DisruptiveBrisket (@Bankkomin) profile picture
Joined September 2009
Photo by DisruptiveBrisket (@Bankkomin)
DisruptiveBrisket retweeted
I had Claude code strategies from 10,000 Substack posts. Claude researched 900 strategies before hitting the weekly limit, including strategies that didn't survive. Everything was tested on QuantConnect using survivorship-bias-free data. I now have two DARWINs ready: one for stocks and CFDs, and another for ETFs. They're largely uncorrelated with RIAG, with a correlation of just 0.25, and they've performed extremely well this year, especially the stock portfolio, which returned over 23%. Now that I think about it, you could take this even further. One of the criteria could be to investigate the author before selecting any of their publications. Have they worked for a fund in the past? Do they have a verified track record that's long enough to establish credibility? You could use these criteria to filter authors before even attempting to automate their strategies. FTMO will start making money too.
Substack might be the best source of alpha. Claude has found over 12,000 posts so far and automated 149 of them, attempting to replicate the authors' results. It's pretty crazy. The less efficient the market and the more heuristic-driven the instrument, the higher the potential Sharpe ratio. But the harder it becomes to find a way to actually trade it.
10
3
2
106
29,968
DisruptiveBrisket retweeted
the overnight range has 94% winrate on ES 🔥 When the 9:30 market opens inside Overnight range there is a 94% probability that price will take out either the range high or the range low proven by 1 years of data by Edgeful
7
11
2
77
10,213
DisruptiveBrisket retweeted
Almost everything has converged to AWS Tokyo. Even Okex moved from Alibaba HK. Few outliers still. Save some time: Binance : ap-northeast-1 Coinbase : us-east-1 Bybit : ap-southeast-1 OKX : ap-northeast-1 Kraken : Equinix London Bitget : ap-northeast-1 KuCoin : ap-northeast-1 Gate.io : ap-northeast-1 HTX : ap-northeast-1 Crypto.com : ap-northeast-1
11
11
1
172
18,099
DisruptiveBrisket retweeted
All second intervals and tape mode are now free on OpenMarket. We're giving you 1s, 5s, 15s and 30s candles on every major crypto exchange, with tape mode on top. Other sources will follow shortly. Tape mode is the order book and every trade, live, one second at a time. Zoom in and you see every single order in the book.
51
48
5
739
41,328
DisruptiveBrisket retweeted
another trading agent outperforming the market +$2,478 in 7 days sharing strategy below (NFA!)
35
16
3
278
40,494
DisruptiveBrisket retweeted
TradingView'ın kendi trending listesinde şu an #6'dayız. Açık ve ücretsiz faydalı çalışmak yaratmak benim sorumluluğum. tradingview.com/script/Tm1cG… Bu başarısını kullanan, test eden, paylaşan ve destek veren herkese veriyorum. Umarım gerçekten de faydalı olmuştur. Bu ilk adımdı. Dahası yolda: tradingview.com/spaces/ata_s…
7
22
203
19,847
DisruptiveBrisket retweeted
Jane Street fired this guy who leaked their method to build regime aware trading agents. he openly shared this guide on reddit that got 40k upvotes. Bookmark this before its lost.
17
88
3
467
87,921
My 0DTE selling bot is now running every trading day. So far, measured live slippage has been very close to my original assumptions. My current workflow: • Download historical option data from Databento • Backtest actual option contracts with Python • Run the same logic live through IBKR • Match live trades against the backtest • Measure the difference in fills and slippage • Use an LLM to reconcile the logs and prepare a detailed daily report The dark blue line is the lower risk, more active version I currently trade, recalibrated using the mean difference observed in my live fills: • Sharpe: 1.43 • Profit factor: 1.26 • Win rate: 68.4% • Fees and calibrated slippage included • No compounding I also plot a more pessimistic slippage assumption. Under that scenario, the Sharpe falls to 1.21. The strategy itself is simple: • Sell 0DTE verticals after the market moves lower, when option premiums are richer • Let roughly 70% of positions expire • Manage fast upside reversals, which are usually where this sleeve loses That loss profile is useful at the portfolio level. A fast rebound can hurt this strategy while helping several of my other, mostly long, positions. I am gaining confidence to increase the position size gradually, but the calibration is still ongoing. I need more live observations before treating any assumption as settled. This chart is a calibrated backtest, not live P&L. Everything 100% automated. All orders are limits. The encouraging part is that another estimated input is being replaced by measured live evidence.
