Prudence
We prefer sober readings over attention-grabbing conclusions.
How the system works
Methodology
How the readings are built, what data they use and where their limits are.
Public sourcesTraceabilityVisible limits
DRL research
DRL Evaluation Protocol
Can a disciplined evaluation protocol distinguish statistically credible and practically feasible DRL portfolio performance from apparent backtest strength?
TD3 is used as a DRL case study, not as a claim that the algorithm is superior.
Principles
Traceability
Each module declares source, state, frequency and main limit.
Privacy
No personal answers, portfolios or wealth data are stored.
No prediction
The system organizes context; it does not forecast prices.
Education
Tools help users think better, not delegate judgment.
Separation
Market context is not the same as personalized advice.
System map
From public data to human decision
Public data / sources
Market data, volatility, flows, quantitative outputs and educational criteria.
Processing
Server-side adapters, static calculations, validations and prudent fallbacks.
Statistical readings
Regime, rotation, levels, drawdowns, metrics and relative signals.
Educational tools
Diagnostic, protection, trends, methodology and reading labs.
Human decision
The person contrasts, asks, documents and decides outside the platform.
Tool by tool
What each tool does, does not do and where it ends
| Tool | What it does | What it does not do | Source / criterion | Main limit |
|---|---|---|---|---|
| Market Regime Dashboard | Organizes volatility, sector rotation and BTC ETF flows. | Does not provide execution signals. | Alpha Vantage, FRED/VIX, Bitbo and pending states where applicable. | External sources may be delayed, fail or change format. |
| Investor Diagnostic | Helps organize horizon, tolerance, behavior and experience. | Does not replace a formal regulatory suitability assessment. | Internal educational questionnaire. | Does not store answers or build a legal user profile. |
| Statistical Levels | Calculates levels, extremes, drawdowns and historical patterns by asset. | Does not predict breakouts or future support levels. | Static historical dataset generated locally. | Quality depends on available history for each asset. |
| DRL Evaluation Protocol | Presents the protocol, assumptions, costs, cash, comparable benchmarks, bootstrap, White Reality Check and practical limits. | Does not prove that TD3 always wins, does not claim broad statistical superiority and does not recommend allocations. | Interactive research paper. TD3 is used as a DRL case study, not as a superiority claim for the algorithm. | Historical and experimental results do not guarantee future behavior. |
| Trends | Turns technological, economic and social changes into educational observation hypotheses. | Does not recommend assets, validate that a trend is a good investment or replace analysis of price, vehicle, horizon and risk. | Editorial framework using sector sources, macro data, institutional reports, Google Trends as an attention signal and holdings/13F as delayed observation with limitations. | A real trend can still be a poor investment if it is expensive, lacks a clear vehicle, is already priced in or does not fit the portfolio. |
| Protect Your Money | Organizes educational red flags before trusting a proposal. | Does not officially verify entities or declare fraud. | Public CNMV / MiFID II criteria. | Should be complemented with official source checks. |
Sources and updates
Data needs visible state
The platform distinguishes automated data, manual data, educational demo data and pending sources. Source state matters as much as the number.
Market
Alpha Vantage for sector ETFs, FRED/VIX, Bitbo, State Street / GLD and pending data where applicable.
Statistical levels
The current dataset is a static snapshot generated by script and committed; the repo does not show evidence of a deploy cron that regenerates it automatically.
Updates
Some modules update server-side with cache; others require explicit regeneration through static scripts.
Fallbacks
If a source fails, the module should show a prudent state instead of presenting demo data as real.
Privacy
The reading happens without storing your financial life
No accountsNo current cookiesNo real analyticsNo personal answers storedNo portfolios storedNo browser persistence
Some selections may live temporarily in the interface or URL so a view can be reproduced, but personal answers and wealth profiles are not stored.
Limits
What the platform does not promise
- This platform does not provide financial, legal, tax or wealth advice.
- It does not recommend assets, products, weights or execution timing.
- It does not predict future prices, returns, volatility or flows.
- It does not replace a formal regulatory suitability or appropriateness assessment.
- Data may fail, be delayed, revised or temporarily unavailable.
- Quantitative models show historical results, not future guarantees.