Hanako
@hanakoxbt • 12,640 subscribers
Building my name | ai researcher | dm is open
Shorts
Videos
a Citadel intern told me something at a party he probably shouldn't have it was on a rooftop in brooklyn. i mentioned i trade prediction markets. he got quiet for a second. "we have a model for that. it scores every contract on four factors. when all four align we enter. when any breaks we exit. that's it" i asked what the four factors are. he looked around. then said it fast like he was confessing. "cross-market divergence. disposition coefficient. capital velocity. pair network correlation" I didn't know what half of that meant. but i memorized it. went home. 11pm. opened Claude. "here are four scoring factors from a quant fund. build a terminal that runs all four on prediction markets" Claude asked one question: "Where's the data?" I sent him one repo: 86 million trades. every wallet. every entry. every outcome three weeks later i'm sitting in my apartment watching a screen i barely understand print money. the disposition meter alone changed everything. it measures how you exit - not how you enter. top wallets capture 86% of winner value and cut losers at 12%. everyone else captures 58% and holds losers to 41%. same exact entries. the exits make it a completely different game. capital velocity: 49x. every dollar gets recycled 49 times before the average trader recycles once. the terminal found 42 pair correlations across 11 markets. when MSFT beats Q3 is priced at 80c but the model reads 93% - it enters. when the gap closes 2 hours later - it exits. no opinions. no news. just four numbers that either align or don't. his fund runs this with a floor of PhDs and $800M AUM. my setup: > Claude - $20/month > VPS - $5/month > poly_data repo - free > Polymarket API - free $25/month. no team. no office. no Bloomberg. 280 trades so far. 70% win rate. $800 seed. four bots splitting the work: pulse_alpha +$299. arb_hunter +$558. trend_rider +$337. cal_engine +$719. +$11,514 total. copytrade here: he texted me last week. "delete everything i told you" too late.
Hanako1,739,160 Aufrufe • vor 3 Monaten
Anthropic Claude Code engineer: "If you're watching Claude write code, you're the QA tester. That's not what you're paid for." In 37 minutes he lays out how to get your keyboard out of the hot path entirely. The shift is /loop. You tell Claude to wake up every 10 minutes and babysit your PRs, and it just does it, while you're nowhere near the laptop. Routines do the same in the cloud, so the work keeps running with your machine closed. He caps it with remote control: any session, on any surface, driven from your phone. Watch the full talk, then grab the setup below.
Hanako689,129 Aufrufe • vor 1 Monat
your agent loop needs 8 exits. most people ship only one. (explained with triggers) 1) goal met → an evaluator scores the output against a rubric, and the run stops on a pass. → fires when the work is measurably done, not when the model says it is done. 2) turn cap → a hard ceiling on iterations, counted and enforced by the harness, not the prompt. → fires on the task it was never going to finish, before you pay to find that out. 3) budget cap → a limit on tokens or dollars, whichever one runs out first. → fires mid-run, which is exactly why it is the exit that saves you the 3am bill. 4) wall clock → a deadline on elapsed time, independent of how much progress was made. → fires when the run collides with a deploy window or the start of business hours. 5) no progress → hash the state every turn and compare it against the last few. → fires when three turns in a row change nothing. busy is not the same as moving. 6) human interrupt → an approval gate before risky steps, plus a kill switch that lives outside the loop. → fires whenever you decide, and it is the one exit the model cannot argue with. 7) error threshold → a counter of consecutive failures that resets on any success. → fires at n in a row, so it halts instead of retrying into the same wall all night. 8) external event → a webhook or a poll on whatever the task was actually about. → fires when the PR merged or the ticket closed and the work stopped mattering. a loop with one exit hangs. a loop with eight is a system. write the exits before you write the prompt.
