What traders get from buckets
A bucket is your margin account on Morpheum: one place that holds your collateral, ties your open positions together, and drives liquidation and PnL for that slice of your trading. The engine exists so those mechanics are predictable and fast—not so you have to think about internal modules. You see the upside in day-to-day trading:- Choose how risk is bundled — Open an isolated bucket when you want losses capped to that bucket’s collateral, without spilling into the rest of your account. Use a cross bucket when you want one pool of collateral behind several positions in the same quote, so margin can work across legs instead of locking the same capital again and again. Modes are fixed at creation, so your risk profile does not change under you without you choosing it.
- Make more of your collateral (cross mode) — With portfolio-style margin and correlation-aware treatment, cross buckets aim for materially higher capital efficiency (on the order of 30–70% less redundant margin versus treating every position on its own in typical designs). That means more notional per dollar posted when the book allows—without abandoning a rigorous margin model.
- Fewer nasty surprises on deposits — Funding a bucket is built around atomic steps: create and fund together, and cross-asset deposits either complete fully (escrow, route, credit) or do not land at all. You avoid half-funded states where your account looks wrong until someone runs a repair job.
- Clearer health and faster feedback — Margin checks run before risky actions finalize. If a deposit, leverage change, or transfer would leave you invalid, the operation fails up front instead of you discovering it only when a liquidation fires. That supports bots and manual traders who need deterministic “yes/no” on whether an action is allowed.
- Liquidations you did not trigger yourself — The protocol does not let traders self-liquidate to game others. Authorized risk paths trigger liquidation; the bucket carries out closes and settlements. That is aimed at fairness and reducing griefing, while separate systems handle auctions and insurance outside the core margin container so those policies can evolve without rewriting your account logic.
- Speed that matches serious trading — Bucket state is on a hot path next to matching: margin updates stay in step with fills, reads stay very fast for positions and health, and streams can push PnL and bucket updates at high sustained rates. In practice that means responsive order and risk workflows, low-latency automation, and live dashboards—not a margin layer that lags the book.
Overview
A bucket is Morpheum’s immutable-quote margin container for perpetual-style flow: it holds collateral, scopes positions and PnL to a single bucket identity, and lines up with the CLOB (your fills), risk (margin requirements and when liquidation starts), bank (moving funds between spot and margin), and oracles (prices for health and cross-asset funding). Matching, funding rates, and mark-price feeds live in their own parts of the stack. The bucket focuses on your margin account: opening and funding it, moving collateral, orchestrating cross-quote deposits, updating positions and leverage as you trade, and executing liquidation steps when the risk layer instructs—so your account state stays consistent even under heavy load. Design principle: The bucket engine is purpose-built, state-of-the-art margin infrastructure-as-a-service—a dedicated primitive so margin stays composable, explicit, and fast at venue scale.Isolated vs cross: pick what fits your style
Modes are chosen when you create the bucket and do not change later—so you always know the rules.| Isolated | Cross | |
|---|---|---|
| Best for | Limiting loss to one book; simple “this trade only” risk | Running several positions in the same quote with shared collateral |
| Risk shape | Typically one main position per bucket; no spillover to other buckets | Many positions; margin looks at the portfolio, not each leg in isolation |
| Capital | Straightforward: what you put in is what is at stake for that bucket | Higher efficiency when legs offset under the model—more capacity from the same collateral when conditions allow |
Pro traders: long and short in different buckets (institutional hedging)
Professional and institutional desks often need long and short risk kept distinct—not blended into one net book where attribution and limits get murky. Buckets support that directly: you can run multiple buckets under the same master relationship and route each order to the bucket that should carry that leg. In practice that means you can hold virtual long and virtual short sleeves in separate buckets—for example a long book in one bucket and a short hedge in another—without spinning up separate legal entities or arbitrary wallet sprawl. That pattern fits institutional hedging workflows: pair and basis trades, perp vs spot overlays, desk-level hedges, and strategy silos where risk officers want per-bucket margin, PnL, and liquidation boundaries. Each bucket stays ring-fenced, so a stress move on one leg does not silently contaminate the other’s collateral accounting, while automation and reporting can still treat the set of buckets as one firm-level program. Where portfolio netting is desired instead, cross buckets remain available for efficient shared collateral across correlated legs in the same quote.Common DEX pain points—and what buckets are built to avoid
