NVG (@nvggnvg) header
NVG (@nvggnvg) profile picture

degen | solo developer | build tools that make you money

Delta-neutral zone
Joined November 2022
Photo by NVG (@nvggnvg)Photo by NVG (@nvggnvg)Photo by NVG (@nvggnvg)Photo by NVG (@nvggnvg)Photo by NVG (@nvggnvg)Photo by NVG (@nvggnvg)Photo by NVG (@nvggnvg)Photo by NVG (@nvggnvg)Photo by NVG (@nvggnvg)Photo by NVG (@nvggnvg)
Pinned Tweet
never liked trading shitcoins. grew my LP capital 3–4x in 2 weeks on Robinhood. just providing liquidity. built Yield-P for this. a free, non-custodial EVM LP terminal. public beta is open. 🧵
1
5
2
21
624
losses happen too 24h in bid-ask: deposit: $3,000 pnl: -$300 return: -10.1% not exactly the screenshot you flex on the timeline. but with musebook dumping this hard, simply holding the token would've hurt a lot more. bid-ask didn't remove the downside it just made the loss way less painful. sometimes -10% is actually the better outcome.
4
98
34 minutes in LP deposit: $3,009 pnl: +$770 return: +25.6% $GPU/USDT on Uniswap v4 not a bad result for basically drawing a range and waiting for the market to do the rest. closed before it even had time to get boring
2
140
16 NFT positions. all on one side of the price. why? say you want to accumulate token musebook lower. instead of one big LP range, YieldP Bid-Ask splits it into 16 levels below spot. less liquidity near price. more liquidity deeper down. musebook drops through the ladder → USDG gradually turns into musebook. not one entry. not one price. 16 levels. and if deeper ranges have less competing liquidity, your share of active liquidity can be higher - meaning a bigger share of fees when volume trades through them. want to sell into strength instead? flip it. 16 levels above spot, gradually selling as price moves up. accumulate lower or sell higher. which side are you taking?
4
96
hood volumes are definitely cooling off but there’s still enough flow to keep catching these pretty consistently Nautilo / USDG deposit: $2k pnl: +$221 return: +11.1% less volume, same game free. lfg
1
84
NVG retweeted
My Delta alternative for LP farm on RH One update to my LP farming guide. I now use Yield-P, built by @nvggnvg, as my main alternative to Delta for opening and managing positions on Robinhood Chain. It covers the full LP workflow in one place with a cleaner UI and less overhead. › discover pools by chain, pair, fee tier, TVL, volume and Fee APR › build Spot, Curve and Bid-Ask setups with custom ranges › track position value, unclaimed fees, IL versus holding, daily P&L and range status separately › collect fees, add or remove liquidity, or close directly from the dashboard › optionally swap through 0x when entering or exiting › review the final P&L and generate a shareable result card I also prefer how positions are opened on Yield-P. The builder makes it easier to see the actual ranges around the current price and understand where each bin will sit before signing. The fee difference is meaningful too. Delta latest manager contract sets the cut at 7.5% of collected LP fees, up from 1%. Yield-P is currently free to use, apart from gas and swap costs. It is non-custodial and does not add another contract layer. LP actions go directly through Uniswap or PancakeSwap, swaps go through 0x and every transaction is signed from your own wallet. It is still a public beta, so bugs and UI risk are part of the deal. But for finding a pool, opening a range, tracking real P&L after IL and closing the position, this is the cleaner setup I use now.
1
2
18
544
just another boring 5 hours LPing deposit: $1500 pnl: +$257 return: +17.2% should’ve just held, obviously
7
130
8/9 that 3–4x was one two-week run, not a promise. the pools got tasty. then came a weekend full of rugs. LPing still carries token exposure, IL and protocol risk. non-custodial doesn't remove UI risk. fees don't make you immune.
1
100
9/9 public beta. expect some bugs. start with the P&L view or explore a Bid-Ask setup. what's the most painful part of your current EVM LP workflow? reply here or DM Welcome yield-p.xyz
48
50%+ APR on oil ticker sounds fake until you realize the setup is actually clean. ticker: BRENTOIL exchanges: Lighter / TradeXYZ (hyperliquid) you’re getting paid on funding while also farming potential retrodrop rewards from TradeXYZ - the top HIP-3 protocols on Unit. this is the kind of setup i like: - real yield today - possible upside tomorrow - no need to pray for “points” with zero liquidity current funding is income. retrodrop is the free option on top.
3
103
it's still very tasty
token: EDGE exchanges: bitget / edgeX APR: 35% so tasty
1
73
funding arbitrage is not the “stable strategy” people like to sell. it’s usually just a fancy way to slowly lose money while thinking you’re being market-neutral. here’s why: funding spreads mostly exist on highly volatile tokens and that’s the first problem — they are extremely hard to manage. price can easily move 2–3x against you, and suddenly your “delta-neutral” position is just a poorly structured directional bet max OI capacity on low liquidity tokens is another constraint you simply can’t scale even if the setup looks good on paper in reality, most of these “arb opportunities” don’t even cover basic entry costs fees, slippage, order book spreads — all of it eats the edge completely then add ADL risk on abnormal tokens and you can get forcibly reduced at the worst possible price exactly when things go wrong and SL/TP? often they don’t work the way you expect in thin liquidity gaps, illiquid books, and execution delays break the model fast so the real question is simple: you’re risking 50% position exposure in a volatile token to make 1–2%? don’t be ridiculous. just farm historical funding instead.
1
2
54
the biggest breakthrough in ai coding wasn’t just better models. it was agents harnesses. they gave llms: - tools - memory - execution loops - retries - terminal access - file editing - evals that’s what unlocked tools like Claude Code and Codex. before that, models were mostly advanced autocomplete. after harnesses, they started acting more like autonomous engineers: debugging, running commands, validating outputs, retrying failures, iterating on tasks. same model. different harness. completely different capabilities.
1
44