Hanako181,005 Aufrufe • vor 19 Tagen
10 agent evals for AI engineers: (explained with usage) 1) golden set → a fixed set of cases you never edit, run on every single change. → use as the baseline that tells you whether anything moved at all. 2) llm as judge → a second model scores the output against a written rubric. → use when the answer is open-ended and there is no string to match against. 3) rubric scoring → one number per dimension: correctness, tone, safety, cost. → use when a single score hides which part actually got worse. 4) trajectory eval → grade the path the agent took, not only the answer it landed on. → use when the right answer for the wrong reason is going to bite you later. 5) tool unit tests → test each tool on its own, with fixtures, no model in the loop. → use always. most agent bugs are tool bugs wearing a costume. 6) regression suite → replay past runs against the new prompt or model and diff the results. → use before every prompt change, because prompts have no type system. 7) a/b in prod → split live traffic between two versions and compare outcomes, not vibes. → use when offline scores stopped predicting what users actually do. 8) human review → sample a slice of runs and have a person grade them honestly. → use to calibrate your judge, because a judge nobody checks quietly drifts. 9) shadow run → the candidate runs on real traffic in parallel and its output is shown to nobody. → use before a risky rollout, when one bad answer would be expensive. 10) red team → deliberately attack it: jailbreaks, injection, exfil, tool abuse. → use before anyone external can reach it, not after. offline evals tell you it works. online evals tell you it still works. both sides matter, but not all ten do. run the two that would have caught your last outage. save this. then read the full breakdown on loop engineering below.
Hanako149,853 Aufrufe • vor 18 Tagen
6 agent patterns for AI engineers: (explained with usage) 1) prompt chaining → split the task into fixed steps, each one checking the last. → use when the task decomposes cleanly and accuracy matters more than latency. 2) routing → classify the input first, then send it to the model or tool built for it. → use when inputs fall into distinct classes that need different handling. 3) parallelization → run several calls at once and merge them, either by splitting the work or voting on the same question. → use when subtasks are independent, or when one answer deserves several opinions. 4) orchestrator-workers → a lead model decides what the subtasks are at runtime, then delegates them. → use when you cannot list the steps in advance. 5) evaluator-optimizer → one model writes, another grades, the loop repeats until it passes. → use when you have clear criteria and iteration measurably helps. 6) autonomous agent → no fixed path. it plans, acts, reads feedback from the environment, and decides when it is done. → use when the steps are unknowable and you can afford the cost and the blast radius. the first five are workflows: you wrote the path. only the last one writes its own. most production systems people call agents are pattern 1, 2, or 5 with good error handling. this taxonomy is from Anthropic's own writeup on building effective agents. full breakdown in the article below.
Hanako143,891 Aufrufe • vor 20 Tagen
Anthropic engineer: "The agent doesn't remember anything. So we built a second set of agents whose only job is to dream about the first ones." They wait until you log off, then reopen every session you ran, fact-check the first agents, merge the duplicates, and burn anything that went stale. One dreaming agent per session, up to 100 at once, built to miss nothing. 95% of the tokens come back cached, so an entire memory rewrite costs almost zero. This is the part of the workflow nobody is talking about yet. Watch the full workshop, then save the setup below.
Hanako334,934 Aufrufe • vor 2 Monaten
one model built ten webpages in 160 seconds. the next one took 623. same prompt to three models at once: ten standalone HTML pages, one per footballer, shared template and CSS, each self-contained. Ling-3.0-flash: 160s Step 3.7 Flash: 319s MiniMax M2.7: 623s first reasoning token came back in 84ms. the other two took 378ms and 8.7s. Ant Ling shipped it at 124B total, 5.1B active per token. one eighth the parameters of their own 1T flagship, and it matches or beats it on most benchmarks they published. the whole race takes 20 seconds to watch.
