Methodology
Last updated:
Please read the Methodology below.
Digital-asset valuation has lacked a defensible standard. Sentiment-driven prices, opaque proprietary models, and speculative forecasts have left domain experts without a consistent answer to a simple question: what is this asset actually worth?
AIREPORT's Intrinsic Engine™ produces Structural Value (SV) — a deterministic, bidirectional fair-value assessment with a documented empirical record across more than 2,100 measured market moments and more than two years of history. Inputs are traceable to their source; outputs are reproducible on demand.
Methodological Highlights
Aggregated market data. Foundation data drawn from CoinMarketCap, which aggregates price, volume, market capitalization, and circulating supply across hundreds of digital-asset exchanges globally.
Appraisal, not prediction. Structural Value provides a fair-value assessment based on each asset's market behavior. It is not a forecast or trading signal.
Bidirectional by design. Universal methodology assesses both overvaluation and undervaluation with a single deterministic formula — no per-asset tuning or custom parameters.
On-chain supply is built in. Circulating-supply changes and supply-normalized comparisons are part of every analysis — ensuring valuation comparisons across assets with structurally different supply profiles remain meaningful.
Documented empirical record. Across a governed frame of 29 digital assets and more than two years of history, 89.7% of assessments on market-priced assets — assets that trade freely rather than tracking a peg — are confirmed by subsequent market behavior within 30 days and 93.2% within 90 days. Full breakdown, including per-class results and how the frame is governed, in Validation.
Reproducible and timestamped. Every calculation is deterministic and saved as a permanent record — suitable for the standard of scrutiny professional work requires.
Who This Serves
AIREPORT is built for professionals who need a defensible, objective valuation of digital assets — not a trading signal, not a speculative forecast, and not a range that requires further interpretation.
For family law attorneys, Structural Value (SV) provides an impartial, mathematically derived figure for equitable distribution of digital asset holdings. One number per asset, backed by a consistent and durable methodology resistant to bias claims. The valuation produced last month and the reassessment produced next week come from the same documented, deterministic process — and when the engine generation advances, the change is dated, tested against the prior generation, and disclosed on this page (see Changeover record), so results remain comparable and defensible.
For financial advisors and wealth managers, Structural Value brings the same analytical rigor to digital asset holdings that they apply to every other asset class. When a client holds digital assets, Structural Value provides a firm number for rebalancing, reporting, and recordkeeping.
What Structural Value Measures
Structural Value assesses an asset's fair value through measures of internal coherence and consistent market behavior. Each value assessment is a statement of what the asset structurally merits at that time.
The mental model is appraisal, not prediction. A property appraiser reports value based on comparable sales and condition factors, not future sale prices. Structural Value reports the fair value of digital assets using historical consistency patterns and integrity metrics.
Market prices routinely deviate from equilibrium. Sentiment, liquidity events, and speculation push price away from fair value — sometimes for days, other times for months. Structural Value represents the gravitational center toward which price tends to return; this is a metaphor for the observable pattern, not a claim about market mechanics. The underlying mechanism is the evidence-weighted, regime-aware calibration described in How the Engine Works.
For professionals, Structural Value is a fair-value reference point, not a price prediction — validated by whether the market reaches the assessed value within defined forward windows. The standard horizons are 30 and 90 days; results for both are published under Validation.
Bidirectional Assessment
The methodology assesses overvaluation and undervaluation through a single deterministic mechanism. When SV sits below market price, the engine assesses overvaluation; when SV sits above market price, undervaluation. Both directions are structurally meaningful; neither is a speculative call.
The same formula runs for every asset the engine evaluates, with no manual parameters, per-asset tuning, or human input. The magnitude of the assessment in each direction scales with the strength of the evidence: weak signals produce small adjustments, strong signals produce larger ones.
For attorneys, this means the methodology is resistant to challenge as directionally biased — it applies the same analysis regardless of whether the result favors or disfavors either party. For advisors, the methodology offers a consistent way to assess a digital asset allocation rather than using asset-specific models.
