01
Research—not investment advice
Salarium is an educational and quantitative research system. Nothing on this website is a recommendation, solicitation, personalized portfolio, suitability determination, or promise of future performance.
Research boundaries
Salarium is designed to make risk and uncertainty visible. These disclosures define what the system is, what its evidence represents, and what no visitor should infer from the interface.
Release evidence
Aug 11, 2026, 11:14 PM UTC
Regenerated from committed reports
Ranking artifact
Jun 10, 2026
25 names · committed · not live
Candidate artifact
Jul 10, 2026
20 monitored names · not live
Execution status
Research only
No brokerage connection or live order routing
Salarium is an educational and quantitative research system. Nothing on this website is a recommendation, solicitation, personalized portfolio, suitability determination, or promise of future performance.
All displayed returns, Sharpe ratios, Sortino ratios, drawdowns, hit rates, turnover, exposure, and related metrics are simulated historical research outputs. They are not live brokerage-account results.
The project evaluates multiple hypotheses. Even with walk-forward discipline, choosing an architecture after observing historical results creates model-selection and overfitting risk. Future market structure can differ materially from the research period.
The Liquid-500 and broad-universe pipelines rely on available price, liquidity, macro, and fundamental inputs. The project retains a documented survivorship-bias limitation, and point-in-time coverage is not complete for every feature or security.
Research deductions cannot fully represent spreads, market impact, borrow availability, financing terms, taxes, operational latency, or portfolio capacity. Real execution can be materially worse than simulated execution.
Ledoit-Wolf shrinkage improves stability but does not make covariance forecasts certain. Correlations and volatility can change abruptly, especially in market stress, causing realized risk to exceed forecasts.
The release architecture contains a hard 1.25x exposure ceiling. That ceiling is permission, not a target. The selected release mandate did not require leverage above 1.00x in the committed evaluation. Leverage can magnify losses and financing costs.
A high model rank or candidate score is not a trade instruction. Security selection, covariance, position caps, persistence buffers, evidence quality, and portfolio-level exposure rules intervene after ranking.
Salarium 1.0 does not connect to a broker, route orders, manage client assets, monitor personal accounts, or provide execution services. The public site reads frozen research and forward paper artifacts.
Users who run, modify, or deploy the source code are responsible for validating data licenses, software behavior, security, regulatory obligations, and financial risk in their own environment.
Artifact freshness
Dates below identify the precise artifacts displayed by the site. The forward paper file refreshes after eligible market closes; it is not an exchange-grade real-time feed or brokerage record.
Forward paper signal
Sep 29, 2026
499 scored · no orders
Release evidence
Aug 11, 2026, 11:14 PM UTC
7c730e84c409
Ranking signal date
Jun 10, 2026
25 committed rankings
Candidate as-of date
Jul 10, 2026
20 monitored candidates
Bottom line
Independent due diligence, professional advice where appropriate, and a clear understanding of loss tolerance are necessary before any real capital decision. A transparent model can still be wrong.
Continue with context