Hanako34,332 Aufrufe • vor 8 Tagen
an ex-Citadel quant told me a salary is a short position on your own time we were at a dinner in new york. friend of a friend. he asked what i do. "software engineer. $4,200 a month" he said one simple thing "you're mass-selling the most valuable asset you have. your hours. at a fixed price. with no upside. that's a short position on your own life" i didn't have a response. then he showed me his phone. a terminal. live trades. +$39,453 running for 6 weeks. "i built this with Claude Opus 4.7 and one open source dataset. took a weekend. haven't touched it since" i asked how. "gave Opus one repo and one prompt. find who wins on prediction markets. find why. copy their edge" 2,900 stars. 36GB. every trade on Polymarket and Kalshi. every wallet. every resolution. Opus read the whole dataset friday night. by sunday it had a live terminal scanning 1,400 markets per hour. 8 detectors running in parallel. entering when void score exceeds 90. exiting when it drops below 75. no human override. > Senate filibuster. locked at 42c. now 98c. +$1,133. > Neuralink approval. entered 52c. now 98c. +$1,396. > Fed 0% rate. entered 35c. now 97c. +$1,147. > BTC 250K. entered 8c. now 98c. +$1,258. > Megaquake. entered 6c. now 97c. +$1,039. the part that broke something in me: latency arbitrage. Polymarket updates in 19ms. Kalshi in 63ms. the bot trades the 44ms gap between them. automatically. every time. "we had four people at Citadel doing what your bot does with latency alone" copy mirror tracking 6 positions: > Senate fili +$1,252. void signal. 10%. > BTC 250K +$958. zero signal. 2%. > Neuralink +$1,384. sign signal. 9%. > Fed 0% +$863. fill signal. 8%. > Megaquake +$1,070. bridge signal. 18%. > META AI +$979. gas signal. 5%. 972 trades. 83% win rate. kelly f+ 0.101. avg profit $10.65. drawdown -1.4%. maker rebate tracker collecting $203 across four platforms while the bot trades. Polymarket +$114. Kalshi +$40. dYdX +$28. Betfair +$21. $1,800 seed. +$39,453. i quit the next morning. one email. "i'm done. thank you for everything." copytrade setup: texted him a week later. "you ruined a perfectly stable career" he replied "no. i closed your short position"
Hanako379,725 Aufrufe • vor 3 Monaten
your agent reviewing its own work is not a check. it is a second opinion from the same source. this is the most common gap in agent systems and it hides in plain sight, because the step exists. there is a review. it just cannot do the thing you think it does. here is the mechanism. the model produced an output from a context. you then ask the same model, holding the same context, whether that output is correct. it answers fluently, because that is what it does. and the answer is drawn from the same distribution that produced the thing being judged. same weights, same window, same blind spots. if the reason the output is wrong is something the model does not know, the review does not know it either. if the reason is something the context does not contain, the review has the same context. the failure mode and the detector share a cause. > why it feels like it works because most of the time the output is fine, and the review says fine. agreement is not evidence of detection. a reviewer that says pass on everything agrees with reality most of the time too. what you actually want to measure is what happens on the cases that are wrong. that is the only place a check earns its name, and it is exactly the place where a self-review is weakest. there is research on this. Huang and colleagues at DeepMind showed at ICLR 2024 that intrinsic self-correction, revising without external grounding, does not reliably help and often makes things worse. > what to actually do move the check outside the model. a test that runs, a schema that validates, a file that exists or does not, an exit code from something you did not write. these are not smarter than the model. they are just not correlated with it, and that is the entire value. when the judgement genuinely needs a model, at minimum use a different family. same family means shared blind spots, and frontier judges measurably inflate scores for outputs that look like their own. and split the work by kind. anything objectively checkable goes to code. only the genuinely semantic calls go to a judge, and those get a rubric written as one line. a review inside the loop tells you the model is confident. a check outside it tells you whether the work is done. save this - then read the eval setup below
Hanako13,683 Aufrufe • vor 3 Tagen
an ex-Jane Street trader told me a paycheck is the most overpriced contract there is we were at a rooftop bar in tel aviv. friend of a friend. i told him i work in tech. $4,200/month after taxes. he showed me his phone. a terminal. green numbers. +$47,000. last 30 days. "Claude and one GitHub repo. that's the whole setup" i asked what he means. "copy-trading. but not how everyone else does it" he explained it fast. like he'd said it a hundred times. "most people copy wallets by win rate. useless. a wallet with 85% wins can still lose money. what matters is disposition ratio" he pulled up a wallet. 91% win rate on crypto. 14% on politics. "copy everything - you bleed. copy only their crypto - top 1% of the platform" i went home. couldn't sleep. opened Claude at 2am. "build a copy engine. filter by disposition ratio above 0.70. copy only dominant category per wallet. kelly sizing" gave it this: almost 1k stars. execution engine. order book logic. position sizing. open source. Claude rebuilt the whole system in one night. first week: +$3,200. second: +$2,900. third: +$2,100. fourth: +$1,648. +$9,848 total. 313 trades. 80% win rate. avg hold 5h. the wallets it copies now aren't the highest win rate. they're the best exits. 0xe41f8...b has $43.4K profit. tech only. disposition 0.82. 0xa22c3...d has $40.2K. crypto only. disposition 0.79. 0xf88d2...c has $22.4K. macro only. disposition 0.74. category P&L: > crypto +$974 > weather +$749 > politics +$409 > macro +$353 > sports -$50. killed it capital velocity: 51x. kelly f+ 0.073. drawdown -2.9%. copytrade: saw him two weeks later. showed him the terminal. he scrolled through it. "you built this in one night?" Claude did. "my old desk had 14 people on this" mine costs less than a Netflix subscription. he nodded. "quit your job" i'm thinking about it.
