Once Upon a (Prime) Time

Editor’s note: this article has been updated to correct a mistake in the Neobrand numbers. Thanks Chris!

When Mark Rosewater previewed [[Once Upon a Time]], eyebrows went up. Free spells and efficient card selection both have a history of ending up on the Modern banned list. In particular, [[Once Upon a Time]] bears a striking resemblance to [[Ancient Stirrings]] – a card living on borrowed time.

Since then, I’ve run tens of thousands of simulations to put the hype to the test. My code (written in Python) is available on GitHub. You can also check out the human-readable explanation of how it works in my Valakut article. All values below apply to seven-card hands playing against a goldfish – no mulligans, no disruption.

Neobrand

Perhaps the most obvious (and concerning) application of [[Once Upon a Time]] is Neobrand, an all-in combo deck that can win on the first turn of the game. Losing to Neobrand is miserable, but luckily doesn’t happen often. The deck’s unreliability and vulnerability to disruption have kept it out of the main stream. [[Once Upon a Time:OUAT]] threatens to make Neobrand more consistent and resilient by improving its access to [[Allosaurus Rider]], [[Simian Spirit Guide]], and lands.

  T1 ≤T2 ≤T3
Neobrand 11% 38% 55%
… [[Serum Visions]] → [[Once Upon a Time]] 13% 40% 56%
… Other stuff1 → [[Ancient Stirrings]] 13% 42% 64%
… Other stuff → [[Once Upon a Time]] 14% 45% 67%

Odds to get a [[Griselbrand]] on the table each turn. Values are cumulative, so "≤T3" is the odds to do so on turn three or earlier. All values ±2%.

Judging from the numbers above, I suspect the concern is overblown. The difference between [[Once Upon a Time:OUAT]] and [[Serum Visions]] is not statistically significant, and it’s not clear that Neobrand has room for more cantrips. After all, it doesn’t play [[Ancient Stirrings]], and [[Once Upon a Time:OUAT]] is usually the same2. Cutting “other stuff” – [[Life Goes On]], [[Pact of Negation]], etc – makes us more likely to get [[Griselbrand]] on the table by turn three, but also increases our risk of imploding3 mid-combo.

Tron

In terms of assembling [[Urza’s Mine:Ur]][[Urza’s Power Plant:za]][[Urza’s Tower:tron]], [[Once Upon a Time]] outperforms4 both [[Ancient Stirrings]] and [[Sylvan Scrying]]. It’s tough to imagine cutting [[Ancient Stirrings]] – it finds [[Relic of Progenitus:much]] [[Karn Liberated:more]] [[Oblivion Stone:than]] [[Ugin, the Spirit Dragon:lands]] – but swapping out [[Sylvan Scrying]] for [[Once Upon a Time:OUAT]] would give the deck a boost.

  T3 ≤T4
Tron 16% 53%
… [[Ancient Stirrings]] → [[Once Upon a Time]] 18% 57%
… [[Sylvan Scrying]] → [[Once Upon a Time]] 21% 60%
… Other stuff → [[Once Upon a Time]] 22% 66%

Odds to have Tron by each turn. Values are cumulative, so "≤T4" is the odds to do so on turn four or earlier. All values ±2%.

It may seem strange that [[Once Upon a Time]] (which looks at five cards) performs better than [[Sylvan Scrying]] (which looks at the whole deck). It comes down to mana cost. We can afford to cast [[Expedition Map:Map]] or [[Sylvan Scrying:Scrying]] by turn two, but not both. With [[Once Upon a Time:OUAT]], we can cast multiple (non-[[Chromatic Star:egg]]) spells in search of turn-three Tron. That gives us a decent shot to assemble turn-three Tron even if there’s only a single land in our opening hand.

[[Once Upon a Time]] lets Tron mulligan a bit less often and assemble Tron a bit more consistently compared to [[Sylvan Scrying]]. On top of that, it adds value later on by increasing access to creatures like [[Ulamog, the Ceaseless Hunger:Ulamog]], [[Walking Ballista]], and [[Wurmcoil Engine]]. I suspect it’ll become a standard inclusion in Tron lists.

Valakut

Titan Shift is built for comfort, not for speed. [[Lightning Bolt:Removal]] and [[Obstinate Baloth:speed]] [[Courser of Kruphix:bumps]] are generally a better fit than cantrips. Titan Breach is another story. The deck plays only eight [[Primeval Titan:win]] [[Summoner’s Pact:conditions]], and regularly steals games with a timely [[Simian Spirit Guide]]. [[Oath of Nissa]] saw play in past builds, and [[Once Upon a Time]] is twice as good:

  T3 ≤T4
Titan Breach 15% 59%
… Other stuff → [[Desperate Ritual]] 35% 70%
… Other stuff → [[Oath of Nissa]] 22% 73%
… Other stuff → [[Once Upon a Time]] 29% 77%

Odds to get [[Primeval Titan]] on the table by each turn. Values are cumulative, so "≤T4" is the odds to do so on turn four or earlier. All values ±2%.

