Champions Shanghai Pick'Ems: Four Groups, Two Slots, and the Math Behind a Top-20% Card
Pick'Ems lock with the Champions Shanghai group-stage opener on Sept 24, one day after Patch 13.06 ships a new rifle. We cross-tabbed the GameGrids settings database against all 16 rosters - 60 tracked players, 57 numeric eDPI records - and the four group medians sit 16 points apart. Here is the defensible card: consensus anchors, one researched gamble, zero patch hype.
Riot's Champions Shanghai Pick'Ems are live, and the last 48 hours produced a wall of filled-out cards: THESPIKE.GG's group-stage guide, Strafe's predictions, Hotspawn's best picks. They are all team-picking on narrative: redemption arcs, miracle runs, home crowds. This piece asks a smaller, meaner question. With the tournament opening September 24 at the Mercedes-Benz Arena - $2,250,000 on the line, $650,000 to the winner - which picks on your card can actually be defended with data you can show someone?
What the card looks like
Sixteen teams, four groups of four, two advancement slots per group. The draw put us here:
You mark two teams to advance in each group, and the reward ladder is public: a "Calculated" title just for submitting, the Stone of Heaven Buddy for the global top 50%, the Pick'Ems Master Buddy for the top 20%, and a "100%" title for a perfect card. Top-20% is where the card stops being a souvenir - and, as we will keep saying, that is a percentile, not a skill floor. Eight picks, one lock time, no re-drafts: the entire game is choosing where to spend risk.
60 pro setups, 4 groups: what we counted
Here is the only dataset none of the prediction pieces bring: our own. Method, stated flat: we took the team field of the player settings database, matched it against the 16 announced Champions rosters, de-duplicated names, and kept entries with a numeric eDPI (DPI x in-game sensitivity). Cutoff: September 18, 2026. Result: 60 tracked players - 13 in Group A, 9 in B, 16 in C, 22 in D - with 57 numeric eDPI records.
The group medians are quoted at the top of this article: 236 / 248 / 240 / 232. Four "different" groups, statistically one population. The old trope - low-sensitivity EMEA discipline versus high-sensitivity NA twitch - does not survive contact with 60 rows of data. Nobody in the eDPI calculator needs to be told twice that VALORANT's pro band is wide and flat.
Where the signal lives is team granularity, and it is genuinely strange:
Read those two facts together and a defensible card starts to write itself: the tournament's aiming consensus is one mid-band population, so upset-chasing against "style" theories is speculation. If a seed looks beatable in our numbers, it is because of who is missing from the numbers - JD Gaming, FUT Esports and TYLOO have zero tracked entries in our database. Three of the sixteen rosters are coverage blind spots, which is a polite way of saying: whatever the team pages and this database show, treat Group A and Group C predictions as the least data-backed ones on the card. We also compared against our earlier tournament pieces, the Champions settings guide and the pro eDPI bands survey - this is the first time the two files have been cross-tabbed by group.
Consensus anchors and the one-slot-per-group rule
THESPIKE.GG's staff card runs 100 Thieves and T1 in A, Global Esports and Vitality in B, TYLOO with Paper Rex in C, Karmine Corp and Nongshim in D. That distribution teaches a rule worth stealing: anchor six slots to structural strength, keep two slots for researched risk. Structural strength here means won-the-qualifier-and-the-stage: 100 Thieves closed Americas Stage 2 and took the Esports World Cup trophy; Karmine Corp crawled out of a 0-5 hole to qualify; Global Esports reverse-swept their way into the conversation. Those are not vibes, they are results under pressure - the only predictor Champions has ever respected.
The genuine gamble shortlist is shorter than the hype suggests. TYLOO advancing from C is the home-crowd pick and it is genuinely 50/50 with a resting G2 bracket. Nongshim's series against Global Esports in the Pacific final went the distance; their five tracked players form the tightest low-variance eDPI block outside EDG (160 to 280, median 216) - nobody in that group is playing Russian roulette with their aim. LOUD are the wildcard with real fireworks and the youngest international stage in the field. And the XLG problem: if their two 550-plus snipers get hot, the bimodal profile cuts both ways - which is precisely why they belong on an upset card and not on yours. One risk per card, chosen from that list, is the highest-expected-value gamble we can defend.
The patch-calendar problem
The underrated variable is the calendar. Patch 13.06 goes live September 23: the Warden rifle at 2,900 credits with an 18-round magazine and a 2x scope (we ran its numbers here), the per-agent gear rework, Agent Mastery tracks, and a practice-queue overhaul arriving in a later update. International rosters get exactly one day of on-patch practice before the first match, and the map pool locked before the patch. Practical translation: Warden appears in zero to two serious scrim packages, most teams board the stable patch, and nobody's group-stage win probability moves because a new gun exists. Cards that got rewritten by patch hype are worse than cards that ignored it. The patch is weather; your picks are climate.
Five card mistakes worth naming
Frequently Asked Questions
When do Champions Shanghai Pick'Ems lock?
At the group-stage opener: September 24, 2026 in Shanghai. Riot's official Pick'Ems hub carries the exact in-client countdown; plan on the first match starting, not the first day ending.
What can you actually win?
Participation pays a "Calculated" title. Global top 50% gets the Stone of Heaven Buddy, top 20% gets the Pick'Ems Master Buddy, and a perfect card gets the "100%" title. The tiers are percentiles of all submitted cards - field size, not your math, decides what 20% takes.
How many picks are on the card?
Eight: two advancing teams in each of the four groups (A-D), the same format every outlet's guide uses.
Which pairs do the prediction guides lean on hardest?
100 Thieves with T1 in A, Global Esports with Team Vitality in B, Paper Rex with TYLOO in C, Karmine Corp with Nongshim RedForce in D, per THESPIKE.GG's guide.
Which second-slot seat is the most contested?
Group D's. NRG's big-event reputation collides with a Nongshim roster that pushed the Pacific final the full distance, with XLG's two-recorded-guns chaos and Karmine Corp's momentum sharing the same table.
What does the site' settings database actually contribute?
One negative result and two flags. Negative: 60 tracked players' eDPI medians (236/248/240/232) show no meaningful aiming difference between groups. Flags: XLG's two setups above 550 eDPI mark the most variance-hungry roster profile we track, and JD Gaming, FUT and TYLOO having zero database coverage marks Groups A and C as our thinnest evidence.
Will the Warden rifle change the group stage?
Probably not on week one. The patch lands one day before opening matches, and a 2,900-credit scoped rifle needs a purpose, not just a hype window. Expect a situational appearance, not a meta, per our Warden math piece.
Who do the data-backed cards favor?
Nobody, which is the honest answer. With group medians sixteen eDPI points apart and four of sixteen rosters untracked, a top-20% card in Shanghai is consensus anchors plus one well-researched risk - plus luck you should not be able to name in advance.
The takeaway
Fill six slots with teams that won something under pressure - 100 Thieves, T1, Vitality or Global Esports, Paper Rex, Karmine Corp, and one Pacific representative - spend your remaining two on the TYLOO/Nongshim/LOUD tier of researched risk, and refuse to rewrite the card because a new rifle exists. The only aiming data that spans all four groups says the groups are one population; your edge is knowing that before the hype does.
Sources are linked inline above; the underlying counts come from the the site player-settings database, cross-checked September 18, 2026.