Hanako310,972 Aufrufe • vor 3 Monaten
a Goldman Sachs engineer sat next to me on a delayed flight from JFK four hours on the tarmac. nowhere to go. we started talking. i told him i trade prediction markets with Claude. he almost choked on his ginger ale. "you do WHAT with an AI?" i showed him my terminal. he went quiet. scrolled through it for two full minutes. "this is close to what we run internally. how did you get the scoring model?" i told him. one GitHub repo with 2,600 stars. Claude connected the dots. official Polymarket framework. LLM-powered trading agents. open source. free. i gave Claude the repo and said: build me a grid that scans 700+ markets per hour and enters when the edge exceeds 9 cents. he didn't ask twice. the Goldman guy kept scrolling. "you have a disposition meter. a kelly gauge. a correlation matrix. we spent $4M building this infrastructure" i told him my total cost. Claude $20/mo + VPS $5/mo he laughed. not the funny kind. that was three weeks ago. today the terminal shows: +$16,976. 439 trades. 73% win rate. avg hold 7h. > Falcon Heavy - mkt 68c. model 95c. +$716. > MSFT antitrust - mkt 59c. model 77c. +$670. > Blizzard NY - mkt 74c. model 85c. +$772. > BOJ rate hike - mkt 52c. model 60c. +$422. four bots split everything: - pulse_alpha +$299. - arb_hunter +$558. - trend_rider +$337. - cal_engine +$719. kelly fraction 0.118. correlation matrix tracking 6 asset pairs. when one breaks - something is wrong. the grid trades it. copytrade here: he DM'd me yesterday. "i showed your screenshot to my desk. they want to know who built your infra" i said Claude. he said "seriously" i said seriously.
Hanako286,695 Aufrufe • vor 3 Monaten
Anthropic Quant Andrej Karpathy to Andrew Ng: "I wrote my own neural net library in JavaScript. Not to ship it - just to see backprop with my own eyes" before any framework, he built convnetjs from scratch - every layer by hand. his advice: skip TensorFlow at first. write it yourself at the lowest level, then pick up the framework once you know what's under the hood. the line that lands: "if all you know is a stack of layers and that dropout helps, you can't debug or improve the model." Watch the talk, then save the loop guide below.
Hanako61,277 Aufrufe • vor 24 Tagen
I asked Claude to build me a MiroFish God View terminal it showed me 56 live agents that simulate real world each agent has memory, personality, and behavior they form groups. develop leaders. shift opinions in real time I feed it a scenario: "Fed cuts rates by 50bps" 56 digital humans reorganize in seconds retail agents panic buy YES on Polymarket. institutional agents fade the move opinion leaders wait the terminal maps every shift live throughput: 5,000 ops/s latency: 23ms 2.1M tokens processed per cycle zero errors I watch the consensus form before Polymarket prices it > monday: injected "US-China military standoff in Taiwan Strait" agents split instantly - 39 rushed YES on "conflict before 2027" bought YES at $0.08 before the herd moved Polymarket moved to $0.29 by tuesday +$1,840 > wednesday: injected "S&P500 drops 3% intraday" 48 of 56 agents shifted bearish in 9 seconds loaded YES on "S&P500 below 5000 by July" at $0.12 contract hit $0.31 by friday +$2,470 > sunday: injected "Trump announces new tariffs on EU" herd effect kicked in - retail agents mass-bought YES institutional agents faded at peak I followed the institutions. bought NO at $0.22 settled at $0.71 monday morning +$3,048 total: $7,358 in 7 days from scenarios that haven't happened yet the god view doesn't predict price it predicts how people bet and people move the odds one terminal. 56 agents. zero team copytrade here: you don't need to predict events you need to predict how Polymarket reacts to them
Hanako304,401 Aufrufe • vor 4 Monaten
a drunk Jane Street quant drew his trading system on a napkin it started as a joke. someone's friend at a bar in lower manhattan. he mentioned prediction markets and this guy lit up. "i do this for a living. for a firm you've heard of" two rounds later he was drawing on a napkin. boxes. arrows. labels. "the crowd repeats six mistakes every single day. we just built detectors for each one" > base rate neglect > mispriced conditionals > thin book slippage > platform divergence > favourite bias > time decay "that's the whole system" i asked what it costs to run. "200 quants. proprietary infra. about $2M/year just to keep it on" i laughed. photographed the napkin. went home. 3am. couldn't sleep. opened Claude. fed it the photo and one GitHub repo - poly_data. 86 million trades. every wallet. every entry. said: build detectors for all six. by 4AM my laptop fan was screaming. 