Before playing Titan Breach at the Pro Tour, I ran the numbers on dozens of different builds: cantrips, rituals, [[Hour of Promise:you]] [[Shefet Monitor:name]] [[Manamorphose:it]]. [[Once Upon a Time:OUAT]] is in a league of its own. It gives almost as much velocity as [[Desperate Ritual]]. But instead of making the deck a glass cannon, it adds resiliency. [[Once Upon a Time]] makes the deck goldfish faster while also increasing access to high-impact sideboard cards and finishers in the face of disruption. Titan Breach has been waiting years for a card like this.

Amulet Titan

Opening hands in Amulet Titan are often one card away from casting a turn-three [[Primeval Titan:Prime Time]]. They just need to find an [[Amulet of Vigor:Amulet]], or an [[Gemstone Mine:untapped land]], or a [[Simic Growth Chamber:bounce land]], or an [[Azusa, Lost but Seeking:Azusa]], or a [[Primeval Titan:Titan]]. [[Once Upon a Time]] finds everything but [[Amulet of Vigor:Amulet]], and it does so for free. In the first few turns of the game, it outperforms [[Ancient Stirrings]] (see below). Other cantrips like [[Explore]] and [[Oath of Nissa]] aren’t even close.

  T2 ≤T3 ≤T4
Amulet Titan 3% 27% 60%
… [[Ancient Stirrings]] → [[Once Upon a Time]] 5% 30% 64%
… Other stuff → [[Explore]] 6% 34% 67%
… Other stuff → [[Oath of Nissa]] 4% 33% 69%
… Other stuff → [[Once Upon a Time]] 7% 40% 75%
… Other stuff → [[Summer Bloom]] 14% 41% 69%

Odds to get [[Primeval Titan]] on the table by each turn. Values are cumulative, so "≤T4" is the odds to do so on turn four or earlier. All values ±2%.

Lists today can’t reliably produce a turn-three [[Primeval Titan]], so they play cards like [[Karn, the Great Creator:Karn]] or [[Golos, Tireless Pilgrim:Golos]] or [[Trinket Mage]] to hold down the fort until turn four. But with [[Once Upon a Time:OUAT]] and sensible mulligans, it’s reasonable to expect turn-three [[Primeval Titan:Titan]] more often than not. The last time we saw numbers like this, [[Summer Bloom]] got banned.

Happily Ever After?

[[Once Upon a Time]] is great in Tron and Amulet Titan – comparable in power level to [[Ancient Stirrings]]. It’s bonkers in Titan Breach, where it finds both [[Simian Spirit Guide]] and [[Primeval Titan]]. It may find a home in Neobrand as well. And that’s just the start. [[Once Upon a Time:OUAT]] is a powerful enabler for any strategy that depends on seeing certain creatures or lands in the first few turns of the game: [[Devoted Druid]], [[Eldrazi Temple]], [[Glistener Elf]], [[Slippery Bogle]], and so on.

Decks built around creatures and lands are ostensibly more “fair” than those using graveyards and the stack, but the decks that want [[Once Upon a Time]] aren’t looking to play fair. I would not be surprised to see this card push something over the line.

  1. “Other stuff” refers to anything not on the critical path to accomplishing the model’s goal. For example, the goal of the Valakut model is to get [[Primeval Titan]] on the table. Cards like [[Lightning Bolt]] and [[Obstinate Baloth]] don’t help with that. As far as the computer is concerned, they’re blanks. 

  2. [[Once Upon a Time]] and [[Ancient Stirrings]] are both green cards (for [[Allosaurus Rider]]), and will both usually find our next land drop. Neobrand doesn’t really spend mana until the turn it wins, so the difference between zero, one, and two mana is not very significant. 

  3. Once [[Griselbrand]] is on the table, the plan is to gain a bunch of life, draw our whole deck, and win with [[Laboratory Maniac]]. If we don’t draw [[Nourishing Shoal:lifegain]] [[Life Goes On:spells]] fast enough, we can run out of steam. 

  4. The model works by exhaustive search, which essentially means it has superhuman “instincts” about the order of the deck. To suppress non-human play patterns, choices between Urza lands are made alphabetically. If it’s already got [[Urza’s Tower:Tower]], it’ll always take [[Urza’s Mine:Mine]] over [[Urza’s Power Plant:Power Plant]] – even if it “knows” the card it’s about to draw is another [[Urza’s Mine:Mine]]. 

© Charles Fyfe 2019 under CC-BY
uno mas