Blog
Trading, markets, and our tools
Research notes, engineering write-ups and plain-language explainers from the team building AlgoBeam. We publish the method alongside the result, including the parts that did not work.
- essays published
- 9 essays published
- topics covered
- 4 topics covered
- last updated
- Aug 21, 2026 last updated
The two we would read first
One on why research quietly goes wrong, one on the architecture that keeps our assistant tied to the bars it reasoned from.
Everything else we have published
Filter by topic, or search titles, excerpts and authors. Nothing here sits behind a paywall.
Showing 7 of 7 posts
- ResearchJul 29, 2026
The 2026 volatility regime, in four measurements
Realised volatility is unremarkable, but its clustering is not. What a year of compressed ranges and violent single-day gaps means for systematic traders.
Mara Vestergaard
8 min read
- EngineeringJul 10, 2026
AlgoBeamTS 1.0: strategy script, anywhere JavaScript runs
Our open-source transpiler hits 1.0. The same AlgoBeamScript strategy now produces identical output in the browser, in Node and on a hosted worker.
Dev Ranganathan
7 min read
- EducationJun 24, 2026
Position sizing is the only edge you fully control
You cannot make the market cooperate and you cannot force a win rate. You can decide, in advance and to the cent, how much a bad idea is allowed to cost.
Amara Osei
8 min read
- ProductJun 5, 2026
Inside the new screener: 180 filters, one query
We replaced a wall of nested dropdown menus with a small typed query language. Here is why, and what it lets you do that menus never could.
Lena Kowalczyk
6 min read
- EducationMay 14, 2026
How to read a market heatmap without fooling yourself
A treemap is a compression of thousands of numbers into one image. Knowing what it throws away is the difference between insight and decoration.
Sofia Marchetti
6 min read
- ResearchApr 28, 2026
Walk-forward analysis, step by step
A single in-sample and out-of-sample split tells you almost nothing. Rolling the split repeatedly tells you whether your parameters have a shelf life.
Tomas Ekwueme
9 min read
- ProductApr 2, 2026
Alerts that actually fire: rebuilding the notification pipeline
Median delivery went from 5.8 seconds after bar close to under one. The interesting part is everything we had to stop doing to get there.
Lena Kowalczyk
7 min read
Editorial standards, stated up front
AlgoBeam is a research tool, so the writing has to hold to the same standard as the product.
Written by the people who built it
Every post is signed by a researcher, engineer or product lead on the team. No ghostwriters, no syndicated filler.
Method before conclusion
We show the measurement, the window and the assumptions first. If a number looks flattering, we say which choice made it look that way.
Nothing here is a signal
We publish research and engineering notes, never recommendations.
The AlgoBeam research digest
One email with the new essays, the changelog items worth knowing about, and a short note on something the market did that we could not explain. No promotions, no trade ideas, one click to leave.
Ready when you are
Read less, test more.
Every idea on this blog is one you can check yourself. Open a chart, fork the strategy and run the backtest on your own data before you believe a word of it.
- No card required
- Cancel anytime
- 14-day Ultimate trial