This holds for stablecoins as well. Because a peg-tracking asset shows little structural divergence between current price and its own historical price behavior, SV converges toward market price by the same mechanism that governs every other asset — the same formula that would place a volatile altcoin's assessed value 12% above its market price produces a near-zero divergence for USDC. The SV calculation applies identically to stablecoins and volatile assets; depeg episodes surface through a dedicated regime-change detector calibrated to peg-tracking behavior.
How the Engine Works
The Intrinsic Engine™ evaluates each asset against its own history across six distinct time horizons, each derived from a mathematically structured reflection period spanning short-term through long-term horizons. Rather than relying on a single lookback window, the system examines how consistently an asset's current price relates to its own structural history across multiple timeframes simultaneously. The empirical record behind the current production engine draws on more than 13,000 dated snapshot computations — each capturing the engine's full set of structural factor measures for one asset, date, and time horizon.
The engine combines three independent inputs:
Structural integrity. A quality measure of the data environment — how stable, consistent, and well-behaved the asset's relationship to its own history is within and among each horizon.
Non-linear consistency. A measure of how consistently price relates to reflection across the six horizons. The non-linear transform dampens noisy or inconsistent signals more than proportionally, preventing weak evidence from producing strong assessments.
Regime-aware calibration. Per-asset confidence is calibrated against real-time market conditions. When recent volatility is contracting — and, as of the current generation, when it is rapidly expanding — the engine expresses its assessments more conservatively, drawing the assessed value closer to the observed market price; full confidence is reserved for stable conditions.
The result is a single synthesized fair-value number — not a range — reflecting the strongest available structural evidence, tempered by current market conditions through the regime-aware calibration described above.
The current Intrinsic Engine generation (v15) powers the production analytics. Each generation tightened what came before — eliminating factors, special cases, and tuning parameters that testing showed were carrying noise rather than signal. The v15 generation extended the regime-aware calibration to expanding-volatility conditions after paired testing on every asset in the validation frame showed the change confirmed more assessments while never converting a previously confirmed assessment into a miss (see Changeover record under Validation).
Data Sources
Market data — price, OHLCV (daily open/high/low/close/volume), market capitalization, and circulating supply — is sourced from CoinMarketCap's aggregated exchange feeds. Aggregation matters: rather than relying on any single exchange's prices (each carrying its own liquidity profile, regional concentration, and reporting cadence), the data reflects activity across the venues where the asset actually trades.
Circulating supply is treated as a first-class input. On-chain supply dynamics — issuance, burns, lockups, and net distribution changes — are reflected in the data the engine consumes, ensuring that valuation comparisons across assets with structurally different supply profiles remain meaningful. Market capitalization is the product of price and supply; an honest assessment requires both inputs to be current and verifiable.
All ingested data is traceable to its origin and timestamped at the moment of ingestion. The data the engine sees today is the data the audit trail preserves for tomorrow.
Asset Integrity Score
The Asset Integrity Score (AIS) is a separate system that rates an asset's data health on a scale of 0–100%.
AIS is built from a focused subset of the engine's structural integrity factors, selected for low mutual correlation so the score captures independent dimensions of integrity — market positioning, model-agreement stability, and historical elasticity. All are measured across the same six composite time horizons, giving a complete view of each asset's analytical strength at any time.
AIS and SV are complementary but independent. SV is the engine's fair-value assessment — the substantiated number a professional reports, with the engine's non-linear and regime-aware calibration already applied. AIS is a separate measurement of the asset itself: how structurally coherent its data behavior is across time. A professional does not use AIS to interpret SV; that calibration is already applied within the SV calculation itself. AIS answers a different question entirely: not what is this asset worth? but how structurally coherent is this asset compared to others in the portfolio?
For a multi-asset portfolio manager, AIS is the asset-characterization signal — it surfaces which holdings behave with consistent structural patterns and which carry noisier underlying data. The fair-value assessment for each holding stands on its own.
Structural Change Detection
AIREPORT monitors portfolio assets for structural changes using automated anomaly detection. It tracks 11 types of analytical events across 7 categories: valuation extremes, integrity deterioration, model regime shifts, supply disruptions, equilibrium bias changes, stablecoin regime changes, and systemic cross-asset patterns.
