Fox vs Marth: Melee matchup win rate
Fox wins 49.2% of games against Marth between equally skilled players, based on 12,103 tournament games from 2002 to 2026 (90% interval 48.2 to 50.4%). The matchup is close to even: the interval includes 50%. In 2026 alone the estimate is 47.9% from 640 games, and that interval includes 50%.
Fox vs Marth by year
| Period | Fox win rate | 90% interval | Games | Raw record | Raw win rate |
|---|---|---|---|---|---|
| 2026 | 47.9% | 44.5 to 50.9% | 640 | 300-340 | 46.9% |
| 2025 | 51.0% | 49.9 to 52.1% | 1,075 | 525-550 | 48.8% |
| 2024 | 49.7% | 48.7 to 50.6% | 858 | 393-465 | 45.8% |
| 2023 | 43.9% | 40.8 to 47.0% | 715 | 339-376 | 47.4% |
| 2022 | 48.4% | 46.4 to 50.1% | 1,256 | 565-691 | 45.0% |
| 2021 | 50.6% | 49.8 to 51.3% | 432 | 224-208 | 51.9% |
| 2020 | 49.6% | 48.8 to 50.4% | 307 | 139-168 | 45.3% |
| 2019 | 46.0% | 44.0 to 48.1% | 1,250 | 626-624 | 50.1% |
| 2018 | 47.7% | 44.9 to 50.4% | 1,275 | 602-673 | 47.2% |
| 2017 | 50.4% | 49.2 to 51.5% | 1,460 | 738-722 | 50.5% |
| 2016 | 49.6% | 48.4 to 50.5% | 1,076 | 545-531 | 50.7% |
| 2015 | 50.3% | 45.5 to 54.4% | 689 | 372-317 | 54.0% |
| 2014 | 48.9% | 48.1 to 49.6% | 354 | 160-194 | 45.2% |
| 2009–2013 | 50.6% | 49.4 to 51.6% | 616 | 307-309 | 49.8% |
| All time | 49.2% | 48.2 to 50.4% | 12,103 | 5,884-6,219 | 48.6% |
The raw record counts every game, so it also reflects who plays each character. The win rate removes that by fitting each player's skill overall, by year and on each character. Single years have fewer games, so their estimates sit closer to 50%.
Fox vs Marth stages
Fox's best stage against Marth is Pokémon Stadium (50.6%) and its worst is Final Destination (42.2%).
| Stage | Fox win rate | 90% interval | Games | Raw record |
|---|---|---|---|---|
| Pokémon Stadium (counterpick) | 50.6% | 48.3 to 52.5% | 676 | 363-313 |
| Yoshi's Story | 49.3% | 46.6 to 51.6% | 880 | 382-498 |
| Dream Land | 49.2% | 47.1 to 51.4% | 882 | 392-490 |
| Fountain of Dreams | 49.2% | 47.0 to 52.0% | 408 | 202-206 |
| Battlefield | 49.0% | 47.0 to 51.0% | 1,275 | 623-652 |
| Final Destination | 42.2% | 40.1 to 44.2% | 1,103 | 516-587 |
Game 1
- Fox strikes first: Fox strikes Battlefield, Marth strikes Yoshi's Story and Dream Land, Fox strikes Final Destination, with no clear last stage (Fountain of Dreams in 23% of resamples).
- Marth strikes first: Marth strikes Yoshi's Story, Fox strikes Battlefield and Final Destination, Marth strikes Dream Land, with no clear last stage (Fountain of Dreams in 23% of resamples).
Best of 3
- After Fox loses a game: Marth bans Pokémon Stadium (61% of resamples), then Fox has no clear pick (Yoshi's Story in 21% of resamples).
- After Marth loses a game: Fox bans Final Destination (100% of resamples), then Marth has no clear pick (Battlefield in 33% of resamples).
Best of 5
- After Fox loses a game: Fox picks Pokémon Stadium (61% of resamples); once Fox has won there, no clear next pick (Yoshi's Story in 21% of resamples).
- After Marth loses a game: Marth picks Final Destination (100% of resamples); once Marth has won there, no clear next pick (Battlefield in 33% of resamples).
With best play on both sides, Fox wins 48.7% of best-of-3 sets (90% interval 46.0 to 50.7%) and 45.5% of best-of-5 sets (42.7 to 48.3%).
Ruleset and method
Game 1: the five starters are struck 1-2-2-1 until one is left. The winner of rock-paper-scissors chooses whether to strike first.
Best of 3: before each later game, the winner of the last game bans one of the six stages, then the loser picks from the rest, never a stage they already won on.
Best of 5: no bans. The loser of the last game picks any of the six stages they have not won on.
Stage win rates are fitted on the 83,758 games since 2022 with a recorded legal stage, with player skill removed the same way; Pokémon Stadium includes Frozen Pokémon Stadium. The plan is the best play for both players by backward induction over the whole set, under the Platfight Melee ruleset (July 2026), assuming both players keep these characters. The percentage after a choice is how often the same choice comes out when the fit is redone on 200 resamples of the sets. Method.