Trust & Transparency
Intelligence, Not Advice
KlirInvest is an educational analytics platform. We help you UNDERSTAND your portfolio — we do NOT tell you what to do.
What We Provide
- • Portfolio analytics & visualization
- • Risk scenario modeling
- • Market regime classification
- • Decision psychology insights
- • Educational tools & glossary
What We Do NOT Provide
- • Investment advice or recommendations
- • Trade execution or portfolio management
- • Price predictions or forecasts
- • Personalized financial planning
- • Tax or legal advice
Methodology
Numerical outputs are computed in code under disclosed model assumptions. Deterministic shocks and probabilistic path simulations answer different questions.
Portfolio Health Score
Model portfolio_health_score@2.1.0 (provisional-basis release). A 0–100 composite of six components: volatility (0.20), tail risk / CVaR (0.25), concentration (0.15), correlation stress (0.15), drawdown depth & proximity (0.15), and data confidence (0.10). Data confidence is coverage-scaled — when the covariance path can only measure a subset of holdings, the component is scaled by that coverage rather than silently reported at full weight.
Last internally reviewed: 2026-07-04
Decision Readiness
A 0–100 summary of recorded personal decision-process penalties: trading frequency against your own baseline (30%), risk-rule breaches (30%), drawdown against your stated tolerance (25%) and session-activity indicators (15%). Market regime is shown separately as context and carries no weight. Shown only after the calibration rules in Help are met.
Risk Metrics
1-day 95% Value at Risk (VaR) and Expected Shortfall via historical simulation on your tracked daily returns (last 365 days, at least 60 days; earlier, a labelled provisional estimate from your current holdings), plus portfolio volatility and maximum drawdown. Ten-day VaR uses a normal approximation, or a fat-tailed simulation in volatile or crisis regimes.
Stress Tests & Scenarios (deterministic)
Reviewed factor shocks are applied to portfolio weights and reconciled to holding and factor contributions in percent and dollars. Built-in, custom, and story scenarios use the same engine; story AI only structures editable assumptions. Rates and credit require separate duration inputs, missing coverage is explicit, and the result assigns no probability, confidence interval, correlation multiplier, or calibrated uncertainty band. When a scenario states a market move, holdings no category describes take beta × that move (market fallback): equity funds and unshocked single stocks, and any other non-cash, non-bond holding with a measured beta — for example gold, silver or other commodity funds when no metal shock is stated, and volatility products — with no minimum fit. When the scenario states only a broad-equity move, only a well-fitting measured beta is used (R² at least 0.20 over at least 120 daily returns). Where a measured and a reference beta both exist, the beta used is a precision-weighted (Vasicek) blend; each fallback line shows the measured beta, its R² and the weight on each. This linear mapping understates crash spikes and ignores futures-roll decay. US-dollar funds take the dollar move a scenario states and are otherwise shown as not modelled.
Last internally reviewed: 2026-07-04
Monte Carlo Simulations (probabilistic)
Path simulation from your holdings' actual return history: the covariance matrix Σ is estimated with Ledoit-Wolf shrinkage on a 400-calendar-day window (about 274 trading days). The measured covariance is used as is: no market-regime adjustment is applied, because regime labels are descriptive, not calibrated volatility forecasts. Tail Risk Mode uses a normal location-scale mixture (df = 5): both volatility and log-return drift vary across paths. Expected conditional diffusion variance is matched to Normal; terminal log returns are not standardized Student-t and their variance need not match Normal. Paths use antithetic sampling for variance reduction. Default run: 10,000 paths. When per-holding history is insufficient, coverage is disclosed and an asset-bucket fallback is used for the affected holdings — never silently.
Last internally reviewed: 2026-07-04
Portfolio Optimizer
Solver-based Min-Variance / Risk-Parity on the measured covariance matrix, with hard position and sector caps. When a sector cap is binding at the optimum, we surface it on the result ("binding: {Sector}").
Last internally reviewed: 2026-06-30
Behavioral Analytics
Confidence-tiered constructs: fewer than 10 recorded actions → no numeric score (patterns only); 10–29 → directional bands (e.g. "80–95"); 30+ → numeric scores. We never fake precision at low sample sizes.