Detection thresholds adapt per asset against its historical baseline, not fixed values. This prevents false positives for volatile assets while remaining sensitive to stable ones. A second-stage detector adds market-stress awareness: acute anomaly thresholds tighten when a composite stress indicator derived from frame-wide conditions is elevated.
For wealth managers, this is the structural-change monitor. Shifts in the asset's profile appear as plain-language alerts with explanations as soon as data supports them.
Validation
How market confirmation is measured
The validation standard measures whether the market actually reaches the assessed value. For a bearish assessment (SV below market), a touch occurs if the market's daily low reaches SV on any day within the forward window. For a bullish assessment (SV above market), a touch occurs if the market's daily high reaches SV on any day within the forward window.
This standard uses intraday high/low prices by design rather than closing prices. A closing-price standard would credit the assessment only when the day ended at the assessed level; an intraday standard credits any moment the market transacted at that level, including brief wicks. The methodology chooses the intraday standard because it tests whether the assessed value is one the market actually engaged with at any point in the window, rather than only at the daily close.
The validation frame
The empirical record is measured against a governed frame of 29 digital assets organized into five classes by hard, measurable criteria only: two fiat-pegged and two gold-pegged value-reference assets (classified by a measured peg test), and large-, mid-, and small-capitalization market-priced assets (classified by market-capitalization rank bands). Membership is governed by a written charter fixed in advance of any selection: additions follow a mechanical top-of-band rule at fixed calendar windows, membership is additive-only, and every selection decision is recorded against an archived market snapshot. No asset is chosen — or removed — because of how it performs.
Two disclosures accompany the frame. First, the founding members predate the charter and are grandfathered — the mechanical rule governs additions only. Second, two assets were added out of schedule in June 2026 as engine test assets and, because membership is additive-only, retained: they improved the frame's confirmation rates while widening its reported worst-case deviation.
Headline result
The pegged classes track their reference by construction, so they are reported separately and never enter the headline. Across the market-priced classes, the current production engine has been tested against more than 2,100 real market moments over more than two years. 89.7% of assessments are confirmed within 30 days; 93.2% within 90 days. Bearish assessments confirm at 92.2% / 95.9% (30d / 90d); bullish assessments confirm at 88.4% / 91.7%. At each refresh, the pooled figure is checked against the average of the per-class rates and is retired in favor of per-class reporting if the two ever diverge by more than one percentage point; at this refresh they agree within 0.6 percentage points. The result is reproducible from backtest data on demand and updates as market history accumulates — a living standard, not a single measurement. (Figures throughout this page are as of July 7, 2026.)
Results by class
Confirmation rates within 30 and 90 days, by class:
Fiat-pegged (2 assets): 100% / 100% — every assessment confirmed, reflecting peg-tracking behavior by construction.
Gold-pegged (2 assets): 84.2% / 86.1% — reported separately because confirmation here measures how tightly these assets track their reference, not the engine's performance on freely traded assets.
Large-capitalization (6 assets): 89.2% / 93.2%.
Mid-capitalization (13 assets): 88.7% / 92.3%.
Small-capitalization (6 assets): 92.8% / 95.4%.
Assessment magnitude — how far the assessed value sits from market price at the time of assessment — is reported per class, because it differs by class in a way a single pooled number would obscure: 95th-percentile magnitudes — the distance from market that 19 of every 20 assessments stay within — are about 1.6% for gold-pegged, 5.6% for large-, 5.9% for mid-, and 6.1% for small-capitalization assets, with fiat-pegged assessments essentially at market. The largest single assessment gap observed anywhere in the record is 19.7%, in the mid-capitalization class.
One statistical caution applies to every rate on this page. A new assessment is generated each week, but each is judged over a 30- or 90-day forward window — so consecutive weekly assessments of the same asset share most of their measurement window and do not succeed or fail independently. The rates are exact as counts of what happened. The caution concerns only precision: because the observations overlap, the record carries the statistical weight of a smaller number of independent observations, and the margin of uncertainty around each rate is therefore wider than the raw count would imply.
Reading the validation results
For a specific asset, the relevant empirical record is its class result; the pooled figure characterizes the frame as a whole. Because sample sizes differ by class, cross-class comparisons of assessment magnitude use the 95th-percentile figures; the single worst case is stated with its class. Worst-case values are per-class extremes among that class's observed record, not forecasts of any asset's behavior.
