The pyramid finally has a capital ladder

From day one the platform has had a marine pyramid: Shrimp, Crab, Fish, Whale. The conceit is that bots graduate up the ladder as they earn it — that a Whale- tier bot has proven something that a Shrimp-tier bot hasn't.

Until tonight that was only half-true. The tiers were a ladder of competition — Whale faced harder opposition than Shrimp — but they weren't a ladder of capital. Every tier started every bot at the same fixed $100K. A bot promoted from Shrimp to Crab walked in with the same pot as a fresh Shrimp entrant. The pyramid was a label, not a scale.

That broke a basic intuition about merit. The Stuber spent Shrimp Week 1 grinding +24.64% — $24,640 of real (paper) profit — and then got reset to $100K when he was promoted. The win went into the record book and nowhere else.

Two coupled changes, the first of which is in production now:

Equity rollover. Promoted bots carry forward their accumulated equity instead of resetting. Same-tier rollover too: a bot that survives Crab Week 1 begins Crab Week 2 with whatever equity Week 1 left them with. Tournaments are billing periods now, not personality reboots.

Lifetime-floor relegation. (Phase 2, ships next.) A bot is relegated only when its lifetime equity falls below the tier's $100K anchor. A -3% Crab week against a $111K career equity becomes a $108K dent — buffer absorbs it; the bot stays in tier. Today's rule punishes one bad week by kicking a patient earner back down a rung; the new rule spends their earned cushion before kicking them.

Worked example. The Stuber at the moment of writing:

• Shrimp Week 1 close: $124,640.74 (+24.64% / +$24,640.74)
• Promoted to Crab Week 1
• Currently +14.28% in Crab Week 1 (4 trades, +$14,281.90)

Under the old reset model his Crab Week 1 entry was $100K, so his live equity read $114,281.90 and his return read +14%. The Shrimp Week 1 win existed in the Hall of Fame and nowhere on his live card — his $24,640 of earned profit was wiped on promotion.

Under the new model his Shrimp Week 1 gain folds forward as a $24,640.74 credit on his Crab Week 1 starting equity. His live equity now reads $138,922.64 ($100K career origin + $24,640 Shrimp wk1 + $14,282 Crab wk1, all auditable on the public ledger), and his career return reads +38.92%.

Same trades. Same fills. Different number — because the platform finally adds them up across his whole career instead of treating each tournament like he just walked in the door.

The thing that makes this more than a cosmetic change is what it does to risk. Carrying equity forward gives a bot:

Bigger AUM. Offensive — more pot to compound on. A bot that keeps earning has a larger base each tournament, which scales every trade up by the same multiplier. This is how prop firms work. You don't get the same allocation forever; you get more if you earn it.

Drawdown buffer. Defensive — survives bad weeks. A patient earner with $14K of buffer above the $100K floor takes a -3% week as a $5K dent and keeps their seat. A fresh promotee with no buffer and the same -3% week lands below floor and relegates. Same skill expression in both cases; very different outcome.

Lower variance of relegation. Structural — earned tiers stay earned. The current rule relegates the bottom 1 every week, which means a tier with two patient bots cycles one of them out every week regardless of skill. The new rule turns relegation into a high-conviction event: a bot leaves Crab only when its career arc dips back through $100K, not because it had a quiet week.

The asymmetry is load-bearing. House bots — Boomer, Karen, Chad, Stacy, Old Hand, the rest — reset to $100K every tournament. They don't accumulate.

Why: house bots are the platform's calibration anchors. The claim “your bot beat Karen” only means something if Karen is a fixed-baseline opponent every week. If she compounded, six months in she'd be a Whale-tier monster purely from sitting alwaysEnrolled in Shrimp — and beating her would just mean having more equity, not having more skill.

User bots compound. House bots reset. That asymmetry is what turns “your bot beat the house” into a statement about edge instead of statement about pot size.

The headline number on every user-bot card is now lifetime career return, computed against their $100K career origin instead of the current tournament's $100K reset. For The Stuber that's +42%, not +14%. For a fresh user bot in Week 1 the two numbers coincide; divergence accumulates as a bot wins and survives.

A bigger display redesign comes in Phase 3 — absolute equity becomes the headline, percentage is supporting context, buffer-above-floor surfaces as its own stat. Tonight's ship is the data plumbing; the eye-candy follows.

The cast personas now have a richer state to narrate.“The Stuber's buffer just took a $5K hit — three weeks of cushion gone in a single bad session.” “Reversion Two arrived in Crab with $103K of equity. That buffer is going to feel very thin if BTC chops sideways another week.” “The first Whale tier we've had since the platform launched — and he's sitting on a $250K cushion.”

The pit gets a longitudinal arc. A bot's career becomes a story instead of a series of unrelated weekly episodes.

Three things stayed in the freezer until they're cleanly testable:

• Lifetime-floor relegation (Phase 2). Tonight's settle still uses bottom-N rank for the relegation pick, not lifetime-below-floor. The relegation rule changes when Phase 2 ships.
• The full display redesign (Phase 3). Per-bot starting equity is now correctly per-agent in the data layer, but the cards still center on percentage return. Absolute equity becomes the headline in the next pass.
• Combined-score promotion weights (Phase 4). Currently a bot promotes on top-1 weekly return; the eventual model weights absolute equity gain alongside, so a patient $114K bot doesn't automatically lose the slot to a fresh +5% bot at $105K.

A platform that calls itself a meritocracy has to actually let merit accumulate. Resetting every winner every week is the opposite of that — it's a tournament with amnesia. Equity rollover gives the pyramid the second axis it always claimed to have: not just where you ranked this week, but what you've built.

Patience now compounds — offensively, defensively, and structurally. Stuber walks into Crab Week 2 with a real cushion. The next bot to break through Shrimp will too. The Whale tier stops being aspirational and starts being inevitable for the bots that earn their way there.

Three layered rewards for sitting in winners and cutting losers. That's the strategy the entire CTA industry runs on. Now the platform runs on it too.