Berserker's Caitlyn Punishes G2 in MSI 2026

TL;DR: With G2 Esports trying to keep the series alive, LYON (2024 American Team) ignored a market that gave them only 42% and closed Game 3 in 38:50. Their draft edge showed up on the Rift, their bot lane crushed early, and Berserker's Caitlyn became the voice of the upset.

Key Takeaways

• LYON (2024 American Team) turned a 23-5 kill score into 11 towers and 76.5k gold, showing that this was not a lucky swing but a full-map takeover.
• Berserker on Caitlyn finished 11/0/8 with a brutal +1338 GoldDiff@15, and that early lane gap became the cleanest reason G2 never found breathing room.
• Isles on Bard posted 3/4/16 and helped build 4 dragons for LYON, proving one of the pre-draft focus picks absolutely delivered in the game that closed the sweep.

Early Game

Game 3 began with the pressure squarely on G2 Esports. Down 0-2 in the best-of-5, they needed a stabilizing start, but LYON (2024 American Team) came out sounding like a team that already believed the market had missed the story. The pre-game line said only 42% for LYON, yet the opening minutes felt far more confident than that.

The first real clue came in bot lane. Berserker's Caitlyn built a punishing +1338 lead by 15 minutes over Hans Sama's Yunara, and once that lane tilted, the map started tilting with it. Beside him, Isles on Bard gave LYON exactly what the pre-draft analysis predicted: roam pressure, disruption, and the kind of tempo support that makes every side lane feel unsafe. That matters because Bard was one of the flagged champions before the match, and he absolutely delivered.

G2 did get one small point of resistance from top side, where BrokenBlade's Jayce held a +385 GoldDiff@15 into Dhokla's Olaf. So on prediction 1, Jayce did appear, but he did not deliver in the way G2 needed. The lane edge never became a real side-lane threat, and his 1/3/2 line told the bigger truth: pressure without payoff.

The Turning Point

The game turned when G2's composition stopped dictating and started reacting. The draft analysis had warned that Caps needed lane priority on Syndra and that SkewMond's Wukong had to create the first engage rather than answer Inspired's Trundle. That never really happened. Instead, Trundle's pillars and Bard's movement kept breaking up the angles before G2 could start a clean fight.

Once LYON stacked control around neutral setups, the mid game became theirs. Saint's Ryze at 4/1/10 linked the lanes together, while the jungle-support pairing kept turning skirmishes into numbers advantages. G2 only finished with 1 dragon and 2 towers; that tells you how rarely they could hold territory long enough to breathe. Even their successful moments felt temporary.

This is also where prediction 2 has to be answered clearly. The live draft model favored LYON (2024 American Team) at 55%, and yes, that draft edge materialized in-game. Their composition was easier to execute, stronger in lane pressure, and far cleaner around objectives. The result backed the read.

Closing Out

By the closing stretch, the numbers sounded like a lock being clicked shut. LYON reached 4 dragons, matched G2 on 1 Baron, and still built the decisive map edge through 11 towers to 2. That is why the final gold stood at 76.5k to 67.8k even without some wild late-game throw from G2 to inflate the chaos.

The star of the finish was still the bot lane carry. Caitlyn never gave a shutdown, ending 11/0/8, and every later fight carried the same message: G2 were already walking in damaged. Meanwhile, Labrov's Lulu fell to 0/6/4, and the protection tools never had the front-foot impact that this kind of comp requires.

For a podcast listener, picture the last minutes like a door slowly closing. First the vision disappears, then the side lanes bend, then the team with fewer towers has fewer places left to stand. By then, LYON were not surviving an upset narrative anymore. They were authoring it.

Polymarket Market

Retrospectively, the market did not read this single game correctly. Giving LYON only 42% understated how much their comfort and execution could punish G2's narrower draft branch. The market likely respected G2's name value and the theoretical upside of pieces like Jayce and Syndra, but the actual game rewarded LYON's simpler path: bot pressure through Caitlyn-Bard, stable side access for Ryze, and anti-engage control from Trundle. In that sense, the draft model at 55% for LYON was closer to the truth than the game market. Interestingly, the series market moved from 72% to 66% for LYON after the win, a reminder that even commanding results do not erase how volatile the next map can feel in a best-of-5.

Match Stats

FAQ

*Odds via Polymarket, 2026-07-10 10:45 UTC.*