RiskModels Concepts — Public Glossary

A reading guide to the vocabulary you'll see in our docs, articles, and SDK, ordered the way the platform works: bottom-up. Start with the risk of a single stock, aggregate positions into a portfolio, then decompose the portfolio's performance and risk. Point-in-time discipline, the regulatory data underneath, and the trust machinery come after.

This page is for analysts, researchers, students, and prospective users. It covers the language, not the implementation.

1. Stock-level risk (bottom-up)

The unit everything aggregates from: decomposing a single equity into factor and residual layers.

Term	Definition
Factor model	A linear model decomposing an asset's return into common factors (market, sector, …) plus a residual. The foundation of the bottom-up unit.
Beta (β)	The slope of an asset's return regressed on a factor; how much the asset moves per unit factor move.
Residual	The part of a stock's return left after the factor model is removed. RiskModels emphasizes the residual as where selection skill lives.
Idiosyncratic	Synonym for residual in classical finance literature. We standardize on residual.
Alpha (α)	Skill-attributable excess return at the position level.
Hedge ratio	The factor exposure you'd offset to neutralize a position's exposure to that factor.
Variance decomposition	Splitting a stock's variance into shares attributable to each factor and the residual. Sums to 1.0.
Return attribution	Splitting a stock's realized return into contributions from each factor. Distinct from variance: a name can be variance-dominated by market yet return-driven by residual.

2. From stocks to portfolios (aggregation)

How position-level risk rolls up. A portfolio is its holdings, weighted — so its exposures are the weighted aggregate of the bottom-up decompositions, plus the covariance between positions.

Term	Definition
Position weight	A holding's share of portfolio value; the weight that scales its risk and return contribution when positions aggregate.
Holdings look-through	Rolling individual holdings up to portfolio-level factor exposure by weight — the bridge from stock risk to portfolio risk.
Peer cohort	A cap-weighted set of similar securities (typically by sector and sub-sector) used as the comparison universe for relative analyses.
Covariance	How two positions move together; why portfolio risk is not just the weighted sum of position risks.
Diversification	The risk reduction from holding imperfectly-correlated positions; the gap between summed position risk and actual portfolio risk.

3. Portfolio performance & risk decomposition

The top of the arc: once positions are aggregated, decompose the portfolio's return and risk the same way — into factor layers, residual, and per-position contributions.

Term	Definition
Portfolio variance decomposition	Splitting total portfolio variance into factor and residual shares, accounting for cross-position covariance. The portfolio-level analog of the single-stock decomposition.
Risk contribution	Each position's (or factor's) share of total portfolio risk, accounting for correlation. Component contributions sum to the total — unlike standalone position risk.
Portfolio attribution	Splitting the portfolio's realized return into contributions from market, sector, subsector, and residual selection.
Active share	The fraction of a portfolio that differs from its benchmark; how much of the book is an active bet versus the index.
Tracking error	The volatility of portfolio-minus-benchmark return; the size of the active risk.
Benchmark fit	How well a portfolio's returns are explained by a benchmark (R², tracking error) — the quality of the comparison.
Sharpe ratio	Return per unit of volatility.
Drawdown	Peak-to-trough decline. Max drawdown is the worst observed.

4. Point-in-time data discipline

How institutional analysis avoids look-ahead bias. Universal vocabulary in regulated industries; the discipline separates serious quant work from naïve backtesting.

Term	Definition
Point-in-time (PIT)	A dataset that respects when each fact was knowable, not just when it was true. Bloomberg and FactSet sell PIT databases as a premium product.
Look-ahead bias	Letting a backtest see information that wasn't yet public on the simulated date. The most common research bug.
Filing lag	The gap between when a fund's holdings reflected reality and when they became publicly knowable via SEC filings. ~45 days for 13F, ~60 days for N-PORT.
Survivorship bias	Backtesting on a universe that excludes delisted or failed entities. Produces optimistic results.

