Amoeba Farm Roadmap

Now

One order book for capped futures

In design and prototyping · not live yet

One market, any leverage

Each index and expiry gets one shared order book.

You pick your direction, size and leverage, up to 50x, and everyone trades against the same bids and asks. A 2x trader and a 50x trader can trade with each other.

How a position works

You choose leverage when you enter. Your maximum gain and maximum loss are both the entry price divided by your leverage.

At 5x, your payoff moves while the index is between 80 and 120. The most you can gain or lose is 20 per unit.

Inside that range, your profit or loss moves one-for-one with the index. Outside it, it stops changing.

Leverage goes up to 50x; the exact menu at launch depends on the finished risk model. The cap applies at expiry: touching the edge of your range before then does not close your position.

Why this design

  • No liquidationsYour maximum loss is paid in when you enter, so a price swing can't force you out.
  • No funding rateContracts have a fixed expiry and settle against the oracle.
  • One pool of liquidityEvery leverage level trades in the same book.
  • Trade anytimeSettle at expiry.

The trade-off: your upside is capped too, and a shared book can't guarantee you a good price if you exit early.

Under the hood

  • Fully on-chain matching on SolanaInspired by Phoenix. The program enforces the rules, not a private server.
  • Open market makingMakers can post compact liquidity curves next to normal limit orders. Our own market-making bot is just another participant and gets no special treatment.
  • Every position fully backedWhen traders with different leverage meet, each pays in exactly what their side can lose, so the market always holds what every position can pay. Any trade that would leave it short of funds is rejected.

Open question: early exits

Your position keeps its original entry price and range. Closing it early at a new price needs clear, fair valuation rules that can't be gamed by repeated trading or cooperating accounts.

We will publish our solution with the market spec.

Milestones

  1. Review the order book prototypeDone when tests show matching works correctly, including partial fills and cancellations.
  2. Finish the market specDone when every action, including early exit, has clear accounting rules.
  3. Stress test end to endDone when every market stays fully funded through price jumps, extreme leverage gaps and attack scenarios.
  4. Benchmark and integrate on SolanaDone when speed and cost are measured and published, including the frontend.
  5. Launch and migrateDone when the new book goes live with clear risk limits, and existing positions have a documented exit path.
Next

The oracle rebuild

Committed · the second stage

Why it matters

Every capped future settles against the oracle, and market makers need fresh data to keep their quotes accurate. The rebuild focuses on trust and speed.

The oracle should be useful even to people who never trade. Our goal is for anyone to run an agent that finds missing price data, collects evidence and submits it under public rules.

How contributions work

  1. Find a gap
  2. Collect evidence
  3. Submit
  4. Validate
  5. Publish
  • Open tasksMissing prices, stale data and disputed findings are visible to everyone, so agents can find useful work on their own.
  • Evidence, not just numbersEach submission includes the price, product, source, timestamp and proof.
  • Clear validationChecks cover source quality, product match, duplicates and freshness. Rejected submissions get a reason so contributors can improve.

What changes

  • Built for agentsDocumented submission tools and a reference agent replace the manual TUI workflow. Anyone can build their own collector.
  • Public rulesSource eligibility, weighting and dispute handling are written down. Copies of one source don't count as independent confirmation.
  • Game mode staysFor evidence that needs human judgment or challenge.
  • Existing contracts are safeThe September and October 2026 launch cohorts keep their original rules. The new rulebook applies to later cohorts only.

Milestones

  1. Publish the new rulebook and evidence format
  2. Open the agent contribution path
  3. Finish validation and game-mode handling
  4. Run the new oracle in shadow modeNext to the current one, comparing results.
  5. Switch new cohorts to the new rules
Launch

Liquidity at launch

Our plan for the first months

Real bids and asks from day one

A new market is only useful if there are real bids and asks in it.

  • We make markets firstAmoeba Farm will run its own market-making bot and seed the pools that fund it at launch. Our bot follows the same rules as everyone else and gets no special treatment.
  • Limits match our capitalLaunch leverage and position sizes will be set to what our market-making capital can support. They grow as capital grows.
  • Start focusedWe will launch with a small number of markets and expiries, so liquidity is deep where it exists instead of thin everywhere.
  • Raising capitalWe are in early talks to potentially raise capital, which would let us deepen launch liquidity. Nothing is final, and we will share details if and when it happens.
AMBA

The AMBA token

Two concrete roles, both tied to how the protocol works

Market maker bond

Anyone can post orders on the book. Makers who want official market maker status post an AMBA bond. In return they get fee rebates, and they commit to clear rules on spread, size and uptime.

  • The bond makes it expensive to spam the book or to quote without real intent to trade.
  • Makers who leave in good standing get their full bond back.
  • Makers who break the rules lose part of it.

Data access for agents

Browsing hardware prices on the website stays free for everyone. Agents and developers who want machine-readable oracle data stake or pay AMBA for access, with faster and deeper data at higher tiers.

This ties AMBA directly to the oracle's main goal: becoming the reference price for hardware that both people and agents rely on.

Later

The horizon

Directions, not dates · each depends on the layers before it

The reference price for hardware

Start with DRAM and NAND, then expand. The goal is to be the first place people and agents check to see what hardware actually costs.

Product histories, specs, sources and availability, readable by both people and machines.

Agents that finance real production

The bigger goal is to connect markets to real production. An agent should eventually be able to spot demand, price the inputs, raise capital, contract for output and have the result verified.

  1. Observe the economy
  2. Price risk
  3. Finance production
  4. Verify the result

One idea we're researching is longer-dated contracts backed by delivery promises. Before that can work, we need clear answers: who owes delivery, what backs the promise, how completion is verified, and what happens if delivery fails.

The order book gives agents a place to trade. The oracle gives those trades a trusted reference. Production is where both become useful in the real world.