The confirmation rate is calibration evidence, not prediction accuracy: it demonstrates that assessed values are levels the market actually transacts at, not arbitrary or theoretical figures. The standard is falsifiable and published — a specific assessed value, in a specific direction, within a stated window; either the market reached it or it did not. If confirmation at this rate were market noise wicking through nearby levels, the failures would be random and unbounded. They are neither: every assessment is made at a disclosed distance from market (the magnitude figures above), so a miss is a bounded, stated level the market did not reach — not an open-ended error — and the misses divide by direction in the pattern reported below. The same standard holds across daily, weekly, monthly, and quarterly windows — 54.7% of assessments confirm within a single day, and confirmation builds as the window extends — so the 30-day figure is a standard reporting horizon, not a window selected for its results. Noise does not produce that structure.
Per-asset consistency
Per-asset 30-day confirmation ranges from approximately 80% to 97% across the market-priced frame, clustered around a median near 90% — essentially the frame-wide rate, indicating the pooled figure reflects the typical asset rather than a few strong performers. Bitcoin's confirmation rate sits in the upper part of this range, consistent with its established structural depth; the methodology applies the same deterministic process to every asset, with no per-asset tuning. The spread reflects each asset's own structural characteristics. Across the frame as a whole, confirmation accumulates as the window lengthens — from daily through weekly, monthly, and quarterly horizons — indicating the headline result reflects sustained engagement with assessed levels rather than a single favorable window.
When assessments are not confirmed
Approximately 10.3% of assessments on market-priced assets are not confirmed by the market within the 30-day measurement window — 7.8% of bearish assessments and 11.6% of bullish assessments. Bullish assessments historically confirm at a lower rate; the record reports the two directions separately rather than blending them. An unconfirmed assessment is not necessarily a wrong one; the market may have been approaching the level without reaching it inside the window. The Results by class section above reports how far assessments sit from market by class — the bounded-output property described under How to Read the Number keeps those gaps disciplined in both directions.
Changeover record (July 2026)
This page previously reported the record of the prior engine generation (v14) against a fixed 15-asset validation cohort; as of July 6, 2026 that record stood at 87.7% / 91.6% (30d / 90d) across more than 1,400 assessments, and it remains accurate as of its date. In July 2026 two deliberate changes took effect together: the production engine advanced to v15 after paired testing across every asset in the frame showed it confirmed more assessments in every class while never converting a previously confirmed assessment into a miss, and the published record moved to the charter-governed 29-asset frame with per-class reporting. Because both changes are dated and both records are published — the prior figure preserved above, the paired-test result isolating the engine's effect, and the per-class results showing the frame's composition — the movement of the headline is fully accounted for rather than left to inference. Under the previous convention, a single pooled worst-case deviation was reported; that convention is retired in favor of the per-class envelope above — 95th-percentile magnitudes by class, with the single largest gap in the whole record still stated (19.7%, mid-capitalization) — because a pooled worst case imputes the widest class's extreme to every asset. Prior published figures remain derivable from the preserved record.
How to Read the Number
Wider gaps carry less conviction. The engine's bounded design dampens the assessment as the gap between SV and market price widens. Large SV–market gaps signal weaker structural conviction, not stronger — a bounded-output property of the methodology.
Market validation through price action. The intraday touches described under Validation are direct market confirmations of the fair-value assessment — they show the assessed level is one the market actually transacted at, not a theoretical estimate.
Built for professional timeframes. Structural Value is designed for professionals who assess value over weeks, months, and quarters — estate valuations, fund reporting, fiduciary due diligence, equitable distribution, and settlement analysis. It is not a trading signal.
Deterministic and auditable. The number is built to withstand professional scrutiny: a documented methodology, a complete audit trail, and a reproducible calculation path from input to output.
What You Receive
AIREPORT generates professional reports showing portfolio-level and per-asset structural valuations alongside integrity scores and structural-change events. Whether the report supports a settlement negotiation, a quarterly client review, or a fiduciary filing, the methodology is consistent and the audit trail is complete.
Contact Information
Questions about the Methodology may be directed to care@keepgood.co.
