BOOM 💥 Big DeepVest update Connect your brokerage — Robinhood, Fidelity, IBKR, Binance, Coinbase & more — or upload a CSV/PDF. Backtest with the new prompt builder, and use revamped NLP screening with prebuilt screens like Momentum, Value, Growth & more. Nothing comes close!

Some good read for this weekend: "Virtue or Mirage? Complexity in Exchange Rate Prediction", where author test whether complex machine learning models beat simple ones in predicting FX rates. Turns out, complexity only helps a little and inconsistently — the humble random walk still wins most of the time papers.ssrn.com/sol3/papers.…

Kids are adorable these days. The scene below is perfect not only because of Joe Pesci’s incredible acting, but also because it was completely improvised — and it was maestro Joe himself who suggested it to Martin Scorsese!

Guess it’s November, not May 2026, when the world economy officially pauses

Calling a data viewer the best AI finance tool is like calling a calculator the next Jim Simons

The reason why Becks is my favourite player of all time is not only because of his tremendous skill and ability to play anywhere in midfield, but also because when he was in the US late in his career and got bored, he transferred to AC Milan and basically became one of the best midfielders in the league in no time, - literally like “yo why not?” - to the point that Milan begged him to stay.

This really happened #nvidia

Seems I have been correct after all

Most tools give you more data. DeepVest gives you more time 🕒. - Turn hours of market research into seconds - Get AI-powered breakdowns of assets, sectors & events - Spend your time on judgment, sizing & client decisions Stop grinding, start deciding.

This makes even perfectererer sense

This makes perfect sense given I have recently increased my ETF exposure to SPY.

Wouldn't you, realizing how much money this has been worth so far? Everyone would. It's not about dumping the whole position due to some bad inside info — it's about wanting to stop being rich on paper.

Thanks for having me on your show @TraderMerlin ! Here’s the YouTube channel of it piped.video/watch?v=WVNLE5…

Trend is not your friend

Good career advice: if you don’t click with the interviewers, you probably won’t enjoy the job.

When your sharpe's higher than your GPA

When I interview people, it’s all about: - Python basics: you obviously know python; next - Libraries: which ones do you use for ML modeling and why? Please say more than scikit learn - Math: all about asking OLS from within, then regularisation, then some math in trees, and if there’s time and will, math behind NN learning, which is an interesting chat to have. But generally, if you don’t know why multicollinearity causes XTX to be singular or what’s gradient descent, then that’s it - Evaluation: really about what to use where and why - Data preprocessing: huge topic, can be talked about for hours, so it’s rather about when we do, when we don’t need to do, and some quick examples. Fillna and trim outliers are boring - Data structures: you’re not a data engineer, and unless you want to create your own ML frameworks, I don’t care - ML algos: another huge topic, and again, best convo is about when to use which and why. You may like xgboost, cool, why? - DL: mentioned above, some maths to see if you even know wtf; and then which arch and why. But this one makes sense when projects use DL, otherwise I care far less - Projects: you have them on your CV, and we talked about it in the beginning

I asked @deepvest_ai for a breakdown. I asked "Why is Chipotle tanking over the last few months? Compare its performance against its competitors within their industry. Analyze its correlations against the overall market. After this run a statistical analysis against some top macro factors." DeepVest called a team of investment agents starting with our search agent to pull relevant context happening with the stock and then switch to calling our asset analysis agents to compute correlations against main competitors and top macro indicators along with pull down fundamentals and technicals on the stock and its peers. Overall, conclusion Chipotle's -40% drawdown reflects genuine fundamental deterioration, not just market noise. The combination of weakening traffic trends, guidance cuts, and bearish technical setup suggests more downside risk than upside potential in the near term. shared chat: deepvest.ai/shared/ea226341-…

I’ll never get why, when the weather is awful everywhere and the pilot says we’re doomed, the flight is smooth—but when it’s perfect outside, with zero wind and cold air, the plane bounces around like a quant’s career.

Me tweeting on topics unrelated to ML/quant

Recently discovered algorithms you’ve never heard of