Indexnews applies predictive modelling to large volumes of market data, producing strategy recommendations that have been backtested against historical performance before being presented to you.
Interface preview: a consolidated dashboard surfacing model confidence, risk exposure, and backtested return ranges across active strategies, updated as new market data is ingested.
Remote-based investors and consultants increasingly operate across multiple markets, time zones, and information feeds. The volume of available data has grown faster than the capacity to interpret it, and manual analysis introduces delay at precisely the moments when timing matters most.
Indexnews addresses this by compressing the distance between raw market signals and a usable recommendation. Predictive models process incoming data continuously, flag asymmetric risk, and surface a narrower set of decisions worth reviewing, rather than a wider set of data points to interpret manually.
A rolling multi-year backtest comparing projected strategy output against realised market performance, segmented by market cycle rather than by a single favourable period.
The platform is built around a single operating sequence: data is ingested, analysed against historical and live conditions, and translated into a specific recommendation with a stated confidence level.
Structured and unstructured market data is pulled continuously from multiple sources, normalised, and checked for completeness before entering the analysis layer. Gaps or anomalies are flagged rather than silently filled.
Predictive models assess current conditions against historical patterns, weighting outcomes by volatility and correlation. Each model is versioned, so a recommendation can always be traced back to the conditions that produced it.
Output is delivered as a ranked, risk-adjusted action with a stated rationale and a backtested performance range, allowing a decision-maker to assess the trade-off before committing capital or time.
Every model undergoes technical validation against out-of-sample data before it is deployed into live recommendations, reducing the likelihood that a strategy performs well only in the conditions it was built on.
Backtested across multiple historical market cyclesExecutive trust depends on being able to see how a conclusion was reached, not only what the conclusion is. The workflow below is the same sequence applied to every recommendation Indexnews produces.
Market feeds, pricing data, and macro indicators are standardised into a common format, with source and timestamp retained for audit purposes.
Multiple models independently score the same dataset. Where models disagree beyond a defined threshold, the recommendation is held for manual review rather than issued automatically.
Recommendations are adjusted against stated risk parameters, including maximum drawdown tolerance and exposure limits, before being presented to the user.
Once delivered, each recommendation is tracked against its own backtested range, and material deviation triggers a review of the underlying model.
Position sizing, exposure limits, and drawdown thresholds are configured before deployment and enforced automatically. No recommendation bypasses these parameters, regardless of projected upside.
Account and portfolio data are encrypted in transit and at rest. Access is restricted on a need-to-know basis, and administrative actions are logged for review.
Indexnews is designed on the assumption that every recommendation will eventually be questioned, whether by an internal risk committee, a client, or the user's own later review. The platform is structured so that assumptions, data sources, and model versions remain visible rather than abstracted away behind a single output.
This approach is deliberately conservative. Where a model's confidence falls below an acceptable threshold, the system withholds a recommendation rather than issuing one with inflated certainty.
The same underlying models are configured differently depending on whether the user is managing personal capital or advising clients from a location-independent base.
A location-independent investor uses Indexnews to monitor multiple positions without maintaining a full-time research desk. Recommendations are reviewed on a set schedule rather than continuously, reducing the operational burden of staying current across time zones.
Scheduled, not constant monitoring
Reduced research overhead
Fixed exposure and drawdown limits
Consultants advising clients on capital allocation use the platform's audit trail to support recommendations with a traceable methodology, rather than relying solely on judgement calls that are difficult to defend after the fact.
Per-engagement analysis
Defensible, documented rationale
Client-specific risk parameters
Managers overseeing several portfolios from a single remote base rely on the ingestion layer to consolidate disparate data sources, avoiding the need to reconcile multiple separate feeds manually before each decision.
Daily consolidated review
Single consolidated data view
Cross-portfolio exposure limits
Each model's live performance is tracked against its own backtested range. When actual performance diverges from the expected range beyond a set tolerance, the model is flagged for review and, if necessary, retrained on more recent data before continuing to issue recommendations.
Latency depends on the data source and the complexity of the model involved. Ingestion and scoring are designed to run on a near-continuous basis, though some recommendations are intentionally held for manual review rather than issued instantly, particularly where model disagreement is high.
The platform is browser-based and does not require local installation, which allows access from any location with a stable connection. Integration with existing reporting or portfolio tools is handled on a case-by-case basis during onboarding.
Backtesting periods vary by strategy and are set to cover more than one market cycle where sufficient historical data exists, so that performance is not assessed solely against a single favourable condition.
The system withholds a recommendation rather than issuing one with artificially inflated certainty. A low-confidence state is shown explicitly rather than suppressed.
Further technical documentation and direct support are available to approved accounts following onboarding.
Request access to view live model output, backtested performance ranges, and the methodology behind current recommendations.
Request AccessNo payment details are required to request access. An account specialist will confirm eligibility before onboarding begins.