20
14
223
21,842
DisruptiveBrisket retweeted
this paper is f*cking insane a quant paper uncovered a factor with a 13+ Sharpe out-of-sample by conditioning purely on market regimes. the numbers: 158.6% annualized returns, 12.0% volatility, -11.9% max drawdown, tested across 20 years of S&P 500 data. the crazy part is this factor isn't always active, it only kicks in during specific drift regimes. most factor models chase what works. this one chases when it works. bookmark before this thread gets buried.
12
39
1
373
110,555
AI-enabled 0DTE Decision Support System Achieved: Claude provides synchronized summaries of Cboe-sourced market-maker hedging flow data ... within seconds of the data's appearance on a live, running html-based dashboard. The System 1. Pull Cboe options data (peak concentrations for calls, puts, GEX-related measures, and SPX spot) ↓↓ 2. Process 10-minute interval snapshots of market maker-informed positioning in SPX 0DTE trading ↓↓ 3. Assemble and present the data as an information read in an auto-refreshing HTML dashboard (image below) →→ Audible alerts sound to announce significant changes in the auction market process (silent during price noise) ↓↓ 4. Within 15 - 20 seconds of the completion of signal processing, Claude presents both a detailed and a summary contextual interpretation of each interval's hedging data. Claude throws light on regime, trader sentiment, changes in key reference levels ... and under specific conditions ... buy/sell signals. ↓↓ Symbiotic Execution - In Summary 5. A python engine and a chatbot supply order-flow analysis, reference levels, and target zones culled from Cboe-sourced, market maker-informed options data. The trader puts it all together — follows the live flow-sheet, considers Claude's take on hedging pressure, reads the tape, and determines timing and order placement. Next Up: Hearing Claude's analysis of the options data read aloud, to enable staying focused on the trading platform.
16
31
2
247
56,973
DisruptiveBrisket retweeted
Citadel pays $650,000 a year for quants. MIT wrote the exact bible to get there & released it for free. 51 pages. Zero to quant. Probability, stats, market making, real interview questions from Jane Street, Citadel, Two Sigma & more. Bookmark and Give it to your agents
26
34
250
25,818
DisruptiveBrisket retweeted
Jane Street, AQR, Ren Tech... All use volatility. Retail was locked out... until now. A 327 page PDF was just released. For free:
7
51
1
339
30,711
DisruptiveBrisket retweeted
Day 15 — Friday August 14, 2026 7 automated strategies, 1 account. Daily P/L: +$252.85 Since Day 1: +$4,489.14 Green or red, posted every day. $ES $NQ $GC $BTC #algotrading
1
1
14
1,113
One indicator. That’s all I need to build a complete breakout strategy from scratch. ATR handles the entry. ATR handles the filter. ATR handles the exit. Here’s exactly how: Entry > Open of the bar + 2.5 × ATR(20). > A hard, calculated breakout level. No guessing. Price either hits it or it doesn’t. Filter > Compare short-term ATR to long-term ATR. > When short-term is below long-term (volatility contraction), the market is coiling - energy is building. These are the higher-quality breakouts. > When short-term is above long-term (already expanding), the move is more likely exhausted. Sit out. > One comparison. Big reduction in false breakouts. Exit > Flip the entry math. > Use a fraction of ATR to set the stop distance from entry. > The market itself decides how much room the trade needs based on current volatility. That’s it. One tool measuring the same thing across all three decisions: how much the market is actually moving. Most traders only use ATR for the stop loss and leave the other two-thirds of its power unused. I don’t. I use it for the entire strategy.