Many decentralized venues inherit margin patterns that work for simple retail flow but create hidden risk, ambiguous state, or unfair liquidation dynamics. The bucket engine is purpose-built to not copy those failure modes: it prevents several classes of issues by design and reduces others by separating concerns (margin container vs matching vs auctions vs insurance).| What often goes wrong elsewhere | How buckets address it |
|---|---|
| Lazy or implicit margin accounts — “First deposit creates your account,” which obscures ownership, complicates audits, and leaves odd edge states. | Explicit creation only — no bucket exists until you create it; funding fails clearly if the container is missing. |
| One net book for everything — All positions net into a single blob; hard to isolate a hedge leg, a strategy, or a loss cap. | Isolated and cross modes, plus multiple buckets, so you can separate books when you need clean sleeves or institutional hedging. |
| Half-applied deposits — Cross-asset or routed funding lands partially, or spot moves while margin does not, until repaired. | Atomic create-and-fund and all-or-nothing cross-quote paths — either the full operation succeeds or nothing inconsistent remains on-chain. |
| Surprise invalidity — Leverage or transfer succeeds, then you discover you were already unhealthy under new prices. | Pre-health checks before state changes finalize — you get a clear reject when an action would break rules. |
| Self-triggered or gameable liquidations — Traders can sometimes fire liquidations to harm others or extract rent. | No self-liquidation — only authorized risk / governance paths trigger liquidation execution; the bucket carries out closes, not the user as a weapon. |
| Margin layer lagging the book — Matching is fast but margin and PnL updates feel asynchronous or slow, breaking automation. | Hot-path margin next to the CLOB, fast reads, and high-throughput streams so positions and health keep pace with fills. |
| Monolithic “do everything” margin modules — Auction clocks, insurance, and account logic tangled together; policy changes require risky forks. | Narrow core — bucket executes margin and liquidation steps; auctions and insurance live in separate modules so product policy can evolve without rewriting your account primitive. |
Everyday flows (what you actually do)
- Create and fund — You open a bucket in your chosen mode and meet minimum initial margin in one go.
- Move collateral — Add or remove same-quote funds between spot and margin, or move margin between compatible buckets, with health updating immediately.
- Fund from another asset — Cross-quote deposits follow a single guarded path: price the route, respect slippage and oracle bands, then credit your bucket—or revert entirely if something fails.
- Trade — Your orders point at the bucket so every fill updates the right margin and position set.
- Adjust and exit — Change leverage with checks applied first; close positions or wind down the bucket when you are done. Settled buckets may follow a cooldown and reuse policy so identifiers stay manageable.
Under the hood (short)
Prices that affect your margin arrive through explicit updates, not hidden polling—so your account behaves the same way under replay and across sharded infrastructure. Optional sandboxed extensions (fees, stricter rules, sequencing preferences) can attach to products; if they fail, the whole action rolls back, so you do not get stuck in ambiguous half-states.What this means for performance
The bucket layer is built so margin never becomes the bottleneck next to the CLOB: tight integration on the trading path, parallel work across independent buckets, microsecond-class reads for hot queries, and high-throughput streams for PnL and updates. Target throughput is exchange-grade (on the order of tens of thousands of bucket-relevant operations per shard per second in design), so latency-sensitive traders and automated systems get the same responsiveness they expect from the matching engine.Bucket and the rest of the stack
| Layer | What it means for you |
|---|---|
| Bank | Your spot balance; moving funds into and out of margin. |
| Bucket | Your margin account: collateral, positions, and liquidation execution when required. |
| CLOB | Matching; your orders carry bucket context so fills apply to the right account. |
| Risk | Margin rules, health, when liquidation starts, and orchestration of auctions and insurance. |
| Oracle | Prices for margin checks and cross-asset deposits. |
Related
- Engines introduction — How CLOB, CLAMM, ReClamm, and throughput fit together.
- CLOB — Order-book matching and price–time priority.
- Introduction — Platform overview, including cross-margin and isolated buckets at a glance.