Methodology
Last updated:
Please read the Methodology below.
Digital-asset valuation has lacked a defensible standard. Sentiment-driven prices, opaque proprietary models, and speculative forecasts have left domain experts without a consistent answer to a simple question: what is this asset actually worth?
AIREPORT's Intrinsic Engine™ produces Structural Value (SV) — a deterministic, bidirectional fair-value assessment with a documented empirical record across more than 2,100 measured market moments and more than two years of history. Inputs are traceable to their source; outputs are reproducible on demand.
Methodological Highlights
Aggregated market data. Foundation data drawn from CoinMarketCap, which aggregates price, volume, market capitalization, and circulating supply across hundreds of digital-asset exchanges globally.
Appraisal, not prediction. Structural Value provides a fair-value assessment based on each asset's market behavior. It is not a forecast or trading signal.
Bidirectional by design. Universal methodology assesses both overvaluation and undervaluation with a single deterministic formula — no per-asset tuning or custom parameters.
On-chain supply is built in. Circulating-supply changes and supply-normalized comparisons are part of every analysis — ensuring valuation comparisons across assets with structurally different supply profiles remain meaningful.
Documented empirical record. Across a governed frame of 29 digital assets and more than two years of history, 89.7% of assessments on market-priced assets — assets that trade freely rather than tracking a peg — are confirmed by subsequent market behavior within 30 days and 93.2% within 90 days. Full breakdown, including per-class results and how the frame is governed, in Validation.
Reproducible and timestamped. Every calculation is deterministic and saved as a permanent record — suitable for the standard of scrutiny professional work requires.
Who This Serves
AIREPORT is built for professionals who need a defensible, objective valuation of digital assets — not a trading signal, not a speculative forecast, and not a range that requires further interpretation.
For family law attorneys, Structural Value (SV) provides an impartial, mathematically derived figure for equitable distribution of digital asset holdings. One number per asset, backed by a consistent and durable methodology resistant to bias claims. The valuation produced last month and the reassessment produced next week come from the same documented, deterministic process — and when the engine generation advances, the change is dated, tested against the prior generation, and disclosed on this page (see Changeover record), so results remain comparable and defensible.
For financial advisors and wealth managers, Structural Value brings the same analytical rigor to digital asset holdings that they apply to every other asset class. When a client holds digital assets, Structural Value provides a firm number for rebalancing, reporting, and recordkeeping.
What Structural Value Measures
Structural Value assesses an asset's fair value through measures of internal coherence and consistent market behavior. Each value assessment is a statement of what the asset structurally merits at that time.
The mental model is appraisal, not prediction. A property appraiser reports value based on comparable sales and condition factors, not future sale prices. Structural Value reports the fair value of digital assets using historical consistency patterns and integrity metrics.
Market prices routinely deviate from equilibrium. Sentiment, liquidity events, and speculation push price away from fair value — sometimes for days, other times for months. Structural Value represents the gravitational center toward which price tends to return; this is a metaphor for the observable pattern, not a claim about market mechanics. The underlying mechanism is the evidence-weighted, regime-aware calibration described in How the Engine Works.
For professionals, Structural Value is a fair-value reference point, not a price prediction — validated by whether the market reaches the assessed value within defined forward windows. The standard horizons are 30 and 90 days; results for both are published under Validation.
Bidirectional Assessment
The methodology assesses overvaluation and undervaluation through a single deterministic mechanism. When SV sits below market price, the engine assesses overvaluation; when SV sits above market price, undervaluation. Both directions are structurally meaningful; neither is a speculative call.
The same formula runs for every asset the engine evaluates, with no manual parameters, per-asset tuning, or human input. The magnitude of the assessment in each direction scales with the strength of the evidence: weak signals produce small adjustments, strong signals produce larger ones.
For attorneys, this means the methodology is resistant to challenge as directionally biased — it applies the same analysis regardless of whether the result favors or disfavors either party. For advisors, the methodology offers a consistent way to assess a digital asset allocation rather than using asset-specific models.
