Agentic Mode

A concentrated range earns only while the price is inside it. The tighter the range, the more it earns, and the more often it has to move. Agentic Mode hands that work to an LP agent. The agent keeps your eUSD LP position in a tight range around the price, keeps the debt at the target ratio, and turns your FABLES into eUSD.

You stay the owner. The position, its vault and everything in it remain yours. The agent can only rebalance and harvest the position. It cannot withdraw, close, or send anything anywhere else.

Turning it on

1. Open your position on own.money/mint/lp, go to the Agent tab in Manage, and click Turn on Agentic mode. This calls setDelegate(agent, 100) on your vault. Managed positions show an agent tag.
2. The agent picks the position up within a few minutes and moves it to the strategy below.
3. To stop, click Turn off Agentic mode in the same tab. The position stays exactly as the agent last left it, and you manage it yourself again.

There is no extra fee for Agentic Mode. You pay the usual stability fee, and swaps in a rebalance pay the pool's normal swap fee.

What the agent does

The ranges are centred on the pool price, not the oracle, so they sit where trades happen. Each change is one transaction: the agent removes the range, swaps in your vault if the new range needs a different mix, and adds it back at the target debt. It keeps your choice of staked or unstaked.

Outside market hours SPY rarely moves much, so a tight ±1% range earns several times what ±10% does. When NYSE opens, the agent widens to ±10% for the session.

On eUSD/USDG both tokens sit near $1, and a 2-tick range takes most of the pool's emissions. If the pool is too thin for the full swap a re-centre needs, the agent swaps what fits and adds the rest over the next checks, once arbitrage has refilled the pool.

How much more it earns

Normal mode is a position you open yourself at the app's default range and leave alone. Agentic Mode runs the strategy above.

"On the position" is FABLES earned on the whole LP position, less the cost of moving the range. At the 200% ratio, half of the position is eUSD you minted, so your own capital earns about twice that. Both are before the stability fee.

These are modelled on today's pools: this week's FABLES emissions at the current FABLES price, the other liquidity in each pool as it is now, and real hourly SPY prices. They are estimates, not promises. They fall as more liquidity joins a pool, and move with emissions and the FABLES price.

They are raw APRs. They leave out the boost from bribes, such as the LP rebate, which add to a pool's FABLES emissions on top of these numbers.

Why this strategy

FABLES emissions in a Fables pool are shared by in-range liquidity, not by dollars. The same dollars spread over a narrower range count as more liquidity, so a tight range earns a bigger share of the emissions and of the swap fees, but only while the price stays inside it. Every time the range has to move, the position pays a swap fee and some price impact. The strategy for each asset is the tightest range that the asset's price lets the agent keep without moving it too often.

SPY

• ±1% outside market hours. When NYSE is closed, SPY's onchain price hardly moves: the typical hourly move is under 0.1% on weekday nights and about 0.02% at weekends. Arbitrage keeps the pool at fair value within seconds. A ±1% range stays in range almost all the time and earns several times what ±10% does over the same hours.
• ±10% while NYSE is open. During the session SPY moves several times more per hour, and most of the day's move happens then. A tight range would fall out of range often, and every re-centre would pay to swap back. When we compared widths on real SPY price history, ±10% earned the most of the widths we tested during market hours, once those costs were counted.
• Switching at the open and the close. The width follows the market's own schedule, about ten switches a week. Outside market hours the agent also re-centres if the price stays out of range for 15 minutes, which ignores a single stray trade but catches a real move.
• 200% collateral ratio. The minimum is 190%, so 200% borrows almost as much eUSD as allowed while leaving a buffer for price moves. The agent repays below 195% so the position never gets near liquidation. It mints back above 225% only at the next open or close, so it doesn't add debt into a fast move.
• Bear mode. Leverage helps in a rising or flat market and hurts in a falling one. In a downturn the agent removes all the debt, so nothing can be liquidated, and keeps earning fees in a wide range. In daily backtests on SPY since 2005, this approach had no liquidations, and it switched to Bear mode for the 2008 and 2022 sell-offs.
• Centred on the pool price. eUSD trades at a small premium in the pool, so the oracle price can sit a little away from where trades happen. Centring on the pool keeps the range where the volume is.

eUSD/USDG

• ±0.01% (2 ticks). Both tokens trade within a fraction of a cent of $1, so the price barely leaves even the tightest range. At today's pool sizes, our simulations show a 2-tick range takes roughly three quarters of the pool's emissions, against about 5% for a ±0.5% range.
• Re-centre after 1 hour out of range. Small drifts in a stablecoin pool usually come back on their own. Waiting an hour avoids paying to swap for noise, and still re-centres a few times a week when the price really moves.
• Swaps sized to the pool. With our range out, there's little other liquidity at the current price. The agent swaps only what fits and adds the rest later, so a re-centre never pushes the price out of its own new range.
• No Bear mode. Both sides are dollars, so there's no market downturn to protect against.

Bear mode (SPY)

The agent reads SPY's and VIX's daily closes. If SPY closes below its 200-day average while its 50-day average is also below it, or more than 8% under its 60-day high, or VIX closes above 30, the agent switches the position to Bear mode:

• It repays all the debt, so nothing can be liquidated.
• It keeps the whole position earning in a wide range, from 33% below to 50% above the price.

After three clean daily closes in a row, it returns to the normal strategy at its next check while NYSE is open.

FABLES

About once a week, the agent claims your FABLES and swaps them into what your vault's harvest setting asks for: eUSD (the default), the asset, or FABLES. The output stays in the position, and the next rebalance adds it to the range.

Limits on the agent

• Daily loss budget. Your vault caps how much equity the agent's actions may lose in a day, measured on-chain at the oracle price. The app sets this to 1%. A rebalance that would go past the budget fails, and the agent waits for the next day. Selling USDG at eUSD's small pool premium counts toward this budget too.
• Price checks. Every swap has to get at least the oracle value less the pool's price band, and the agent sets a tighter minimum from its own quote. Rebalances need a fresh oracle price.
• Health check. A rebalance can't leave the position below the minimum collateral ratio.
• Rate limits. The agent waits at least 10 minutes between actions on a position, except when it repays debt, and makes at most 20 calls a day.

Risks

Agentic Mode has the same risks as eUSD LP, with these differences:

• Tighter ranges move more often. Each move pays a swap fee and some price impact. The agent sizes swaps to the pool's depth to keep this small, but it is a cost.
• The agent can be wrong or offline. If it stops, the position stays where it was, which is the same as an unmanaged position. Bear mode follows daily closes, so it reacts within a day, not within the hour.
• Approval is a trust decision. The loss budget, the price checks and the health check bound what the agent can do, but within them you rely on its judgement. You can revoke at any time.