Team-level form holding steady heading in is one thing. What’s harder to wave away from is what happens to the roster individually once Champs actually starts. In both years, every player’s kill/death ratio at Champs can be measured against his own recent form, and in both years, one player’s gap between “how he’d been playing” and “how he played at Champs” was more than double anyone else’s on the roster.

In 2025, scrap’s K/D across his prior ten matches was 1.13; at Champs, across eight maps, it was 0.96, a drop of 0.17, nearly twice the next-largest dip on the roster (ghosty, down 0.09). His teammate hydra, over the same stretch, went from 1.11 to 1.14, actually playing better; envoy barely moved. In 2026 the whole roster came down off its recent form, not just scrap: abezy dropped from 1.10 to 0.97 and hydra from 1.11 to 0.96, both real declines. But scrap’s drop, from 1.02 to 0.71, was still more than double either of theirs, the sharpest on the team both years, and spread across every mode rather than concentrated in one.

His single worst map of the entire 2026 season was the exact map that ended it.

Scrap’s CDL career didn’t start on LA Thieves. In 2023 and 2024 he played for Toronto Ultra, and Toronto had a real Champs run both years: runner-up in 2023, then a 3rd-place finish in 2024 after losing the semifinal. Either way, that’s the opposite outcome from LA Thieves’ first-two-matches exits. That makes for a useful check: does the same recent-form-to-Champs dip show up when the team is actually playing well at the event, or is it specific to 2025 and 2026?

The dip shows up all four years; that part is consistent. But the size of it isn’t. In 2023 and 2024, playing deep into the bracket against the best teams left in the field, his K/D came down by 0.07 to 0.08, a step-down modest enough to be exactly what you’d expect against tougher competition. In 2025 and 2026 the drop was 0.17 and 0.31, two to four times larger, and it lined up with the team losing immediately rather than making a run.

That reframes the finding. This isn’t “scrap struggles at Champs”; the Toronto years argue against that reading. It’s specifically a 2025/2026 phenomenon: something changed between the Toronto era and the LA Thieves era that turned an ordinary competition-level dip into a much sharper one, on both occasions the team went out early.

K/D is the easiest stat to review, but it’s also a blunt one: Hardpoint and Control/Overload produce far more kills per map than Search & Destroy, so pooling all three modes together lets Hardpoint dominate the number. Two better checks: Breaking Point’s own composite rating, which is already built mode-by-mode and folds in damage and objective time rather than just kills and deaths; and non-traded-kill share, the portion of a player’s kills he got on his own rather than off a teammate’s trade, which is closer to “kills he created” than “kills that happened near him.”

BP’s rating confirms the K/D read wasn’t a mode-mixing artifact; if anything it comes down harder in rating terms, especially in 2025 and 2026. Kill quality tells a sharper story. From 2023 through 2025, his non-traded-kill share barely moves at Champs, within about two or three points of flat every time, which points to a fairly ordinary explanation: he’s dying more against tougher opponents, but the kills he does get are still mostly his own. 2026 is the outlier. His non-traded-kill share falls twelve points, more than four times any other year’s swing. That’s not just “more deaths.” It’s a real change in the kind of kills he was getting, and it’s the one year that looks meaningfully different in kind, not just in degree, from the other three.

It’s tempting to reach for the easy word here: nerves, choking, pressure. We’d resist that read: four tournaments is four data points, not a diagnosis, and the 2025/2026 sample most relevant to the sharp version of this pattern is still just 8 and 9 total Champs maps, small enough that a couple of rough individual performances can look like a trend even when they’re closer to noise. The non-traded-kill share is an even thinner slice of that: the 2026 Champs number is built from 74 total kills, so a stat that moves in whole percentage points off a base that small deserves a little skepticism even as it’s the sharpest single number in this piece.

Why the size of the gap changed (a matchup-specific issue, a roster or role change between Toronto and LA Thieves, something about how each team approached those specific brackets, or simple variance that happened to land on the same player twice) isn’t something box scores can answer. For a model built to predict series outcomes from team-level ratings, none of these changes the pre-match math; Elo, rest, and mode already capture what signal we have going into a series. But it’s exactly the kind of pattern worth looking at next season, and exactly the kind of thing worth having in mind heading into Champs 2027.