You might have heard about "Prop AMMs" on Solana
We built a lightweight dashboard:
- streams realtime data from 2 Prop AMMs
- plots continuous-depth curves => can see sub 1 bps
- cross-pool router simulator => can see sub 0.5 bps
00:00
u8
285 posts
- new panel in the PropAMMs tab bid/ask depths in realtime at 1/2/5/10bps!
00:00 - Gauging interest for a potential Breakpoint talk that deep dives into prop AMMs, including reverse engineering techniques and how they really function. pAMM alpha incoming Only like or engage if you're VERY interested in such a talk! Topics would include: - Reverse
- finally reverse engineered a Solana Prop AMM took all of last Sunday and most of today. started with virtually 0 sbpf knowledge. now I need to encode these procedures into an agent and give it the same tools I used
- Atomic Arb bots are making great use of DFlow JIT for onchain searching Latest strat seems to be: 2 DFlow JIT instructions, each one with 5 candidates (tessera, humidifi x 3, solfiv2) in opposite directions sent via Jito. if not profitable, doesn't land. no failed tx feesDFlow JIT routing is the future of trading on Solana. For perspective, HFT hires teams of PhDs to squeeze out basis points of edge, and retail on Solana gets that for free through our API. DFlow JIT routing is exclusively available through hitting our API directly Hit us up
- This might be controversial, but your transactions should be able to fight back against malicious sandwiching validators. I built a simple program to do exactly that. You can't know at runtime if slippage is natural market movement or a sandwich attack. But if your swap lands
- In our atomic arb dashboard we show Jito tip pct (total jito tips / total arb revenue) Saw that it was >100% for today. Thought it might be a bug But nope, someone actually tipped 420 SOL on a +2 USDC arb
- Replying to @MevRefundThis is because the tx that created the opportunity (0x912c5) was propagated when the slot was about to close so few searchers had a chance to get their bundle included. Searcher:proposer value capture seems correlated to how late an opportunity appears in slot-time.
- Checked the chain and noticed @JupiterExchange has now introduced JIT routing for Jupiter Ultra. Great to see. As a reminder, this means the router's onchain program checks what each pool would output and picks the best one at execution time
- Indeed! And you can confirm for yourself (*) $ ./target/release/solfi-sim spreads 1 (*) Using the solfi-sim repo
Daily reminder that spreads on Solana are sub-1bp! Use DeFi. - Routers always(*) operate on stale data (t-X), which means routing can't be perfect when pool conditions change between quote generation and onchain execution If you're from the EVM world, Helius Backrun Rebates is similar to: - bloXroute BackRunMe - CowSwap MEV Blocker -users and apps should be able to earn the rewards that their transactions create with this, they now can before, they were cut out of the process there's a lot of misinformation that goes on with this stuff, generally due to dishonesty or just a lack of understanding but this
- This seemed compelling so I looked into it Here's how they determined the pool address: - the deployer's first attempt failed - first attempt was 366 blocks before the successful liquidity add - first attempt contained the pool address, publicly and onchainI took a quick glance at the $HAWK drama to see if there's any merit to the speculation. The initialize transaction wasn't bundled, but there is something EXTREMELY suspicious. The sniper knew the address of the pool dozens of blocks before the initialize tx even landed.
