NEXUSTRADE / VIDEOS
AI trading.
See it in action.
Real builds. Honest experiments. A closer look at AI trading, one video at a time.
Latest shorts
0:32Transcript
algotrading
20 steps to build a trading strategy that actually works
A 20-step loop for building a trading strategy, from Austin Starks. Start with a simple idea and backtest it, add layers and backtest again, then find a promising candidate and inspect its backtest events rather than trusting the summary numbers. Reason through the trace, create new strategy variants, re-run the backtests, and pick the best version. The second half is the discipline that decides whether any of it pays. Run walk-forward optimization, reason through the rules the optimizer produced, and reject the strategy outright if those rules cannot be interpreted. Test out of sample, throw away the failures and start from scratch, and paper-trade the winners for three months while keeping a journal of the idea. Repeat the whole loop across ten strategies and deploy only the most consistent one.
1:38Transcript
market cap
Buying the Five Largest Companies: A Market-Cap Strategy Backtest
Buy the five largest US companies by market cap, 20% in each. Every two weeks, re-sort the list and reset all five back to 20%: trim whatever ran up, add to whatever drifted down, and swap out anything that fell out of the top five. One rule, and nothing to tune. In the January 2020 to August 2026 backtest, the strategy turned $100,000 into $415,000, against $246,000 for the S&P 500. Dividends are counted on both sides and trading costs are charged at a tenth of a percent per order, which is a fair approximation for stocks this liquid. Taxes are not modelled at all, and resetting five positions every two weeks in a taxable account generates short-term gains. The costs are real too. The worst drawdown was 44.9%, about eleven points deeper than the index, and you are holding five companies instead of five hundred. Open the linked portfolio to read the rules and run the backtest yourself. Performance figures reflect the recording, not current returns.
1:04Transcript
ai avatar
How I Built My AI Avatar: Voice Cloning, HeyGen and Remotion
How do you turn a still photo and a cloned voice into a talking-head video? Austin Starks walks through the seven-step workflow behind his AI avatar, created to share AI trading content. The process starts with OpenRouter, HeyGen, an agentic coding tool, RunPod and Hugging Face. Generate a portrait, clone your voice with Dots TTS, turn a script into narration, and give the image and audio to HeyGen. Finally, use Remotion and footage from your camera roll to build the b-roll. Watch the walkthrough, then read the linked voice-cloning guide for the part of the process that turns your own voice into reusable narration.
1:20Transcript
software design
How to Design Software Systems With AI Before Writing Code
A working demo is only the beginning. Austin Starks explains the workflow he uses to design and build software systems with AI: write the design first, review it adversarially, and only then implement the code. The design should examine existing patterns, cover the full stack, consider how the system scales, and include concrete code examples. A separate design review catches problems before implementation. Once the design is ready, a fast coding model can build it, followed by an adversarial code review. The video closes with the checks that support shipping: unit tests, integration tests, end-to-end tests and CI/CD. Explore the linked course to learn more about building AI agents from scratch.
0:36Transcript
algotrading
20 things to do before you deploy an AI trading bot
Before deploying an AI trading bot, work through this 20-item research checklist from Austin Starks. It covers technical, fundamental, and economic indicators; alternative data; look-ahead and selection bias; overfitting; and walk-forward analysis. The checklist also includes probability, uncorrelated markets and strategies, the ReAct paper, tool calling, MCP servers, backtesting, and paper trading. This short video names the topics to investigate before going live; it is a starting point for further study.
0:47Transcript
ai trading
I Put $25,000 Into an AI Trading Strategy
Austin Starks put $25,000 of his own money into an AI trading strategy and made the strategies public so anyone could inspect his work. In this Public Portfolio Challenge update, he explains why he questioned the backtest results, audited the data, and kept looking for bugs before and after going live. Watch the story, then open the linked portfolio to follow the challenge. Returns mentioned in the video describe the account at the time of recording.