5. Regulatory data

The filings the platform consumes (see also the Filings section for the calendar). Universal vocabulary for institutional analysis of US-listed funds and managers.

Term	Definition
13F	SEC filing (Form 13F-HR) listing institutional managers' US-listed equity holdings. Quarterly, 45-day lag.
N-PORT	SEC filing reporting registered fund (mutual fund / ETF) holdings monthly. Only every third month is publicly disclosed.
N-CEN	Annual SEC filing for registered funds; carries adviser relationships and share-class detail.
CIK	The SEC's identifier for any filer (manager, fund, adviser).
Security master	The canonical record of what is this security and what was it called when? — the foundation under any survivorship-corrected analysis.

6. What RiskModels publishes

The artifacts and surfaces we make available externally.

Term	Definition
RiskModels API	The public REST + MCP surface for decomposition, attribution, and exposure analyses. Documented at riskmodels.app.
`riskmodels-py` (SDK)	The Python client for the API. `pip install riskmodels-py`.
Position snapshot	A one-page institutional analysis of a single equity: factor decomposition, attribution, residual quality, peer context.
Fund snapshot	A one-page tearsheet for a single fund: portfolio attribution, holdings, NAV, benchmark fit.
AOM	Analysis Object Model — our publicly-versioned protocol for describing analyses (subject + lens + view) before they hit any endpoint.

7. How we keep results trustworthy

Engineering vocabulary that lets institutional buyers evaluate whether outputs are reproducible.

Term	Definition
Deterministic	Same input → byte-identical output on re-run. Required for cache correctness and auditability.
Provenance	The metadata recording how an artifact was produced: subject, parameters, as-of date, model version.
Render-once	Web, PDF, PNG, and agent surfaces all read the same canonical artifact. Drift across surfaces is impossible by construction.

8. The RiskModels engine (ERM3)

The hierarchical orthogonal decomposition that produces everything above. The math is standard; the cascade is ours. Full derivation in the .

Term	Definition
ERM3	Equity Risk Model v3 — the engine. Decomposes any US equity into market → sector → subsector → residual via a hierarchical orthogonal cascade.
Hierarchical cascade (L1/L2/L3)	L1 = market (SPY); L2 = sector ETF (e.g. XLK) fit on the L1 residual; L3 = subsector ETF (e.g. SOXX) fit on the L2 residual. Each level explains what the prior one left over.
Link beta (λ)	The coefficient that strips a higher level's exposure from a lower one, so each β captures only the incremental effect of its own level — no double-counting.
Explained risk (ER)	The variance share attributable to each level. ER(L1) + ER(L2) + ER(L3) + residual = 1.0 by construction, verified at runtime to ±0.1%.
Replication equation	The raw-ETF identity that reproduces a stock's return from its hedge legs — only tradeable ETF returns, no orthogonalization at trade time. Verified to ±0.1% for every stock, every date.
Robust beta (Huber-M)	Betas estimated by 252-day rolling Huber-M regression, robust to the fat tails (earnings gaps, single-day reclassifications) that distort ordinary least squares.
Vasicek shrinkage	L2/L3 betas shrunk toward the log-mcap-weighted peer mean within a 4-digit industry cohort; short-history and small-caps are pulled more. Market beta (L1) is not shrunk.
L-star rule	The L* rule: a cost-aware, learned selector that picks the shallowest level whose expected out-of-sample variance reduction outweighs the added hedging cost of going deeper — describing a name at the coarsest level that still captures its risk. Scored point-in-time, so the dispatch is backtest-safe.
Geometric attribution bridge	Telescoping compound differences (market → sector → subsector → residual) so multi-period factor contributions sum exactly to the gross compound return — not the naive daily sum, which overstates by volatility drag.
Residual mean-reversion signal	A 5-day cumulative L3 residual return, z-scored by the stock's own 60-day residual volatility. A combo-input factor for multi-signal stacks, not a standalone strategy.

