About SIGMA IV
An independent research and technology company working across scientific software, quantitative finance and machine learning.
How it started
SIGMA IV began in quantitative finance and software development. Over nearly five years that work has covered allocation and rotation models, backtesting, risk management and the data systems behind them. Along the way, the founder worked as an analyst for BlockPhi as part of a partnership, and then built an independent data analytics platform for digital-asset markets.
The founder's background is in biomedical research. The statistical problems in a laboratory are close to those in markets: small samples, noisy measurements and many comparisons. That overlap led to SCALAR, analysis software that lets researchers check every step behind their results.
Machine learning connects both fields. It is a feature in SCALAR, a tool in quantitative research, and the subject of ongoing research into scientific image analysis.
What it works on
- Laboratory analysis is spread over spreadsheets, scripts and chart tools, and is hard to check.SCALAR keeps the whole route in one program and shows the calculation behind every answer.
- Backtests often look better than reality because of hindsight, missing costs or luck.Quantitative research that tests for exactly those effects before a result counts.
- Models are often trusted without being compared with chance.Machine learning with held-out validation and baselines, in SCALAR and in research.
Principles
- Show the calculation
- A number comes with the steps that produced it.
- Label the evidence
- Working software, research and plans are kept apart.
- Move methods between fields
- A technique proven in one discipline is tested in another.
- Keep data local
- Where possible, analysis runs on the user's own computer.