This holds for stablecoins as well. Because a peg-tracking asset shows little structural divergence between current price and its own historical price behavior, SV converges toward market price by the same mechanism that governs every other asset — the same formula that would place a volatile altcoin's assessed value 12% above its market price produces a near-zero divergence for USDC. The SV calculation applies identically to stablecoins and volatile assets; depeg episodes surface through a dedicated regime-change detector calibrated to peg-tracking behavior.
How the Engine Works
The Intrinsic Engine™ evaluates each asset against its own history across six distinct time horizons, each derived from a mathematically structured reflection period spanning short-term through long-term horizons. Rather than relying on a single lookback window, the system examines how consistently an asset's current price relates to its own structural history across multiple timeframes simultaneously. The empirical record behind the current production engine draws on more than 13,000 dated snapshot computations — each capturing the engine's full set of structural factor measures for one asset, date, and time horizon.
The engine combines three independent inputs:
Structural integrity. A quality measure of the data environment — how stable, consistent, and well-behaved the asset's relationship to its own history is within and among each horizon.
Non-linear consistency. A measure of how consistently price relates to reflection across the six horizons. The non-linear transform dampens noisy or inconsistent signals more than proportionally, preventing weak evidence from producing strong assessments.
Regime-aware calibration. Per-asset confidence is calibrated against real-time market conditions. When recent volatility is contracting — and, as of the current generation, when it is rapidly expanding — the engine expresses its assessments more conservatively, drawing the assessed value closer to the observed market price; full confidence is reserved for stable conditions.
The result is a single synthesized fair-value number — not a range — reflecting the strongest available structural evidence, tempered by current market conditions through the regime-aware calibration described above.
The current Intrinsic Engine generation (v15) powers the production analytics. Each generation tightened what came before — eliminating factors, special cases, and tuning parameters that testing showed were carrying noise rather than signal. The v15 generation extended the regime-aware calibration to expanding-volatility conditions after paired testing on every asset in the validation frame showed the change confirmed more assessments while never converting a previously confirmed assessment into a miss (see Changeover record under Validation).
Data Sources
Market data — price, OHLCV (daily open/high/low/close/volume), market capitalization, and circulating supply — is sourced from CoinMarketCap's aggregated exchange feeds. Aggregation matters: rather than relying on any single exchange's prices (each carrying its own liquidity profile, regional concentration, and reporting cadence), the data reflects activity across the venues where the asset actually trades.
Circulating supply is treated as a first-class input. On-chain supply dynamics — issuance, burns, lockups, and net distribution changes — are reflected in the data the engine consumes, ensuring that valuation comparisons across assets with structurally different supply profiles remain meaningful. Market capitalization is the product of price and supply; an honest assessment requires both inputs to be current and verifiable.
All ingested data is traceable to its origin and timestamped at the moment of ingestion. The data the engine sees today is the data the audit trail preserves for tomorrow.
Asset Integrity Score
The Asset Integrity Score (AIS) is a separate system that rates an asset's data health on a scale of 0–100%.
AIS is built from a focused subset of the engine's structural integrity factors, selected for low mutual correlation so the score captures independent dimensions of integrity — market positioning, model-agreement stability, and historical elasticity. All are measured across the same six composite time horizons, giving a complete view of each asset's analytical strength at any time.
AIS and SV are complementary but independent. SV is the engine's fair-value assessment — the substantiated number a professional reports, with the engine's non-linear and regime-aware calibration already applied. AIS is a separate measurement of the asset itself: how structurally coherent its data behavior is across time. A professional does not use AIS to interpret SV; that calibration is already applied within the SV calculation itself. AIS answers a different question entirely: not what is this asset worth? but how structurally coherent is this asset compared to others in the portfolio?
For a multi-asset portfolio manager, AIS is the asset-characterization signal — it surfaces which holdings behave with consistent structural patterns and which carry noisier underlying data. The fair-value assessment for each holding stands on its own.
Structural Change Detection
AIREPORT monitors portfolio assets for structural changes using automated anomaly detection. It tracks 11 types of analytical events across 7 categories: valuation extremes, integrity deterioration, model regime shifts, supply disruptions, equilibrium bias changes, stablecoin regime changes, and systemic cross-asset patterns.