Going deeper

Decomposition philosophy — the Hierarchical Orthogonal Decomposition article series: why residual matters more than headline beta.
Same label, different bet — why two analysts who both "own NVDA" may be running entirely different theses.

For analysts using the SDK, the concepts above are enough. For everything else, the API documentation is authoritative.

RiskModels Concepts — Public Glossary

This page is for analysts, researchers, students, and prospective users. It covers the language, not the implementation.

1. Stock-level risk (bottom-up)

The unit everything aggregates from: decomposing a single equity into factor and residual layers.

Term	Definition
Factor model	A linear model decomposing an asset's return into common factors (market, sector, …) plus a residual. The foundation of the bottom-up unit.
Beta (β)	The slope of an asset's return regressed on a factor; how much the asset moves per unit factor move.
Residual	The part of a stock's return left after the factor model is removed. RiskModels emphasizes the residual as where selection skill lives.
Idiosyncratic	Synonym for residual in classical finance literature. We standardize on residual.
Alpha (α)	Skill-attributable excess return at the position level.
Hedge ratio	The factor exposure you'd offset to neutralize a position's exposure to that factor.
Variance decomposition	Splitting a stock's variance into shares attributable to each factor and the residual. Sums to 1.0.
Return attribution	Splitting a stock's realized return into contributions from each factor. Distinct from variance: a name can be variance-dominated by market yet return-driven by residual.

2. From stocks to portfolios (aggregation)

How position-level risk rolls up. A portfolio is its holdings, weighted — so its exposures are the weighted aggregate of the bottom-up decompositions, plus the covariance between positions.

Term	Definition
Position weight	A holding's share of portfolio value; the weight that scales its risk and return contribution when positions aggregate.
Holdings look-through	Rolling individual holdings up to portfolio-level factor exposure by weight — the bridge from stock risk to portfolio risk.
Peer cohort	A cap-weighted set of similar securities (typically by sector and sub-sector) used as the comparison universe for relative analyses.
Covariance	How two positions move together; why portfolio risk is not just the weighted sum of position risks.
Diversification	The risk reduction from holding imperfectly-correlated positions; the gap between summed position risk and actual portfolio risk.

3. Portfolio performance & risk decomposition

The top of the arc: once positions are aggregated, decompose the portfolio's return and risk the same way — into factor layers, residual, and per-position contributions.

Term	Definition
Portfolio variance decomposition	Splitting total portfolio variance into factor and residual shares, accounting for cross-position covariance. The portfolio-level analog of the single-stock decomposition.
Risk contribution	Each position's (or factor's) share of total portfolio risk, accounting for correlation. Component contributions sum to the total — unlike standalone position risk.
Portfolio attribution	Splitting the portfolio's realized return into contributions from market, sector, subsector, and residual selection.
Active share	The fraction of a portfolio that differs from its benchmark; how much of the book is an active bet versus the index.
Tracking error	The volatility of portfolio-minus-benchmark return; the size of the active risk.
Benchmark fit	How well a portfolio's returns are explained by a benchmark (R², tracking error) — the quality of the comparison.
Sharpe ratio	Return per unit of volatility.
Drawdown	Peak-to-trough decline. Max drawdown is the worst observed.

4. Point-in-time data discipline

How institutional analysis avoids look-ahead bias. Universal vocabulary in regulated industries; the discipline separates serious quant work from naïve backtesting.

Term	Definition
Point-in-time (PIT)	A dataset that respects when each fact was knowable, not just when it was true. Bloomberg and FactSet sell PIT databases as a premium product.
Look-ahead bias	Letting a backtest see information that wasn't yet public on the simulated date. The most common research bug.
Filing lag	The gap between when a fund's holdings reflected reality and when they became publicly knowable via SEC filings. ~45 days for 13F, ~60 days for N-PORT.
Survivorship bias	Backtesting on a universe that excludes delisted or failed entities. Produces optimistic results.

5. Regulatory data

The filings the platform consumes (see also the Filings section for the calendar). Universal vocabulary for institutional analysis of US-listed funds and managers.

