Three ways to be the liquidity instead of the taker.
Each mandate is stated with its mechanics, its parameters and — at the same length — the conditions under which it is the wrong thing to do. A desk that only publishes the first two is selling something else.
| IAscending LadderGet paid to take profit.Read the mandate→ | IIFlat HarvestFee farming on the flat.Read the mandate→ | IIIDelta-Neutral LPHedged LP. Later, and honestly.Read the mandate→ | |
|---|---|---|---|
| What it is for | Distributing size into strength | Earning on a coin going nowhere | A fee stream with the direction removed |
| Where the liquidity sits | Single-sided, above market | Two-sided, wide, around spot | Two-sided, plus a short perp |
| Quote capital | None | Required | Required, plus margin |
| How it fills | At your price or better | Both sides of the range | Not selling — holding for fees |
| Primary risk | Price never arrives | A trend starts and you stay | Funding, and the hedge liquidating |
| Exit trigger | Filled, or pulled | Regime break → pull to spot | Regime break, or funding |
| Horizon | Weeks | Days to weeks | Days to weeks |
| Status | Primary mandate | Selective — regime gated | Selective — perp required |
Ascending Ladder
Get paid to take profit.
Your exit is placed as single-sided liquidity above market. It fills like a limit ladder — at your prices, never below — and earns trading fees on every order that crosses it while it waits.
How it works
- 01
The position is deposited as one-sided token liquidity across a geometric ladder of DLMM bins, starting a few percent above spot and reaching as far above as the mandate allows. No quote capital is required.
- 02
As price rises through a bin, the pool swaps your tokens into quote at that bin's price. You are the ask. Buyers lift you instead of you hitting them.
- 03
Every trade that transits a bin you own pays you the pool's fee on that volume — including trades that route through and come straight back down without filling you.
- 04
Bins are re-weighted, not re-priced. Once a rung fills it stays filled: the proceeds sit in quote and are swept, so nothing is round-tripped back into the token on a retrace.
Parameters
- Ladder floor
- +2% – +8% over spot
Close enough to catch ordinary strength
- Ladder ceiling
- +25% – +120%
Set against the mandate's target, not a forecast
- Bins
- Derived
Fixed by the pool's bin step and the range — the desk reports the count, it is not a dial
- Distribution
- Flat / front-loaded / back-loaded
Front fills sooner and lower; back fills higher and less often
- Quote capital required
- None
Single-sided by construction
- Fill discipline
- Proceeds swept to quote
Filled rungs are never re-exposed
When this is right
- You were going to sell into strength anyway, and the only real question was at what price.
- The token still has upside volatility — something has to lift the ladder for it to fill.
- Your timeline is measured in weeks, not in the next hour.
- The position is large enough that a market sell would move the price against you meaningfully.
When this is wrong
- Price never comes up. Then you do not sell. You are still long, you have collected fees, and that is all — this mandate cannot manufacture a bid.
- Price gaps through the entire ladder in one candle. You are fully filled at your average, and you watch it trade higher without you.
- You need the proceeds on a deadline. A ladder is a price commitment, not a time commitment; forcing it early means market-selling the remainder anyway.
- The pool is so thin that even the buys lifting your ladder are the only volume. Fees will be nominal and the fills will be slow.
It will not sell what the market will not buy. If price stays flat or falls, the ladder sits there earning fees and you remain long — which is exactly what a limit order would have done, minus the fees.
Flat Harvest
Fee farming on the flat.
A wide two-sided range on a coin going nowhere with real volume. Fees accrue with every trade; divergence loss only bites when the price actually trends. The work is knowing which of those two you are in.
How it works
- 01
Liquidity is placed in a deliberately wide band around spot — wide enough that the position is rarely knocked out of range and rebalancing stays rare.
- 02
Fees accumulate roughly linearly with volume. Divergence loss grows with the square of the log price move, so it is almost nothing while price is chopping and unbounded once a trend starts.
