Ali H. Askar

this is f*cking Legendary build a regime detector using filtered probabilities that eliminates lookahead bias while your backtest looks perfect (Full Framework) if I had this a year ago, I would've caught the lookahead bias on day one instead of chasing ghosts for six months bookmark before the thread gets buried

"Lecture Notes: Convex Optimization" (by Andreas Habring): arxiv.org/abs/2607.11664

Every crypto matching engine already generates Level 3 (market-by-order) data. Almost none of them sell it. I pulled the Binance, OKX, and Kraken API docs to figure out why. The answer splits into three separate questions. A breakdown: #marketmicrostructure #cryptotrading

With the context set, if you are curious about spinlocks, you MUST read this. A classic John Regehr blog.regehr.org/archives/2173

Great overview of Parallel C++ about concepts such as false sharing, double buffering, spinlocks, and much more. youtube.com/playlist?list=…

C++ · Physics · Calculus — Integrated Stem Bundle for CS students creatifystoredp.gumroad.com/l/mbpoy

A fantastic read blog.weineng.me/posts/slowest_…

"Lecture notes on random matrix theory: the results, the applications, and the analytical tools" (by Joseph W. Baron): arxiv.org/abs/2607.07868

AI didn’t eliminate programming. It revealed what programming was really about. The code was never the point. Thinking was. My first Thinking with AI episode is live. youtube.com/watch?v=cEoxTg… #AI #SoftwareDevelopment #Programming

hmm... when retail is extremely long, short. when extremely short, long?

A visual introduction to information theory arxiv.org/abs/2206.07867

Binance is arguably the world's largest crypto exchange. Feb 2026: Polymarket lists a small 5-minute Bitcoin contract. Since then, Binance order flow spikes in the final seconds before settlement; prices revert soon after. New paper with David Dai and Shihao Yu ( @ShihaoY ): Settlement Manipulation in Prediction Markets (papers.ssrn.com/sol3/papers.cf…)🧵 The finding, up front: Traders push Bitcoin's price in the final seconds to decide the contract. That push makes the price less informative, yet Bitcoin's market more liquid. Market makers are largely insulated; ordinary traders lose $7.6M in two months.

1/ Everyone knows XTX and Gerko. But there’s a whole tier of Russian-rooted trading firms no one is yapping about online. Some are quietly elite, some are cautionary tales. A field guide 🧵

A year ago today, the best model we had was GPT o3. How quickly has intelligence leapfrogged in just the span of a year's time. Conventional, human-written code is officially dead. You can basically assume that a year from now, no kinks from agentic engineering today will remain, as they will be completely eradicated by models 10x smarter. Human-shaped bottlenecks in agentic engineering will be all that remain, and agentic engineering will be the new skill ceiling for software engineers.

“Calculus in Context” science.smith.edu/~callahan/intr… Download the entire 869-page Mathematics PDF ebook +plus+ more!

todays read is vasily volkov's phd thesis on gpu latency hiding. it's quite long, so i think it'll keep me occupied for the next few days. www2.eecs.berkeley.edu/Pubs/TechRpts/…

Lots of really good fundamental work in today's Moontower edition H/T @KrisAbdelmessih. Take some notes and work through it in excel/python/R over the long weekend. 🫡

The above referenced book is the textbook. Here is the encyclopedia => "Mastering Modern Time Series Forecasting: A Comprehensive Guide to Statistical, Machine Learning, and Deep Learning Models in Python" Complete encyclopedic reference goes in-depth on every method, extensive benchmarks, & advanced topics: amzn.to/41RPK25

The uncomfortable truth about Forecasting methods: valeman.medium.com/why-im-writing… Quotes from the article: "The forecasting establishment spent decades protecting its territory. The result is a generation of practitioners who were taught that statistical methods are inherently superior to machine learning for time series, that uncertainty quantification means Gaussian prediction intervals, and that the M-competitions settled the debate." "None of this is true. And the cost of believing it — in bad forecasts, wasted compute, and missed opportunities — is real." "This book is an attempt to set the record straight. Not with opinion, but with evidence, code, and proper evaluation."

This paper outlines some interesting patterns for HFT. arxiv.org/pdf/2309.04259