Term	Definition
13F	SEC filing (Form 13F-HR) listing institutional managers' US-listed equity holdings. Quarterly, 45-day lag.
N-PORT	SEC filing reporting registered fund (mutual fund / ETF) holdings monthly. Only every third month is publicly disclosed.
N-CEN	Annual SEC filing for registered funds; carries adviser relationships and share-class detail.
CIK	The SEC's identifier for any filer (manager, fund, adviser).
Security master	The canonical record of what is this security and what was it called when? — the foundation under any survivorship-corrected analysis.

6. What RiskModels publishes

The artifacts and surfaces we make available externally.

Term	Definition
RiskModels API	The public REST + MCP surface for decomposition, attribution, and exposure analyses. Documented at riskmodels.app.
`riskmodels-py` (SDK)	The Python client for the API. `pip install riskmodels-py`.
Position snapshot	A one-page institutional analysis of a single equity: factor decomposition, attribution, residual quality, peer context.
Fund snapshot	A one-page tearsheet for a single fund: portfolio attribution, holdings, NAV, benchmark fit.
AOM	Analysis Object Model — our publicly-versioned protocol for describing analyses (subject + lens + view) before they hit any endpoint.

7. How we keep results trustworthy

Engineering vocabulary that lets institutional buyers evaluate whether outputs are reproducible.

Term	Definition
Deterministic	Same input → byte-identical output on re-run. Required for cache correctness and auditability.
Provenance	The metadata recording how an artifact was produced: subject, parameters, as-of date, model version.
Render-once	Web, PDF, PNG, and agent surfaces all read the same canonical artifact. Drift across surfaces is impossible by construction.

8. The RiskModels engine (ERM3)

The hierarchical orthogonal decomposition that produces everything above. The math is standard; the cascade is ours. Full derivation in the .

Term	Definition
ERM3	Equity Risk Model v3 — the engine. Decomposes any US equity into market → sector → subsector → residual via a hierarchical orthogonal cascade.
Hierarchical cascade (L1/L2/L3)	L1 = market (SPY); L2 = sector ETF (e.g. XLK) fit on the L1 residual; L3 = subsector ETF (e.g. SOXX) fit on the L2 residual. Each level explains what the prior one left over.
Link beta (λ)	The coefficient that strips a higher level's exposure from a lower one, so each β captures only the incremental effect of its own level — no double-counting.
Explained risk (ER)	The variance share attributable to each level. ER(L1) + ER(L2) + ER(L3) + residual = 1.0 by construction, verified at runtime to ±0.1%.
Replication equation	The raw-ETF identity that reproduces a stock's return from its hedge legs — only tradeable ETF returns, no orthogonalization at trade time. Verified to ±0.1% for every stock, every date.
Robust beta (Huber-M)	Betas estimated by 252-day rolling Huber-M regression, robust to the fat tails (earnings gaps, single-day reclassifications) that distort ordinary least squares.
Vasicek shrinkage	L2/L3 betas shrunk toward the log-mcap-weighted peer mean within a 4-digit industry cohort; short-history and small-caps are pulled more. Market beta (L1) is not shrunk.
L-star rule	The L* rule: a cost-aware, learned selector that picks the shallowest level whose expected out-of-sample variance reduction outweighs the added hedging cost of going deeper — describing a name at the coarsest level that still captures its risk. Scored point-in-time, so the dispatch is backtest-safe.
Geometric attribution bridge	Telescoping compound differences (market → sector → subsector → residual) so multi-period factor contributions sum exactly to the gross compound return — not the naive daily sum, which overstates by volatility drag.
Residual mean-reversion signal	A 5-day cumulative L3 residual return, z-scored by the stock's own 60-day residual volatility. A combo-input factor for multi-signal stacks, not a standalone strategy.

Going deeper

Decomposition philosophy — the Hierarchical Orthogonal Decomposition article series: why residual matters more than headline beta.
Same label, different bet — why two analysts who both "own NVDA" may be running entirely different theses.