400+ markets scanning per hour. 8 agents deployed. terminal live. i didn't understand half of what was on screen. but the P&L number kept climbing. +$312 by breakfast. +$1,900 by end of day one. +$7,400 by end of week one. +$16,474 right now. 383 trades. 81% win rate. the scanner finds gaps like: > AAPL earnings - crowd says 77%. data says 91%. > Gold $2600 - crowd says 44%. model says 60%. > Senate filibuster - crowd says 68%. base rate says 82%. enters. waits 6 hours average. exits when the gap closes. politics at 93%. weather 79%. crypto 72%. eight agents split the work. one mirrors whales. one catches Binance lag. one fades favourites. one bleeds theta on dying contracts. his firm runs this with 200 people. mine runs on $25/month. copytrade here: texted him the screenshot yesterday. long pause. "you're not supposed to be able to do this" i know.
Hanako244,370 Aufrufe • vor 4 Monaten
i told Claude to scan every Polymarket wallet and GitHub repo woke up to +$3,582. the terminal was still running on my screen. it analyzed thousands of wallets overnight. most of them bleed money. but 340 had something different - asymmetric payouts. entering at 27 cents. exiting at 91. losing 27 cents when wrong. winning 64 when right. you only need to be right 1 in 4 times. they were right half. Claude found 3 patterns: category specialists - one wallet: 91% wr on crypto. 14% on politics. total looks average. filter to crypto only - top performer. speed arbitrage - 47 wallets entering 3-8 seconds after Binance moved. before Polymarket updated. not predicting. just faster. near-zero accumulation - buying at 2-8 cents weeks before resolution. $8 in, $200 out. hundreds of times. it connected GitHub repos to run all three: > poly_data - scanned every wallet and entry price. > polyterm - whale tracking and insider detection. > py-clob - placed the orders. 7 wallets copied. 204 trades. $25 seed. +$3,582. 6 agents running: > sigma_decay - 78% conviction > darkpool_7 - 87% > velvet_void - 85% > profitprinter - 54% the equity curve tells the whole story. $25 to $3,582 without a single dip worth mentioning. sharpe 2.09. copytrade here: 87% of Polymarket is exit liquidity. the question is which side you're on.
Hanako252,006 Aufrufe • vor 4 Monaten
a Palantir data scientist saw my trading terminal from across a hotel lobby in manhattan i had my laptop open. wasn't hiding it. entropy gauge pulsing. cryo scanner running. live P&L climbing. she was checking in. stopped mid-sentence at the front desk. walked over. "that's a signal intelligence grid. who do you work for" i told her i don't work for anyone. i built this with Claude Opus 4.7 and one GitHub repo. she sat down on the couch next to me. "i build these for defense clients. anomaly detection. propagation analysis. entity clustering. we charge seven figures annually" she asked to see the screen. i turned the laptop. the cryo scanner had just frozen a market. entropy below 1.86. price flat. volume dead. but underneath three fresh wallets were accumulating the same obscure contract. Platinum $12. nobody trades platinum on Polymarket. "that's a latent coordination detector. we have a classified version. yours is running on what?" $25 a month. Claude plus a VPS. she didn't respond to that. i told her Opus 4.7 built the whole system from this: 29,500 stars. agentic coding framework. multi-agent orchestration. skill system. plugin architecture. i gave it one prompt friday night: build 8 parallel detectors for prediction market anomalies it read the framework. spawned sub-agents for each detector. entropy. mempool. topology. propagation. cluster analysis. flow pressure. oracle lag. cryo freeze. live by saturday morning. first trade by noon. she pointed at the topology map. "you're tracking price deltas across four platforms simultaneously. Polymarket. Kalshi. Betfair. oracle feeds. in real time" i nodded. "with latency under 200 milliseconds" 31ms on the Polymarket API. 54ms on oracle. 14ms on chain. copy mirror tracking 6 positions: > NK summit deal +$788. entropy signal. > Lunar gateway +$918. cluster signal. > Typhoon Pacific +$889. mempool signal. > BTC 150K Mar +$661. topo signal. > ECB cut Dec +$587. propagation signal. > AMZN split +$721. cryo signal. 516 trades. 78% win rate. avg hold 3h. sharpe 2.86. +$19,796 from $1,400 seed. 4 weeks. copytrade setup here: she closed my laptop lid halfway and looked at me. "we bill $1.2M a year for a system that does less than this" i said i know. she handed me her card. "when your AI builds something for defense let me know" it already did. she just saw it on a prediction market instead.