- 03
Width is chosen against realised volatility, not against a target APR. A narrow band earns more per dollar and dies faster; a wide band earns less and survives.
- 04
The position is pulled to spot — not rebalanced, pulled — when the regime detector says trend. Sitting through a trend is how range positions turn into bad directional ones.
Parameters
- Range width
- ±18% – ±45%
Set against realised vol, not against a target yield
- Rebalance band
- Rare, by design
Every rebalance realises loss and pays gas
- Exit trigger
- Regime break → pull to spot
Pull, not re-centre
- Fee tier
- Venue dependent
Only entered where the tier compensates the width
- Sizing
- Capped share of in-range liquidity
Being the whole range is its own risk
When this is right
- Volume is high and price is going nowhere. This is the entire thesis.
- The venue's fee tier is high enough that transiting volume actually pays for the risk being taken.
- The range can be set wide enough to make rebalancing rare without making the position pointless.
When this is wrong
- A trend starts and the position is still there. Divergence loss compounds and the position quietly becomes a long in whatever is being sold to it.
- Volume dries up. A wide range with no flow earns nothing and still carries the exposure.
- The chop is fake — a series of one-way gaps that look like range on a daily candle and are not.
- Someone else's liquidity floods the same band. Your share of the range collapses, and so does your share of the fees.
Every dollar this earns is rent for taking the other side of informed flow. In chop that rent is free money. In a trend you are the exit liquidity — so the only thing that matters is getting out before the trend is obvious.
Delta-Neutral LP
Hedged LP. Later, and honestly.
Short the LP's delta on a perp and the fee stream stops being a directional bet. It is deployed, on the tokens that have a perp at all — which is the smaller half of them. It imports funding risk, liquidation risk and a much larger operational surface, and it re-hedges on a threshold rather than continuously.
How it works
- 01
A constant-product position has a known delta: its value moves as the square root of price, so exposure is 1/√r of position value. That number is hedgeable.
- 02
Shorting that delta on a perpetual leaves the fee stream and removes most of the direction. As price moves the delta drifts, so the hedge has to be re-struck — continuously, and at a cost each time.
- 03
Funding is the tax. It is not a constant, it is not always in your favour, and on the assets this desk works with it is set by whoever is most levered that week.
- 04
The hedge is sized off the position, not off a model: the base token still standing in the range is the exposure, so that is exactly what gets shorted. The delta formula below explains why an LP position carries direction at all; it is not what decides the size.
- 05
The hedge leg can be liquidated. An LP position cannot. Bolting one onto the other means the combined position has a failure mode neither leg has alone.
Parameters
- Status
- Deployed, selective
Only where the token has a perp with real depth
- Hedge instrument
- Perpetual futures
Venue and depth dependent
- Re-hedge policy
- Threshold band
Re-struck when drift exceeds the band, on a one-minute floor — not continuously
- Hedge authority
- Trade-only delegate
The desk can adjust the hedge and cannot withdraw the margin
- Remaining constraint
- Funding + venue availability
Not the maths — the maths was always settled
When this is right
- There is a liquid perp on the same asset, with depth that survives the size being hedged.
- Funding is negative or flat over the holding period, so the short leg is paid rather than paying.
- The fee stream is large enough to clear funding, re-hedge slippage and the cost of carrying margin.
When this is wrong
- Funding flips against the hedge and stays there. The fees are real and the funding bill is bigger.
- There is no perp. Most tokens where this desk is useful do not have one with meaningful depth.
- The re-hedge loop lags a fast move, which is precisely when the delta is moving fastest.
- Margin on the short leg gets called during the move the hedge existed to protect against.
On most of the tokens where the first two mandates work, there is no perp to hedge with, and where there is one the funding bill can exceed the fees. This mandate is switched on, and it is still the one the desk turns away most often.
The full derivations are in the paper.
Figures on this site are computed from AMM formulas to illustrate mechanics. They are not a track record, a forecast, or a promise of return.