can’t drive.) Every Vehicle that I include, and even the best Pilots like Kotori, Greasefang, and Depala, may have explosive moments on-board, but they are within the bounds of my design goals and should have enough answers to not run away with every game they resolve in.
A few Vehicles are beginning to outclass each other enough that I’ll consider cutting them entirely once the raw quantity of options is high enough. I did make 19 total changes with the release of LCI, but only duplicate Vehicle copies were cut: Bomat Bazaar Barge and Brute Suit for being at a crowded mana value with minor effect, Daredevil Dragster for the performance noted above. By the time the Death Race set rolls around, we may be in need of major restructuring, but until then we’ll keep driftin’!
The Museum of Modern
Our second cube seeks to encapsulate the Modern format as it was at various points from 2012–2018, with Twin legal, Tarmogoyf as king, and Birthing Pod waiting. This time, the analysis will focus on specific deck archetypes as well as broad overall strategies like Aggro, Combo, Control, and Midrange. As before, this will be followed by a little color breakdown.
The Museum was drafted all six times in full pods for a total of 48 players. This time I’ve associated the match points earned each draft with one of 11 deck archetypes, then (as before) combined their points and averaged across the number of drafts they were played in. In cases where the player uploaded a decklist, I was able to get the archetype specifically but in a few cases was forced to guess. Here’s how that data turned out:
Burn was the most successful deck followed closely by both Naya Zoo and Twin Combo, Zoo and Twin with the more statistically significant successes due to their 4 additional plays. Zoo and Twin were also the most popular decks, provided we count various color combinations of Control as different decks. Death’s Shadow had the worst showing, but with only 2 decks and such a risk-inherent win condition, this is forgivable. Another potential lesson comes from the Misc Aggro decks averaging less than a single match win: stick to established aggro decks like Humans or Zoo, or risk being outclassed by more consistent and supported game plans. This isn’t something I hope to discourage, but if a drafter asks me, I like having this sort of data to let them know the historic risks to breaking the mold.
While white was the most popular color to include in decks, Green was the color to deliver the best winrate at 57%. Naya was the most popular color scheme, coming together in the form of 6 Zoo decks plus a Burn deck, cementing Aggro game plans as known effective answers to decks attempting to assemble combos or control the board with Limited resources. Dimir and Simic decks not materializing was entirely expected, given the era and that our fixing is good enough, you should usually just go into a third color from those pairs. The color hit hardest here is definitely Black, with the worst winrate and every single 0-3 list containing the color. I will be trying to give it a few more tools as part of the changes resulting from the data and player feedback:
- Apostle’s Blessing can support Death’s Shadow, and a second Resto wasn’t really necessary given White’s dominance, and Felidar Guardian.
- We have another Spell Pierce and other spells available to counter noncreature spells or protect a combo, and Gearhulk will help round out a “Titan” for each color.
- No graveyard deck came together in all 48 pools this year, and I’m not sure I’ve ever seen one that makes the Spellbomb appealing; meanwhile Grave Titan can go in Pod or be a top end in Tokens or Control.
- Burn did so well it doesn’t need the second Rift Bolt, but the player-suggested Abbot could also be useful in Humans or Death’s Shadow.
- Become Immense works well with Death’s Shadow and in Zoo decks, and as well in graveyard decks as the second Satyr would.
- Since we have self-mill, Faithless Looting, Nahiri, and Titans, a value reanimator strategy could be cool with Unburial Rites, leaving one copy of Voice as a more anti-Blue sideboard piece.
- Lastly and sadly, Cruel Ultimatum simply isn’t viable at this power level, but there are shenanigans aplenty to be had with an additional copy of Ghost Quarter, as requested by PleasantKenobi.
Seven changes out of 480 cards isn’t much, but I’m mostly happy with where the cube is, and we do have to be careful when curating a Museum.
The Amonkar Desert
Our third and final cube is my primary project, and the only one to have been at the previous CubeCon. As mentioned in the previous sections, I received draft pools for every player, but only deck lists for a few. However, we get a special advantage in a Desert Cube: seeing all of the players lands! For the uninitiated, check out this article which provides a history of Deserts, but what you need to know is there’s no basic land station! So, for the data covered in this section, I was able to identify the colors a player was running, whether they could splash (usually for a Gold card), and up to 27 of the nonland cards I believe they were most likely to have included in their maindeck. This is also the only cube mentioned today that is not drafted 100% each time: players draft 3 packs of 18 for a total of 432 of the 480 cards, so some cards missing won’t be because they ended up in a sideboard, but stayed in the box all along.
Another note which will influence much of the data I am about to cover: Amonkar was drafted all six times by 47 total drafters, which resulted in a nice 69 matches. Seven of those matches ended in a draw, with some number more going to time and resulting in a winner within the 5-turn overtime. Over 10% of matches going to time is not acceptable, so we’re going to do our best to find solutions to that.