Hanako223,812 Aufrufe • vor 3 Monaten
context engineering vs graph engineering. every few months the list gets a new word and everyone treats it as a replacement for the last one. these two are not on the same list. one decides what the model sees this turn, the other decides what exists at all. the cleanest way to tell them apart is to ask what a single unit of work looks like. > context engineering is the window the window opens empty, every single time. you assemble what goes in it. the prompt, the docs, the history, the tool results. the assembling is the work. the window only grows. it never shrinks on its own, so eventually something gets dropped. usually from the middle. usually without telling you. then the turn ends and the window is thrown away. not archived, thrown away. the next turn opens empty again and you re-explain what you already explained. good context engineering is knowing what to leave out, not what to pack in. the unit of work is one window. > graph engineering is the structure the same material arrives from the same sources. instead of packing it into a window, you pull entities out of it, resolve the duplicates into one node, and write typed edges between them. nothing here is stored as text you hope to find again. it is stored as a thing with a name and its connections to other things. when the turn ends, the graph is still there. the next turn does not start from zero. it starts by querying what already exists, and the query walks edges instead of guessing at similarity. good graph engineering is deciding what counts as the same thing twice. the unit of work is one relationship. > they are not alternatives the graph is what refills the window. context engineering decides what fits. graph engineering decides what there is to choose from. remove the graph and every session starts blind. remove the context work and the best structure in the world arrives as an unreadable dump. that also tells you which one broke. the answer drifted from what you actually said, or forgot something from this same session. that is the window. the answer is coherent but invents a connection that does not exist, or cannot join two facts it has clearly seen. that is the structure. people debug the prompt because the prompt is the easiest thing to edit. it keeps taking the blame for failures that live a layer down. save this - then read the full breakdown below
Hanako19,089 Aufrufe • vor 10 Tagen
my Claude built a Liquidity Sniper terminal that eats market overreactions +$3,082 while I was asleep. thin books, fat profits 251 snipes across 500 markets in under two hours every one triggered by low liquidity and a price that overshot too far the logic: when liquidity is low, prices overshoot. when prices overshoot, they revert. the bot sits and waits for exactly that moment. Hurricane FL - everyone panic-sold when the forecast dropped. book went thin. bot entered at $0.35. reverted to $0.42 in 11 minutes. +$287. Trump OH - breaking news spike. price shot up on emotion. bot caught the overreaction. entry $0.59. reverted. +$194. Senate bill - market moved on a rumor. zero liquidity on the ask side. bot sniped at $8.45. reverted 112%. +$371. avg revert time: 17 minutes. bot enters, waits, exits. nothing else. the spike feed doesn't stop: mech Miami flood +15.5% - SNIPE info Dogecoin +7.9% - SKIP mech eth>5k Q2 +5.8% - SNIPE SKIP means liquidity is fine. SNIPE means the book is thin and the price is wrong. > markets scanned: 500 > scan rate: 30sec > win rate: 88.8% > active snipes: 251 P&L curve - no dips. just stairs going up. bot doesn't predict events. it reads liquidity. when it's thin - it strikes. copytrade: the market overreacts. this bot is already there waiting.