For analysts using the SDK, the concepts above are enough. For everything else, the API documentation is authoritative.

RiskModels Concepts — Public Glossary

This page is for analysts, researchers, students, and prospective users. It covers the language, not the implementation.

1. Stock-level risk (bottom-up)

The unit everything aggregates from: decomposing a single equity into factor and residual layers.

Term	Definition
Factor model	A linear model decomposing an asset's return into common factors (market, sector, …) plus a residual. The foundation of the bottom-up unit.
Beta (β)	The slope of an asset's return regressed on a factor; how much the asset moves per unit factor move.
Residual	The part of a stock's return left after the factor model is removed. RiskModels emphasizes the residual as where selection skill lives.
Idiosyncratic	Synonym for residual in classical finance literature. We standardize on residual.
Alpha (α)	Skill-attributable excess return at the position level.
Hedge ratio	The factor exposure you'd offset to neutralize a position's exposure to that factor.
Variance decomposition	Splitting a stock's variance into shares attributable to each factor and the residual. Sums to 1.0.
Return attribution	Splitting a stock's realized return into contributions from each factor. Distinct from variance: a name can be variance-dominated by market yet return-driven by residual.

2. From stocks to portfolios (aggregation)

How position-level risk rolls up. A portfolio is its holdings, weighted — so its exposures are the weighted aggregate of the bottom-up decompositions, plus the covariance between positions.

Term	Definition
Position weight	A holding's share of portfolio value; the weight that scales its risk and return contribution when positions aggregate.
Holdings look-through	Rolling individual holdings up to portfolio-level factor exposure by weight — the bridge from stock risk to portfolio risk.
Peer cohort	A cap-weighted set of similar securities (typically by sector and sub-sector) used as the comparison universe for relative analyses.
Covariance	How two positions move together; why portfolio risk is not just the weighted sum of position risks.
Diversification	The risk reduction from holding imperfectly-correlated positions; the gap between summed position risk and actual portfolio risk.

3. Portfolio performance & risk decomposition

The top of the arc: once positions are aggregated, decompose the portfolio's return and risk the same way — into factor layers, residual, and per-position contributions.

Term	Definition
Portfolio variance decomposition	Splitting total portfolio variance into factor and residual shares, accounting for cross-position covariance. The portfolio-level analog of the single-stock decomposition.
Risk contribution	Each position's (or factor's) share of total portfolio risk, accounting for correlation. Component contributions sum to the total — unlike standalone position risk.
Portfolio attribution	Splitting the portfolio's realized return into contributions from market, sector, subsector, and residual selection.
Active share	The fraction of a portfolio that differs from its benchmark; how much of the book is an active bet versus the index.
Tracking error	The volatility of portfolio-minus-benchmark return; the size of the active risk.
Benchmark fit	How well a portfolio's returns are explained by a benchmark (R², tracking error) — the quality of the comparison.
Sharpe ratio	Return per unit of volatility.
Drawdown	Peak-to-trough decline. Max drawdown is the worst observed.

4. Point-in-time data discipline

How institutional analysis avoids look-ahead bias. Universal vocabulary in regulated industries; the discipline separates serious quant work from naïve backtesting.

Term	Definition
Point-in-time (PIT)	A dataset that respects when each fact was knowable, not just when it was true. Bloomberg and FactSet sell PIT databases as a premium product.
Look-ahead bias	Letting a backtest see information that wasn't yet public on the simulated date. The most common research bug.
Filing lag	The gap between when a fund's holdings reflected reality and when they became publicly knowable via SEC filings. ~45 days for 13F, ~60 days for N-PORT.
Survivorship bias	Backtesting on a universe that excludes delisted or failed entities. Produces optimistic results.

5. Regulatory data

The filings the platform consumes (see also the Filings section for the calendar). Universal vocabulary for institutional analysis of US-listed funds and managers.

