The Answer:
What Everyone Got Wrong
The TL;DR. Sixteen months of arguments collapse into four findings — each one proved out in the sections below.
Solo rank and customs success are two different ladders.
Crossplay is off in ranked, so solo rank is input-fair skill.
Ice is the GOAT. It's settled.
Down a touch right now — that's form. By everything that lasts, he's #1.
Bots and Iron Man don't warp who wins.
They pad the stat sheet, not the scoreboard. The ban arguments are debunked.
The "healer prison" was a myth.
Donovan and Michael said they were stuck on healer. The data disagrees.
The Roster
Find yourself first: tap your name → your card. Eleven players, eleven trading cards. Tap a card to flip it — full dossier, role split, signature & throw hero, six-quarter trajectory, and the persistence gap that explains the win rate.
Rise & Fall
The Overall blend, quarter by quarter (cumulative) — each line ends exactly on its standings bar below. The chart is how everyone got there.
The standings
There is no one "skill" number — skill is contestable, so the default Overall score blends all four lenses into one (each normalized, then averaged), and you can switch to any single lens to see what makes it up. Each names exactly what it measures and what it can't.
The Rank Ladder
Crossplay is off in ranked — so the rank you reach is pure, input-fair skill. Left: the solo rank each regular climbed to. Right: how they're winning customs right now (last 60). The two barely line up — solo rank doesn't predict who's hot today.
Is everyone at the right rank?
Win rate drifts toward ~50% at your own ceiling — so a win% above 50% on few games means you're underranked, with room to climb.
Hours Played
The single biggest confound in the whole record. Ice has played more than anyone — by a mile. Customs (lobby) + ranked + quick-play, all tracked time. Practice isn't cheating, but it's the edge nobody else comes close to.
Tracked match time only — customs from full lobby data, ranked & quick-play from each player's match feed.
Stacking:
The Real Lever
The headline finding. Not heroes, not bots — team construction is the single biggest controllable force in the record. The lopsided nights weren't about who picked Iron Man; they were about who was on whose team. Exhibit A: April 25, 2025.
Exhibit A — the 4/25 blowout
One stacked side, one starved side. Same lobby, same heroes, opposite outcomes.
Best & worst duos · recent form (2026)
Win% when two regulars share a team, over the last few months (15+ shared games). The spread between the best pairing and the worst is the whole argument.
⚔ The Team Simulator
Build two teams and see the odds. Tap a player to cycle them A → B → bench. Empty slots fill with bots — which are neutral (our own data: bot count doesn't move the scoreboard), so a side's strength is the average Elo of the humans you pick. Hit Balance to auto-split everyone you've picked into the two fairest teams.
Roles & Healing
Who actually plays what — and who's actually healing. Two myths die here: the healer prison (the loudest complainers heal the least) and the idea that anyone was locked into a role.
The role split — by playtime
Every player's Tank / Healer / DPS mix across 16 months. The “forced healers” spend the majority of their time on DPS or tank; the real supports chose the back line.
Who's actually healing — last 60 games
Average healing per game, counting every game (0 for the ones you played a non-healer). No opinions — just output.
The Moneyball Theory
The strategy Kyle laid out — and got clowned for. The data says it isn't an opinion. It's math.
Skill is relative, not absolute. Put each player in the role that maximizes the team's combined score — even if your best DPS is the one who should play healer. “Individual skill doesn't matter at all, it's relative to the other skills on the team… I'm always just looking for the highest net score.” He called it macroeconomics 101 — why countries trade. Michael named it the Moneyball strat; Donovan said it “makes perfect sense.”
“I completely disagree with all of this. People need to stop thinking they are the main character and play other roles.” It answered a question about preference that Kyle never asked — not the effectiveness claim he actually made — and later that same night Zach told the group to “ignore people and play who you want,” the exact main-character move he'd just condemned.
Settle The Argument
All 19 disputes from the chat. The claim → the data verdict → what actually happened. CONFIRM · REFINE · OVERTURN. Tap to open.
D17 & D18 — the bot question, measured
The chat blamed bots for everything. The honest answer depends on the lens. By the fairest lens — bots on the team you're fighting (the only one that tests whether the enemy can react) — the effect is small: win rate barely budges, so bots warp the stat sheet more than the scoreboard. By the own-team lens it looks larger (a single bot on your side reads ~% — small samples, and partly the lobby compensating you elsewhere). We lead with the fairer enemy-bot lens; win% vs the enemy's bot count:
The Fact Check & Credibility
Every checkable claim from 16 months of chat, graded against the data and windowed to when it was said. This is the thesis engine — it feeds the credibility scoreboard that closes the case in §10 The Verdict, right below.
The claims, one by one
Accuracy = share of a person's checkable claims that held up (true / vindicated = 1, mostly-true or exaggerated = ½, false = 0). Pure opinions and unverifiable claims (e.g. "90% of ban nights") are excluded from the score, not graded. The per-person rollup is the Verdict, immediately below.
The Verdict
Sixteen months, every argument graded. This is the scoreboard that closes the case — who was right, who was wrong, and who you should believe next Friday night. Accuracy when each person makes a checkable claim:
The bot & Iron-Man panic outran the evidence.
The Receipts
The legendary nights, rebuilt game-by-game from the lobby data. The memories were real — mostly.
The Lobby Tier List
Every hero with 12+ games this year as someone's main, ranked by the group's win rate on them — recent form (2026). (Confounded by who plays each.)
Most-contested heroes
The "claimed mains" — heroes 2+ regulars logged real time on. The overlap the chat fought about was real.
Superlatives
The data-certified awards.
🎤 From the Chat — what the stats can't measure
The awards no database can hand out: the arguments, the roasts, the predictions, and one line so good it got canonized.
Law of the Lobby
The rules the group discovered the hard way — codified, and reconciled against the data. Where the numbers back a rule (or correct it), it's tagged 📊 by the numbers.