10
44
2
373
34,258
DisruptiveBrisket retweeted
Yesterday I shared an SPX put-selling strategy -- 95.4% win rate, profit factor 2.48, no losing years. People kicked the tires and found real flaws. They were right. So I bought 8 more years of data and re-ran everything. It changed the strategy. 🧵
29
15
3
342
85,034
DisruptiveBrisket retweeted
Even though nobody guessed it, I'm sharing it anyway: Sell one SPX put credit spread per week at the Multi-Day Saty ATR -1 level. That's the whole strategy. 326 backtested trades since May 2020 95.4% win rate Profit factor 2.48 +$53,328 on a single contract Max drawdown $4,605 Zero losing years, 2026 is 32 for 32 Everything below, free.
This is not made up -- if anyone can guess what this strategy involves, I will share it freely 2026 had 32 trades, all 32 won -- profit factor is literally = infinity
138
199
15
3,181
739,814
DisruptiveBrisket retweeted
Anthropic engineer: “90% of our engineers were using self‑improving loops. Now everyone shifted to building agentic Graphs" “No more prompting.” In just 10 minutes, she builds her entire Claude Code setup and workflow live from a blank terminal. This is more valuable than most $1000 agentic courses. Watch this video, Then save the article below if you want to become a graph architect before everyone else catches up.
33
166
12
1,593
430,803
DisruptiveBrisket retweeted
KARPATHY JUST KILLED THE PROMPT ERA WITH A SINGLE DOCUMENT prompts are easy. loops are hard. and writing fifty prompts a day is the work nobody does twice. he shifts the burden to the harness. you define the contract once. the model writes, reviews, restarts, and reconciles. you keep judgment. it keeps the loop. the throughline is the same in every rule: the human owns the spec and the boundary. the model owns the execution and the bookkeeping. planner never touches code. generator never grades itself. state lives on disk, not in context. 9 rules. start with one feature, not ten. most people are still typing prompts. this turns Claude into an agent that finishes the job on its own. here is the official document from Karpathy explaining the architecture
Readers added context they thought people might want to know
This document is not from Andrej Karpathy; he has not shared or authored it on X or GitHub. x.com/karpathy github.com/karpathy
75
361
32
2,633
667,884
DisruptiveBrisket retweeted
I found 7 free Polymarket trading bots on GitHub for 7 different trading situations… Each of these bots comes with a detailed step by step setup and usage guide in English: 1. This bot includes 118+ ready to use strategies and tools for trading on prediction markets (Momentum, Binance-Polymarket latency, Penny Clipper, Smart Routing, Expiry Fade, DCA bots and more). Built by a Cambridge computer science student who won a hackathon with this trading bot. GitHub: github.com/alsk1992/CloddsBo… 2. This bot automatically manages all your Polymarket limit orders to maximize liquidity rewards. GitHub: github.com/lihanyu81/polymar… 3. A weather bot from a Chinese dev, that analyzes multiple sources in real time, like forecasts, airport data and aviation observations (METAR + SPECI) to get the most accurate temperature data and generate a detailed weather report for a specific city and day. GitHub: github.com/yangyuan-zhen/Pol… 4. A bot that automatically searches for arbitrage opportunities between Polymarket and Kalshi. GitHub: github.com/Composio-HQ/polym… 5. This is a bot-toolkit that includes copy trading, arbitrage, market making, whale alerts, spread farming, sports trading and more… GitHub: github.com/HarrierOnChain/Pr… 6. A smart money trading bot - it looks for the most successful traders in selected markets, filters them by Pnl + win rate, and then creates a list for automated copy trading. GitHub: github.com/MrFadiAi/Polymark… 7. A large collection of 20+ free trading bots for prediction markets. GitHub: github.com/aarora4/Awesome-P… Every bot here has a Dry Run mode, so you can test it on real markets without risking any funds.
23
56
1
354
42,603