@0xngmi @patfscott, pls need you to check this

сс defillama.com/protocol/colle…

thought i found a niche prediction market with an inefficiency but turns out they added this recently and just didnt announce it aaaaaaaa @DefiLlama on cc revenue kept/vaulted pulls count the full pack price as revenue and sell backs subtract it later so daily revenue is basically when whales decide to sell back not margin. today went 1.46m -> 0.69m intraday, for a market on october revenue thats kinda a big deal, worth a note in the methodology

lol risk free

haha, I was recently working on a similar idea, but a bit more complicated. The idea was to maintain the actual index weights: new buys go into assets that have dropped below their target %, while sells come from assets that have grown above their target weight.

finally played with uniswap v4 hooks as i see it, it only works if your pool actually gets order flow, so you need: - something unique (a pair) - target pools to trade against right? but it also needs real demand e.g. a meme/usdc pool rebalanced against meme/stock + stock/usdc: - arb is expensive: launchpad hook 1-3% + v3 0.3-1%, so the pool drifts 1.3-4% before rebalancing pays and lps eat it - you're stuck holding a meme that usually dies in days - 3-4 swaps of gas per buy, so routers prefer the bots' direct path couple questions: - how do you handle hedging in your setup? - how do you get flow into your pool, uniswap allowlist or your own frontend/bots? - who pays the arb gas in your hook, the swapper? doesn't that make routers skip your pool?

gg

If there wasn’t a signing thing, even auto-signing feels slow. It would be much cooler to just deposit into a balance and then trade like papertrade did

ik, i meant actual numbers. 99% per contract but realized ~90%, part goes to refilling lp after trader wins. imo 5m cap makes the house super efficient but fragile, no buffer so any trader win pauses payouts

excess over 5m gets swept right away, so lp has no buffer. any move where traders net win pulls it below the cap, and stakers go from ~99% to 1% until losses refill it. so time at the cap showed how often trader pnl swings against the house, i.e. how stable staker income is

1. yeah fair, time above 5m doesnt matter. at the cap ~90% of trader losses go to stakers (oct 14: 5.43m of 5.93m), below cap less (oct 11: 59%) 2. agree, i kept mint flat. with decay 11.2b is around nov 5-28 depending on H (35-17m/day), supply caps ~12.6b. per 1m paper ~695 -> ~565 by oct 21, ~470 mid nov

Would be cool to have stocks/RWAs or other efficient crypto markets where you need to burn or get paper to launch Not sure about lowcap/midcap tokens or memes, though. Since there’s no order book and the price comes from an oracle, it could be relatively easy to manipulate the price and drain the LP, like what happened on launch day

I think it only makes sense if you expect: a) PAPER to become tradable on DEXes b) volume to stay where it is now or keep growing LP vs the $5M cap: - above $5M only ~8–11% of the time - staked supply is growing ~8%/day Weekly model (staker payouts stay at today's level) Oct 15 -> Oct 21: - Staked: 7.9B -> 11.2B PAPER - Stake growth: +7.5% -> +5.1% per day - Paid to all stakers: ~$5.3M per day, flat - Per 1M PAPER: ~$690 -> ~$485 per day (-30%)

That was an expensive sleep, F

Also tried playing Monte Carlo at the end, almost got back to profit, then hit a losing streak and ended up at $0.00675 per PAPER At this point, the payback period is ~23h (if the hype continues) The key metric now: staking yield per 1M PAPER per hour, and how fast it decays

true, farming is going faster now, but closing still takes 5 - 20 sec

I love how we all optimized our latency, legs, cost per PAPER and all that… …and then got fucked by the relayer queue, which wasn't mentioned once in the docs 🙃 How's your 100–200× on BTC with the close at 70% to liquidation, colocation in Tokyo and < 5m LP going? lol @papertrade_xyz

exactly, you just need to figure out how to connect each “piece” now and ofc experiment experiment experiment

this

Two years ago, this was a team with infrastructure. Now almost every stage of an on-chain strategy has a tool an agent can call directly.