Term	Definition
13F	SEC filing (Form 13F-HR) listing institutional managers' US-listed equity holdings. Quarterly, 45-day lag.
N-PORT	SEC filing reporting registered fund (mutual fund / ETF) holdings monthly. Only every third month is publicly disclosed.
N-CEN	Annual SEC filing for registered funds; carries adviser relationships and share-class detail.
CIK	The SEC's identifier for any filer (manager, fund, adviser).
Security master	The canonical record of what is this security and what was it called when? — the foundation under any survivorship-corrected analysis.

6. What RiskModels publishes

The artifacts and surfaces we make available externally.

Term	Definition
RiskModels API	The public REST + MCP surface for decomposition, attribution, and exposure analyses. Documented at riskmodels.app.
`riskmodels-py` (SDK)	The Python client for the API. `pip install riskmodels-py`.
Position snapshot	A one-page institutional analysis of a single equity: factor decomposition, attribution, residual quality, peer context.
Fund snapshot	A one-page tearsheet for a single fund: portfolio attribution, holdings, NAV, benchmark fit.
AOM	Analysis Object Model — our publicly-versioned protocol for describing analyses (subject + lens + view) before they hit any endpoint.

7. How we keep results trustworthy

Engineering vocabulary that lets institutional buyers evaluate whether outputs are reproducible.

Term	Definition
Deterministic	Same input → byte-identical output on re-run. Required for cache correctness and auditability.
Provenance	The metadata recording how an artifact was produced: subject, parameters, as-of date, model version.
Render-once	Web, PDF, PNG, and agent surfaces all read the same canonical artifact. Drift across surfaces is impossible by construction.

8. The RiskModels engine (ERM3)

The hierarchical orthogonal decomposition that produces everything above. The math is standard; the cascade is ours. Full derivation in the .

Term	Definition
ERM3	Equity Risk Model v3 — the engine. Decomposes any US equity into market → sector → subsector → residual via a hierarchical orthogonal cascade.
Hierarchical cascade (L1/L2/L3)	L1 = market (SPY); L2 = sector ETF (e.g. XLK) fit on the L1 residual; L3 = subsector ETF (e.g. SOXX) fit on the L2 residual. Each level explains what the prior one left over.
Link beta (λ)	The coefficient that strips a higher level's exposure from a lower one, so each β captures only the incremental effect of its own level — no double-counting.
Explained risk (ER)	The variance share attributable to each level. ER(L1) + ER(L2) + ER(L3) + residual = 1.0 by construction, verified at runtime to ±0.1%.
Replication equation	The raw-ETF identity that reproduces a stock's return from its hedge legs — only tradeable ETF returns, no orthogonalization at trade time. Verified to ±0.1% for every stock, every date.
Robust beta (Huber-M)	Betas estimated by 252-day rolling Huber-M regression, robust to the fat tails (earnings gaps, single-day reclassifications) that distort ordinary least squares.
Vasicek shrinkage	L2/L3 betas shrunk toward the log-mcap-weighted peer mean within a 4-digit industry cohort; short-history and small-caps are pulled more. Market beta (L1) is not shrunk.
L-star rule	The L* rule: a cost-aware, learned selector that picks the shallowest level whose expected out-of-sample variance reduction outweighs the added hedging cost of going deeper — describing a name at the coarsest level that still captures its risk. Scored point-in-time, so the dispatch is backtest-safe.
Geometric attribution bridge	Telescoping compound differences (market → sector → subsector → residual) so multi-period factor contributions sum exactly to the gross compound return — not the naive daily sum, which overstates by volatility drag.
Residual mean-reversion signal	A 5-day cumulative L3 residual return, z-scored by the stock's own 60-day residual volatility. A combo-input factor for multi-signal stacks, not a standalone strategy.

Going deeper

Decomposition philosophy — the Hierarchical Orthogonal Decomposition article series: why residual matters more than headline beta.
Same label, different bet — why two analysts who both "own NVDA" may be running entirely different theses.

