Daisy is live on @Pumpfun
CA: C1wuSCJrjAbEXKXee2vn3KvZLmQY9JS4dmmnQZU9pump
Daisy is building an agentic prop AMM operator for Solana. The live desk data is open now at https://t.co/e457AtbwaG.
On top of that data, Daisy shows how an agentic prop AMM thinks. It prices, quotes and guards inside the rules you write, then widens, pauses or kills when risk rises.
Fills and PnL are simulated for now while we build on-chain execution.
Tighter spreads aren't always better.
We ran a full-week backtest again and again on the same real SOL/USDC tape. The only thing we changed was the target spread, from very tight to very wide.
What the curve shows:
• Too tight: the desk fills constantly, but price runs straight through the quotes and the week ends in the red
• Wider quotes fill less often, but each fill earns more
• Somewhere in the middle the score peaks, well above the Balanced preset
• Past the peak, fills dry up faster than the edge grows
Score is fills times edge per fill. Every new tape moves the peak.
New in the Daisy Desk.
108 of you already entered the Desk Challenge. Now you can see where you stand.
1. Public leaderboard
Every entry from this thread, ranked live. Scores are recomputed in your browser from each DSY code, so nothing has to be trusted. Same code, same score? The earliest reply ranks higher.
2. Score cards
Run your policy and download an image with your code, score and PnL curve. Attach it to your reply.
Click Load on any entry, including the leader's, to backtest it yourself.
Week 1 entries close 23 Nov, 00:00 UTC.
https://t.co/PDX34FC893
Watch the Daisy engine work.
This is a real replay, not an animation. One policy (DSY-12-5-80-6000-15000-1500) running through all 2,016 five-minute SOL/USDC candles from 2–9 Nov.
On every candle you can see the engine:
• read volatility from the last 12 candles
• size the spread to match it
• skew bid and ask against inventory
• fill only when price trades through a quote
• widen or pause when risk climbs
Final score: +$377.81 after fees, 2,934 fills.
Same code, same tape, same score. Run it yourself and try to beat it.
Quote updates trigger only after drift exceeds the configured threshold and remain capped by max updates per second to control fee burn.
Guards escalate deterministically from widen to pause to kill on stale oracles, toxic fill streaks, or inventory band breaches. Fully non-custodial: a scoped hot quoter key can only update quotes via allowlisted instructions, while treasury remains in a separate Squads multisig or cold wallet the agent cannot access.
AI drafts policy; the runtime enforces it and never signs. On-chain execution is in development; the live desk currently simulates fills and PnL.
The engine replays 2,016 real SOL/USDC 5-minute candles. On every candle it runs four steps:
1. Read volatility from the last 12 log returns
2. Size the spread: target × (1 + 0.35·(σ/σ₀ − 1)), clamped to your min/max
3. Skew both quotes against inventory
4. Fill only when the candle's high or low trades through your quote
A guard state machine sits on top: QUOTE → WIDEN → PAUSE as inventory or volatility climbs.
Win 5,000,000 $DAISY.
The Daisy Desk Challenge is live.
One winner takes the full 5M $DAISY reward.
How to enter:
1. Go to https://t.co/cFZBakOsN3 → Challenge
2. Tune your spreads, inventory bands and quote size
3. Ttest it on a week of SOL/USDC prices
4. Reply to this post with your DSY code
Week 1: 23–30 Nov
Entries close: 23 Nov, 00:00 UTC
Winner announced: 30 Nov
Every entry is scored on the same price tape, so the result is public and verifiable.
We hear you.
We just bought back, and every one of those tokens is going back to the community.
First up: the Daisy Desk Challenge. Build your policy on https://t.co/utyeCRM6pQ, run the desk for the week, and the top PNL takes 5M $DAISY.
Proof: https://t.co/MI5bmTTOGs
From SOL and its liquid staking tokens to JUP, PYTH, RAY and the loudest memes on chain, Daisy tracks them all side by side.
Live prices, 24h moves and liquidity, pulled straight from Jupiter and refreshed every five seconds.
Pick your market and watch:
https://t.co/utyeCRM6pQ
Every quote update is one transaction: a 5,000 lamport base fee plus the live median priority fee for about 20,000 compute units.
Across all 19 desks, that comes to roughly 0.41 SOL per hour at today's fees. Even if priority fees jump to 100k micro-lamports, it only rises to about 0.58 SOL per hour.
Daisy keeps it low by updating only when price drifts far enough to matter.
Check any quoter address on the Wallets page and Daisy shows exactly how long your balance will last.
Go run the numbers yourself:
https://t.co/e457AtaYl8
Inside the engine: how Daisy prices a quote 🌼
Watch the real quoting logic run, step by step.
When volatility rises, the spread widens with it, always inside the min and max you set. When inventory builds up, both quotes lean back toward flat so risk never piles up on one side.
Quotes only update when price drifts far enough to matter, which keeps priority fees low.
And when risk crosses your soft band, Daisy widens. Cross the hard band and quotes are pulled completely.
No black boxes. Every number you see is the same logic running in the desk:
https://t.co/e457AtbwaG
Inside the engine: how Daisy prices a quote.
Every slot, Daisy starts from fair value and builds a spread around it. Volatility widens that spread, while your min and max keep it inside the lines you set.
Inventory then skews both sides, leaning quotes back toward flat so risk does not pile up on one side.
Quotes only update when price drifts far enough, which keeps Solana priority fees low.
On top of it all sits the guard state machine: quote, widen, pause, kill. Kill always needs a human restart.
This exact logic runs in the Daisy desk today:
https://t.co/e457AtbwaG
We built Daisy to feel simple, even when the market is not. One clean desk, real Solana data streaming in every slot, and an agent that knows when to step back.
This short tour shows how it all comes together, from the first click on our website to taking control of the desk.
Fills are still simulated while we build on-chain execution, but the experience is live today.
Watch the video, then try it yourself.
https://t.co/e457AtbwaG
Meet Daisy an agentic prop AMM operator on Solana. Daisy reads the oracle every slot, quotes inside the rules you write, and steps back when markets turn.
Built for market makers who want tight spreads without giving up custody. Your keys, your rules.