I was a judge both years, and I don’t recall hearing last year that my “Desert cube has gone to time, again”. So is something that changed between the two events the culprit? The first thing that comes to mind: I changed the GW archetype from +1/+1 counters to tokens, and recently added a lot of Food support from LTR as well. Both creature tokens and Food do a great job at prolonging games. Let’s look at some color data first this time:
Lots to take in, but first on the topic of draws: On the left side tables we see Black had the most in raw quantity, but as a proportion of play Green drew the most frequently. We also see Red was the color to be when looking to maximize the chance of winning, and White with the highest loss rate. Despite the draws, the bottom-right table shows that Green did have the highest play:points ratio. And finally, Blue managed to hold its own in winrate while being the least played color.
At the top-right, we see there were as many tricolor 3-0s as color pair 3-0s, although the vast majority of 1-2 records played 3 colors, perhaps reflecting the unreliable nature of mana fixing in a Desert. Boros had the best showing of a particular color pair, and Mardu and Sultai were the most popular tricolor decks. Overall, the colors were fairly balanced in plays:points ratio, with Blue just a touch underplayed.
Gwen’s list above is pretty typical of the GW deck, it has 15 cards that produce tokens, 10 cards that gain life, yet it never drew a game. Neither did the one other Green-White deck, so it seems that change from last year isn’t likely to be the issue. We’ve further narrowed the likely colors to Black and Green, so lets look at potential cards responsible for all those draws. Similar to Eiganjo Drift above, I assigned match point values to every card, then averaged them together across the 6 drafts, which also allows me to see which cards were in decks with multiple draws. 69 cards met this criteria, and among them 19 were Black, 16 were Green, and 4 were Black-Green. That’s right, all 4 of the 4 Golgari cards.
Of these 4, Old Rutstein was the least criminal, having been in 2 decks with draws out of 5 total. Choking Miasma was in 2 out of 4. But Uurg, Spawn of Turg and Spider Spawning, hoo boy: 3 draws each, Uurg out of 4 total decks and Spawning out of 5. Uurg feels like an easy cut, everything about it screams to slow down the game, but Spawning hurts. A popular limited archetype historically, but one that undeniably makes the game ending much harder on resolution, unless the Spawning player is already ahead.
So the data is helping me to identify some cards that need to either be given another chance, or be cut and replaced with similar cards that lean a more game-ending direction. Luckily for me, Lost Caverns of Ixilan has a very on-theme archetype in the form of Descend and I believe Akawalli, the Seething Tower replacing Uurg is the first of many swaps in the right direction. I’ll detail all the changes coming to this environment at the end of this section, but for now lets continue looking at outlier cards!
Cards like Forbidden Alchemy aren’t the most compelling here, only appearing in one deck and going 2-0-1 isn’t bad but it also tells us nothing about its consistency. Conversely, I believe this shows Alharu to be the real winner of the weekend, appearing in all six drafts and present in a White deck that went 2-1 or better every single time. Blossoming Tortoise and Lord Skitter also had great showings hot off the presses of Wilds of Eldraine. Meanwhile, at the bottom of the scorecard:
Dying to Serve was the most consistently poor-performing card, with a meager 10 match points total across 6 draft pools in Black, but a special shout-out goes to Tiller Engine, averaging just a single match point per draft. Two other fun card-specific facts: Arcane Proxy was likely the least-played card in the entire cube, drafted exclusively in pools without much or any Blue mana. And for those fans of Regular old Magic cards, Night Clubber was the most average card, with 6 plays and a 4.5 avg point value (But don’t worry Anthony, Scrapwork Mutt was very close with 4.67)
Now for the hard choices. After reviewing the LCI spoilers I knew I wanted to bring in around 25 cards, and even when the swaps are logical they can still be difficult. Confessor for example is coming out for Miner’s Guidewing. The bird is a way better 1-drop White Common and Confessor had a pretty bad showing at CubeCon in table above… but he’s been in the cube since its earliest iteration. So it goes for many of these changes:
The bulk of these changes can be described as cutting poor performers or cards with very defensive effects, and removing the ability for many cards to produce creature tokens repeatedly. A few card-specific notes: Astrid Peth from WHO is one of the most exciting additions in recent memory, tying together WG Food, WR Aggro, and even has utility with WB life and WU flicker. Stonerise Spirit hopes to help bolster the minimal evasion in the format, as I hate when a huge flier ends a game by itself because the enemy just has no removal, but this has a small evasive body and an on-theme way to temporarily grant others that evasion. Trail of Crumbs was a suggestion by Sam Black to help the Food archetype do more than just gain life. Lastly, Balmor turns a durdly flying token generator into a way for Wizards to take the aggressive position early.
I don’t want to remove Spider Spawning just yet, but it’s certainly on notice. Some of the high performers like Irencrag Pyromancer and Lord Skitter will also remain on my radar if they prove oppressive over time. This environment to me should still feel like a puzzle with many different lines of play as the game goes on, but not one that has so many options and game objects that it stalls out entirely.
In closing thanks so much for reading! This is as much for my own personal reference as anyone else’s, but for those who make it this far I’d love to hear feedback. Should I have analyzed the data another way? Do you want to see MORE of the data? These are all segments of a bigger file, but I’ve tried to cut out most of the noise. My next drafts of these environments will be the most interesting, as I have the additional perspective of CubeCon to know what to watch out for. And with any luck, I’ll be able to do this all again after next year’s event. I hope to see you there!