For analysts using the SDK, the concepts above are enough. For everything else, the API documentation is authoritative.

RiskModels Concepts — Public Glossary

This page is for analysts, researchers, students, and prospective users. It covers the language, not the implementation.

1. Stock-level risk (bottom-up)

The unit everything aggregates from: decomposing a single equity into factor and residual layers.

Term	Definition
Factor model	A linear model decomposing an asset's return into common factors (market, sector, …) plus a residual. The foundation of the bottom-up unit.
Beta (β)	The slope of an asset's return regressed on a factor; how much the asset moves per unit factor move.
Residual	The part of a stock's return left after the factor model is removed. RiskModels emphasizes the residual as where selection skill lives.
Idiosyncratic	Synonym for residual in classical finance literature. We standardize on residual.
Alpha (α)	Skill-attributable excess return at the position level.
Hedge ratio	The factor exposure you'd offset to neutralize a position's exposure to that factor.
Variance decomposition	Splitting a stock's variance into shares attributable to each factor and the residual. Sums to 1.0.
Return attribution	Splitting a stock's realized return into contributions from each factor. Distinct from variance: a name can be variance-dominated by market yet return-driven by residual.

2. From stocks to portfolios (aggregation)

How position-level risk rolls up. A portfolio is its holdings, weighted — so its exposures are the weighted aggregate of the bottom-up decompositions, plus the covariance between positions.

Term	Definition
Position weight	A holding's share of portfolio value; the weight that scales its risk and return contribution when positions aggregate.
Holdings look-through	Rolling individual holdings up to portfolio-level factor exposure by weight — the bridge from stock risk to portfolio risk.
Peer cohort	A cap-weighted set of similar securities (typically by sector and sub-sector) used as the comparison universe for relative analyses.
Covariance	How two positions move together; why portfolio risk is not just the weighted sum of position risks.
Diversification	The risk reduction from holding imperfectly-correlated positions; the gap between summed position risk and actual portfolio risk.

3. Portfolio performance & risk decomposition

The top of the arc: once positions are aggregated, decompose the portfolio's return and risk the same way — into factor layers, residual, and per-position contributions.

Term	Definition
Portfolio variance decomposition	Splitting total portfolio variance into factor and residual shares, accounting for cross-position covariance. The portfolio-level analog of the single-stock decomposition.
Risk contribution	Each position's (or factor's) share of total portfolio risk, accounting for correlation. Component contributions sum to the total — unlike standalone position risk.
Portfolio attribution	Splitting the portfolio's realized return into contributions from market, sector, subsector, and residual selection.
Active share	The fraction of a portfolio that differs from its benchmark; how much of the book is an active bet versus the index.
Tracking error	The volatility of portfolio-minus-benchmark return; the size of the active risk.
Benchmark fit	How well a portfolio's returns are explained by a benchmark (R², tracking error) — the quality of the comparison.
Sharpe ratio	Return per unit of volatility.
Drawdown	Peak-to-trough decline. Max drawdown is the worst observed.

4. Point-in-time data discipline

How institutional analysis avoids look-ahead bias. Universal vocabulary in regulated industries; the discipline separates serious quant work from naïve backtesting.

Term	Definition
Point-in-time (PIT)	A dataset that respects when each fact was knowable, not just when it was true. Bloomberg and FactSet sell PIT databases as a premium product.
Look-ahead bias	Letting a backtest see information that wasn't yet public on the simulated date. The most common research bug.
Filing lag	The gap between when a fund's holdings reflected reality and when they became publicly knowable via SEC filings. ~45 days for 13F, ~60 days for N-PORT.
Survivorship bias	Backtesting on a universe that excludes delisted or failed entities. Produces optimistic results.

5. Regulatory data

The filings the platform consumes (see also the Filings section for the calendar). Universal vocabulary for institutional analysis of US-listed funds and managers.