Detection thresholds adapt per asset against its historical baseline, not fixed values. This prevents false positives for volatile assets while remaining sensitive to stable ones. A second-stage detector adds market-stress awareness: acute anomaly thresholds tighten when a composite stress indicator derived from frame-wide conditions is elevated.
For wealth managers, this is the structural-change monitor. Shifts in the asset's profile appear as plain-language alerts with explanations as soon as data supports them.
Validation
How market confirmation is measured
The validation standard measures whether the market actually reaches the assessed value. For a bearish assessment (SV below market), a touch occurs if the market's daily low reaches SV on any day within the forward window. For a bullish assessment (SV above market), a touch occurs if the market's daily high reaches SV on any day within the forward window.
This standard uses intraday high/low prices by design rather than closing prices. A closing-price standard would credit the assessment only when the day ended at the assessed level; an intraday standard credits any moment the market transacted at that level, including brief wicks. The methodology chooses the intraday standard because it tests whether the assessed value is one the market actually engaged with at any point in the window, rather than only at the daily close.
The validation frame
The empirical record is measured against a governed frame of 29 digital assets organized into five classes by hard, measurable criteria only: two fiat-pegged and two gold-pegged value-reference assets (classified by a measured peg test), and large-, mid-, and small-capitalization market-priced assets (classified by market-capitalization rank bands). Membership is governed by a written charter fixed in advance of any selection: additions follow a mechanical top-of-band rule at fixed calendar windows, membership is additive-only, and every selection decision is recorded against an archived market snapshot. No asset is chosen — or removed — because of how it performs.
Two disclosures accompany the frame. First, the founding members predate the charter and are grandfathered — the mechanical rule governs additions only. Second, two assets were added out of schedule in June 2026 as engine test assets and, because membership is additive-only, retained: they improved the frame's confirmation rates while widening its reported worst-case deviation.
Headline result
The pegged classes track their reference by construction, so they are reported separately and never enter the headline. Across the market-priced classes, the current production engine has been tested against more than 2,100 real market moments over more than two years. 89.7% of assessments are confirmed within 30 days; 93.2% within 90 days. Bearish assessments confirm at 92.2% / 95.9% (30d / 90d); bullish assessments confirm at 88.4% / 91.7%. At each refresh, the pooled figure is checked against the average of the per-class rates and is retired in favor of per-class reporting if the two ever diverge by more than one percentage point; at this refresh they agree within 0.6 percentage points. The result is reproducible from backtest data on demand and updates as market history accumulates — a living standard, not a single measurement. (Figures throughout this page are as of July 7, 2026.)
Results by class
Confirmation rates within 30 and 90 days, by class:
Fiat-pegged (2 assets): 100% / 100% — every assessment confirmed, reflecting peg-tracking behavior by construction.
Gold-pegged (2 assets): 84.2% / 86.1% — reported separately because confirmation here measures how tightly these assets track their reference, not the engine's performance on freely traded assets.
Large-capitalization (6 assets): 89.2% / 93.2%.
Mid-capitalization (13 assets): 88.7% / 92.3%.
Small-capitalization (6 assets): 92.8% / 95.4%.
Assessment magnitude — how far the assessed value sits from market price at the time of assessment — is reported per class, because it differs by class in a way a single pooled number would obscure: 95th-percentile magnitudes — the distance from market that 19 of every 20 assessments stay within — are about 1.6% for gold-pegged, 5.6% for large-, 5.9% for mid-, and 6.1% for small-capitalization assets, with fiat-pegged assessments essentially at market. The largest single assessment gap observed anywhere in the record is 19.7%, in the mid-capitalization class.
One statistical caution applies to every rate on this page. A new assessment is generated each week, but each is judged over a 30- or 90-day forward window — so consecutive weekly assessments of the same asset share most of their measurement window and do not succeed or fail independently. The rates are exact as counts of what happened. The caution concerns only precision: because the observations overlap, the record carries the statistical weight of a smaller number of independent observations, and the margin of uncertainty around each rate is therefore wider than the raw count would imply.
Reading the validation results
For a specific asset, the relevant empirical record is its class result; the pooled figure characterizes the frame as a whole. Because sample sizes differ by class, cross-class comparisons of assessment magnitude use the 95th-percentile figures; the single worst case is stated with its class. Worst-case values are per-class extremes among that class's observed record, not forecasts of any asset's behavior.
