Sets

Mystery Booster Commander forecast

Last updated · Scryfall data of Oct 25, 2026 · locked

Predicted Commander popularity for cards 1–60 of Mystery Booster Commander Edition, which become Commander-legal November 13, 2026. Playtest and acorn-stamped cards are excluded.

Browse these cards in search

On past sets this model's scores had a 0.52 rank correlation with the actual popularity for commanders and 0.62 for all cards (1 is a perfect order, 0 is no relation).

Method

Manayard parses mechanics from known Commander legal cardsThis part is changing frequently but currently includes most mechanics, mana cost, colors, types, current popularity
Text modelWords and phrases learned from older cards
Pairwise modelLearns which of two same-release cards got played more, using facts from Manayard's card parser.
Blended average
Apply to new set

For more details see How the tests were run

Power

Power is a card's expected strength in a game, learned from older cards and separate from popularity. Three models see only what is known when a card is revealed: its rules text, type, mana value, colours and a parsed-effect score (lib/card_power.js).

Two read 17lands limited win rates of cards printed before the set; the third reads Cube Cobra draft Elo, which is how often drafters pick a card over others. A cheap language model (Haiku) also rates each card from its text alone, 0 to 10. The percentile from the three models and the percentile of that rating are averaged, so 0.50 is a typical older card.

Trained before 2024 and tested on 3,909 cards from 2024 on, the three models correlated 0.36 with win rate (mana value alone: −0.03), and adding the language model's rating raised it to 0.40 on a 500-card sample. The panel under a card lists the rules-text terms that moved the text model most.

Results

Each category is ranked separately. Model scores are percentiles among older cards, from 0 to 1. Power is separate from popularity (see Method). Select the Blend or Power heading to sort. Select a card to see its image and what raised or lowered its scores.

Commander candidates 59 cards

#CardSetText modelPairwise modelBookmark
1Maular, the Next EvolutionMBC0.990.830.990.99
2Balefang the UnslayableMBC0.970.820.960.98
3Homer, the HermitMBC0.970.280.960.98
4Olag and Miau, New FriendsMBC0.970.510.960.98
5Davvol, Evincar of RathMBC0.970.280.980.96
6Meatsqueak, Hoard LordMBC0.960.600.950.98
7Chief Magistrate of MercadiaMBC0.960.940.930.99
8Flitwing, Lyev DetectiveMBC0.960.350.950.97
9Tresserhorn's Lord, ReturnedMBC0.950.410.950.96
10Tolabow, Loch RascalMBC0.950.660.980.93
11The Unluckiest PlaneswalkerMBC0.950.460.950.95
12Seluma, Light of AysenMBC0.950.760.950.95
13Dyfed, the Guiding HandMBC0.950.550.950.95
14Sifa GrentMBC0.940.780.940.94
15Grizzlegom, Hurloon HeroMBC0.940.830.920.96
16Selenia, the Cursed HeartMBC0.940.660.920.96
17The EverforgerMBC0.930.360.950.92
18Miss HighwaterMBC0.930.780.910.94
19Venser, Visionary TravelerMBC0.920.880.930.91
20Boss Uramon, Shadow's ReachMBC0.920.620.860.98
21The Weaver KingMBC0.920.740.890.95
22Grandmother GobyMBC0.920.650.900.94
23Nephilim EpochalMBC0.910.570.890.93
24Blor the ImperviousMBC0.910.570.910.91
25Uugguu, the OmniplasmMBC0.890.280.900.88
26Ashaya's Enduring BondMBC0.880.470.930.84
27Valko IndorianMBC0.870.340.930.82
28Worzel, the ProtectorMBC0.870.910.820.91
29Yume, Chronicler of ValorMBC0.860.360.900.82
30Ekthi, Contaminator PriestMBC0.860.560.850.86
31Autumn Willow, HarmonyMBC0.850.670.830.87
32Massimo, the MagicianMBC0.850.400.850.85
33Feroz, Ulgrotha's WardenMBC0.850.880.800.90
34Sahir, Visitor in DarknessMBC0.850.690.880.81
35Rikala, Homarid KingMBC0.850.580.820.87
36Jabs, Mistress of MockeryMBC0.840.210.930.76
37The Madcap JesterMBC0.840.700.910.77
38Olinda the ObliviousMBC0.840.740.870.81
39Gustha EbbasdotterMBC0.830.510.780.87
40Lyna, Veil of VengeanceMBC0.830.410.720.93
41Zagorka, Mother of SanctumMBC0.820.450.810.83
42Chira, All InMBC0.800.280.900.70
43Itazura, Lingering WickMBC0.780.700.780.78
44Grakk the PacifistMBC0.760.380.820.71
45Glava, Five-Advents MageMBC0.740.250.790.70
46Thomil, the DestroyerMBC0.740.940.740.73
47ArzakonMBC0.720.800.680.76
48GreensleevesMBC0.720.900.720.71
49Tsagan, Raider WarlordMBC0.710.430.850.57
50The Muckslinger GangMBC0.710.670.630.78
51Kuroki, Thief of TalentsMBC0.680.930.720.65
52Sash and Waistcoat, UnmenMBC0.680.650.660.69
53Whtz, the BibliophileMBC0.640.470.650.64
54Joven and ChandlerMBC0.630.700.710.56
55Jandor, Fortuned TravelerMBC0.620.570.770.47
56Elda, Conjurer of SpectacleMBC0.590.250.630.55
57Nivea, Beloved BattlemageMBC0.510.690.660.36
58Matoc, LavamancerMBC0.460.830.540.37
59Istvan, Butcher of ElnMBC0.420.530.480.37

Other spells 1 card

#CardSetText modelPairwise modelBookmark
1Dragon's SmileMBC0.980.350.960.99

Limits

27 of 60 cards still contain text the parser cannot fully interpret. Popularity scores predict how often a card is played, not how strong it is, and are not probabilities. Power is learned from limited win rates and draft picks, so it measures strength in limited play, not in Commander, and ignores synergy with a specific deck. On 3,909 cards from 2024 on, the models had a 0.36 rank correlation with win rate (0.40 with the language-model rating, on 500 cards); most of the spread in win rate is not predictable from card text. Popularity scores will be graded against real popularity after release.