Last internally reviewed: 2026-06-30
Tax-Loss Harvesting
Losses are bucketed into Short-term, Long-term, and Term unverified. Positions with estimated (rather than real) purchase dates are never folded into Short/Long — they surface with a fix note explaining how to promote them.
Last internally reviewed: 2026-06-30
Methods and sources
The literature below supports the methods our engines implement. Parameter values are internal calibration unless a source is named against them in the model-parameters section. None of this constitutes third-party validation of KlirInvest.
- Covariance shrinkage — Ledoit & Wolf (2004), “Honey, I Shrunk the Sample Covariance Matrix”, Journal of Portfolio Management.
- Tail correlation / stress uplift — Longin & Solnik (2001), “Extreme Correlation of International Equity Markets”, Journal of Finance; Ang & Chen (2002), “Asymmetric Correlations of Equity Portfolios”, Journal of Financial Economics.
- Equicorrelation stress convention — Engle (2002), Dynamic Conditional Correlation; Roncalli, Handbook of Financial Risk Management, §11.3.
- Modified (Cornish-Fisher) VaR — Cornish & Fisher (1937); Favre & Galeano (2002), “Mean-Modified Value-at-Risk Optimization”.
- Fat-tail degrees of freedom — Platen & Rendek (2008) on empirical equity-return tail indices; standard RiskMetrics-t practice.
- Return conventions — GIPS treatment of Modified Dietz and chain-linked time-weighted return; CFA Institute Sharpe-ratio convention.
- Behavioural constructs — Barber & Odean (2000) on attention-driven trading and net returns.
Model Validation
We compare model forecasts with later outcomes as data accumulates. These checks are early and provisional: they are not a completed or independent validation, and statistical tests run only once enough observations exist.
Platform summary
Early and provisional — not completed or independent validation.
The platform summary above is public. Sign in to see the detailed checks for your own account.
Model Parameters
The tunable constants the current risk and simulation engines use — value, units, rationale, source, sensitivity, and last review date. Grouped by engine. Values that no current engine reads are listed separately as retired.
Behavioral Validation
Behavioral Construct Validation
Validation status: constructs are internally consistent and confidence-tiered by sample size; convergent validation planned at n≥30 per user; outcome-linkage study at cohort scale.
What moves each score
- Discipline (disposition effect):
- Rises when you realise losses about as readily as gains; falls when gains are sold while losses are held (Odean 1998 proportion of gains realised minus proportion of losses realised)
- Measured from sales replayed against your transaction ledger in the observation window; positions you have not sold do not move it
- Consistency (turnover):
- Rises as annualised turnover (traded value relative to portfolio value) falls: 0% scores 100, 258% a year or more scores 0
- Round-trips bought and sold within 30 days are listed for review; simulations do not count as trading
- Diversification (concentration):
- Rises as weight is spread across more holdings: score = (1 − HHI) × 100 on current holding weights, cash excluded
- Uses weights only, not correlations; 1/HHI is the equally weighted equivalent number of positions, not a count of independent bets
- Process signals:
- Read only from trades with a written reason and stated decisions; with none logged there is no score
- Falls when momentum-language entries, exits after a market drop or loss, or same-ticker re-entry within 24 hours of a loss sale are detected in that logged activity
Data Sources
Yahoo Finance data comes from its public chart endpoint, not a licensed data feed. Providers that need an account are used only while that account is configured; when a provider fails the next one in the list is tried, and a missing value is shown as unavailable rather than filled in.
Data is cached and refreshed according to tier limits. Intraday data may have 15-minute delays on the Explore plan.
Known Limitations
Understanding the boundaries of our analytics
Security Posture
AI Transparency
KlirInvest uses AI to help explain complex concepts and provide context:
- AI explains calculations — it NEVER computes numbers
- Numerical outputs come from calculation code; AI commentary is separate
- AI responses are compliance-scored and audited
- Prohibited: advice, predictions, urgency, recommendations
AI Role: Analyst (explains) — NOT Advisor (recommends)
Regulatory Compliance
KlirInvest is not authorised or regulated by the Financial Conduct Authority, and is not an investment adviser, broker, portfolio manager or financial planner in any jurisdiction. We provide educational analytics tools only.
All users acknowledge this via mandatory disclosure during onboarding and before using decision-support features. Users are responsible for their own investment decisions and should consult qualified financial professionals for personalised advice.