Term	Definition
13F	SEC filing (Form 13F-HR) listing institutional managers' US-listed equity holdings. Quarterly, 45-day lag.
N-PORT	SEC filing reporting registered fund (mutual fund / ETF) holdings monthly. Only every third month is publicly disclosed.
N-CEN	Annual SEC filing for registered funds; carries adviser relationships and share-class detail.
CIK	The SEC's identifier for any filer (manager, fund, adviser).
Security master	The canonical record of what is this security and what was it called when? — the foundation under any survivorship-corrected analysis.

6. What RiskModels publishes

The artifacts and surfaces we make available externally.

Term	Definition
RiskModels API	The public REST + MCP surface for decomposition, attribution, and exposure analyses. Documented at riskmodels.app.
`riskmodels-py` (SDK)	The Python client for the API. `pip install riskmodels-py`.
Position snapshot	A one-page institutional analysis of a single equity: factor decomposition, attribution, residual quality, peer context.
Fund snapshot	A one-page tearsheet for a single fund: portfolio attribution, holdings, NAV, benchmark fit.
AOM	Analysis Object Model — our publicly-versioned protocol for describing analyses (subject + lens + view) before they hit any endpoint.

7. How we keep results trustworthy

Engineering vocabulary that lets institutional buyers evaluate whether outputs are reproducible.

Term	Definition
Deterministic	Same input → byte-identical output on re-run. Required for cache correctness and auditability.
Provenance	The metadata recording how an artifact was produced: subject, parameters, as-of date, model version.
Render-once	Web, PDF, PNG, and agent surfaces all read the same canonical artifact. Drift across surfaces is impossible by construction.

8. The RiskModels engine (ERM3)

The hierarchical orthogonal decomposition that produces everything above. The math is standard; the cascade is ours. Full derivation in the .

Term	Definition
ERM3	Equity Risk Model v3 — the engine. Decomposes any US equity into market → sector → subsector → residual via a hierarchical orthogonal cascade.
Hierarchical cascade (L1/L2/L3)	L1 = market (SPY); L2 = sector ETF (e.g. XLK) fit on the L1 residual; L3 = subsector ETF (e.g. SOXX) fit on the L2 residual. Each level explains what the prior one left over.
Link beta (λ)	The coefficient that strips a higher level's exposure from a lower one, so each β captures only the incremental effect of its own level — no double-counting.
Explained risk (ER)	The variance share attributable to each level. ER(L1) + ER(L2) + ER(L3) + residual = 1.0 by construction, verified at runtime to ±0.1%.
Replication equation	The raw-ETF identity that reproduces a stock's return from its hedge legs — only tradeable ETF returns, no orthogonalization at trade time. Verified to ±0.1% for every stock, every date.
Robust beta (Huber-M)	Betas estimated by 252-day rolling Huber-M regression, robust to the fat tails (earnings gaps, single-day reclassifications) that distort ordinary least squares.
Vasicek shrinkage	L2/L3 betas shrunk toward the log-mcap-weighted peer mean within a 4-digit industry cohort; short-history and small-caps are pulled more. Market beta (L1) is not shrunk.
L-star rule	The L* rule: a cost-aware, learned selector that picks the shallowest level whose expected out-of-sample variance reduction outweighs the added hedging cost of going deeper — describing a name at the coarsest level that still captures its risk. Scored point-in-time, so the dispatch is backtest-safe.
Geometric attribution bridge	Telescoping compound differences (market → sector → subsector → residual) so multi-period factor contributions sum exactly to the gross compound return — not the naive daily sum, which overstates by volatility drag.
Residual mean-reversion signal	A 5-day cumulative L3 residual return, z-scored by the stock's own 60-day residual volatility. A combo-input factor for multi-signal stacks, not a standalone strategy.

Going deeper

Decomposition philosophy — the Hierarchical Orthogonal Decomposition article series: why residual matters more than headline beta.
Same label, different bet — why two analysts who both "own NVDA" may be running entirely different theses.

For analysts using the SDK, the concepts above are enough. For everything else, the API documentation is authoritative.