The confirmation rate is calibration evidence, not prediction accuracy: it demonstrates that assessed values are levels the market actually transacts at, not arbitrary or theoretical figures. The standard is falsifiable and published — a specific assessed value, in a specific direction, within a stated window; either the market reached it or it did not. If confirmation at this rate were market noise wicking through nearby levels, the failures would be random and unbounded. They are neither: every assessment is made at a disclosed distance from market (the magnitude figures above), so a miss is a bounded, stated level the market did not reach — not an open-ended error — and the misses divide by direction in the pattern reported below. The same standard holds across daily, weekly, monthly, and quarterly windows — 54.7% of assessments confirm within a single day, and confirmation builds as the window extends — so the 30-day figure is a standard reporting horizon, not a window selected for its results. Noise does not produce that structure.
Per-asset consistency
Per-asset 30-day confirmation ranges from approximately 80% to 97% across the market-priced frame, clustered around a median near 90% — essentially the frame-wide rate, indicating the pooled figure reflects the typical asset rather than a few strong performers. Bitcoin's confirmation rate sits in the upper part of this range, consistent with its established structural depth; the methodology applies the same deterministic process to every asset, with no per-asset tuning. The spread reflects each asset's own structural characteristics. Across the frame as a whole, confirmation accumulates as the window lengthens — from daily through weekly, monthly, and quarterly horizons — indicating the headline result reflects sustained engagement with assessed levels rather than a single favorable window.
When assessments are not confirmed
Approximately 10.3% of assessments on market-priced assets are not confirmed by the market within the 30-day measurement window — 7.8% of bearish assessments and 11.6% of bullish assessments. Bullish assessments historically confirm at a lower rate; the record reports the two directions separately rather than blending them. An unconfirmed assessment is not necessarily a wrong one; the market may have been approaching the level without reaching it inside the window. The Results by class section above reports how far assessments sit from market by class — the bounded-output property described under How to Read the Number keeps those gaps disciplined in both directions.
Changeover record (July 2026)
This page previously reported the record of the prior engine generation (v14) against a fixed 15-asset validation cohort; as of July 6, 2026 that record stood at 87.7% / 91.6% (30d / 90d) across more than 1,400 assessments, and it remains accurate as of its date. In July 2026 two deliberate changes took effect together: the production engine advanced to v15 after paired testing across every asset in the frame showed it confirmed more assessments in every class while never converting a previously confirmed assessment into a miss, and the published record moved to the charter-governed 29-asset frame with per-class reporting. Because both changes are dated and both records are published — the prior figure preserved above, the paired-test result isolating the engine's effect, and the per-class results showing the frame's composition — the movement of the headline is fully accounted for rather than left to inference. Under the previous convention, a single pooled worst-case deviation was reported; that convention is retired in favor of the per-class envelope above — 95th-percentile magnitudes by class, with the single largest gap in the whole record still stated (19.7%, mid-capitalization) — because a pooled worst case imputes the widest class's extreme to every asset. Prior published figures remain derivable from the preserved record.
How to Read the Number
Wider gaps carry less conviction. The engine's bounded design dampens the assessment as the gap between SV and market price widens. Large SV–market gaps signal weaker structural conviction, not stronger — a bounded-output property of the methodology.
Market validation through price action. The intraday touches described under Validation are direct market confirmations of the fair-value assessment — they show the assessed level is one the market actually transacted at, not a theoretical estimate.
Built for professional timeframes. Structural Value is designed for professionals who assess value over weeks, months, and quarters — estate valuations, fund reporting, fiduciary due diligence, equitable distribution, and settlement analysis. It is not a trading signal.
Deterministic and auditable. The number is built to withstand professional scrutiny: a documented methodology, a complete audit trail, and a reproducible calculation path from input to output.
What You Receive
AIREPORT generates professional reports showing portfolio-level and per-asset structural valuations alongside integrity scores and structural-change events. Whether the report supports a settlement negotiation, a quarterly client review, or a fiduciary filing, the methodology is consistent and the audit trail is complete.
Contact Information
Questions about the Methodology may be directed to care@keepgood.co.