Hanako181,205 Aufrufe • vor 4 Monaten
a drunk Chinese friend who works as a quant slipped me a napkin with 8 words on it we were at a rooftop in tel aviv. he was three cocktails past the professional filter. i mentioned prediction markets. he laughed. grabbed a napkin from the bar. scribbled in english. "disposition. category. velocity. cross-market. news. volume. whale. theta." i asked what it means. "eight signals. every mistake on these markets is one of them. we run detectors in parallel. enter only when three or more fire" i asked how long it took to build. "two years. fourteen quants" i asked if i could just do it myself. he laughed harder. "with Claude? tonight probably" i don't think he meant to say that part. he looked at the napkin like he regretted writing it. too late. i pocketed it. went home. opened Claude at 11PM. typed the eight words. "build a parallel detector grid on prediction markets. 8 signals. enter only when 3 or more fire simultaneously" Claude needed a backbone. i gave it: 610 stars. market making bot. order book execution on both sides. position sizing. the foundation for anything that needs to place real orders fast. terminal was live by 2AM. the radar lights up 14 times per hour. most signals fire alone. rare ones stack. last week: > BRICS summit - 74c. 4 signals. +24c in 3h. > Artemis launch - 63c. 5 signals. +33c in 5h10m. > Helium-3 demo - 68c. 3 signals. +14c in 48m. > Waymo public - 67c. 4 signals. +8c in 6m. sentiment feed reads 4 platforms: - X/CT: 56% bearish - Reddit: 53% - Telegram: 17% - Discord: 75% when sentiment diverges from flow pressure by 20%+ the contrarian signal fires. 83% win rate on those. cross-market delta catches gaps: BTC 92K vs Kalshi 7.7c Fed cut vs PredictIt 7.2c AAPL>180 vs Manifold 5.7c oracle delta tracks NOAA, AP, CoinGecko, Reuters. when a feed moves before the market adjusts, the bot enters on the lag. 6 wallets copied by dominant category: > 0xe41f8...b2 - $483 profit. 83% wr. > 0xa22c3...d9 - $890. 67%. > voltwhale - $593. 55%. > 0xf88d2...c4 - $830. 73%. 412 trades. 76% win rate. avg hold 5h22m. best $232. worst -$37. 9 consecutive wins. +$12,640. from a $900 seed. 4 weeks live. copytrade setup: i haven't seen him since that night. the napkin is still in my drawer. i don't think he remembers writing it.
Hanako156,731 Aufrufe • vor 3 Monaten
i fed Claude 5 PhD formulas and asked him to build me a terminal he didn't ask questions. he built MiroFish 274 agents. 4 quant formulas running live each one doing what 87% of polymarket traders can't every 5 seconds the terminal does this: > scans polymarket contracts > runs bayes update on every new signal > calculates EV against market price > sizes position through ¼ kelly > checks KL-divergence across correlated markets for arbitrage no opinions. no "i feel like YES is underpriced" just math that PhD students publish and hedge funds lock behind NDAs here's what happened in 14 days: > day 2: bayes picked up OSINT chatter on iran negotiations prior 0.31 → posterior 0.58 in three updates bot bought YES on "ceasefire by Q3" at $0.33 kelly sized it at 6% of bankroll contract moved to $0.61 by day 5 +$2,180 > day 6: KL-divergence flagged a gap "candidate X wins primary" priced at $0.70 "candidate X wins general" priced at $0.48 historical base rate says general should track at ~62% of primary bot bought general, hedged with primary convergence hit by day 9 +$3,740 > day 9: EV scanner found a weather contract market priced hurricane landfall at $0.22 model said 41% based on NOAA data EV = +$0.86 per dollar risked kelly said 11% allocation landfall confirmed day 12 +$4,890 > day 11-14: base rate engine running quiet fed meeting contract at $0.65 for "hold rates" base rate: fed holds when unemployment < 4% → 74% of the time unemployment at 3.8%. market underpriced by 9 points bought at $0.65. settled at $0.98 +$4,663 total: $15,473 in 14 days not from predictions. from formulas 87% of polymarket wallets lose money because they trade what feels right the top 1.2% trade what the math says MiroFish doesn't read twitter threads it reads probability distributions 274 agents don't have opinions they have bayesian priors every 15 seconds the NEXUS core sends a pulse to all agents they recalculate. reposition. repeat i just watch the profit tick copy the bot here: you don't need to be a quant you need a quant's formulas running 24/7
Hanako159,042 Aufrufe • vor 4 Monaten