Opportunity Alignment Rules

This page documents the current rule-based methodology behind the Performance page's opportunity-alignment labels.

It is written for business review. The goal is to make clear which rules are current implementation facts and which rules may need business sign-off.

What The Model Is Trying To Answer

Opportunity alignment asks:

At the time Alpha traded, did the available dashboard context support or oppose the trade direction?

It does not answer:

whether execution was good
whether the trade made money
whether the trade should have been placed
whether Alpha had a different strategy intent

The current model is a deterministic rules engine, not a learned model.

Current Trade Intent Assumption

Until Alpha provides strategy intent, the model uses direction as a proxy:

Alpha direction	Current assumed strategy
`buy`	Long volatility
`sell`	Short volatility
anything else	Unknown

This is a major assumption. It can be wrong for hedges, rolls, spread legs, closing trades, directional calls/puts, and portfolio overlays.

Signals Currently Used

The scoring model uses these opportunity-context fields:

Signal	Field	Included in score?
DVol percentile	`dvol_percentile`	Yes
VRP	`vrp`	Yes
Gamma regime	`gamma_regime`	Yes
Stationarity	`stationarity_status`	Yes
Smile richness	`smile_rich_cheap`	Yes when sourced; currently not sourced for Alpha proxy rows
Term structure	`term_structure_state`	Yes

Important boundary:

BTC proxy liquidity is not part of the opportunity score.

Liquidity is displayed as context coverage/backdrop only. Execution-quality liquidity needs MSTR/COIN traded-contract bid/ask, spread, OI, volume, and quote age at execution time.

Score Construction

Each row starts at 50 when:

trade direction is buy or sell
at least one scoring signal is observed

The model then adds drivers and detractors.

If context is unavailable, unmapped, or direction is unknown, the score is null and the label is insufficient_data.

Headline Labels

Score range	Label	Performance grouping
`70-100`	`aligned`	Supportive
`55-69`	`partially_aligned`	Supportive
`45-54`	`mixed`	Mixed/no-edge
`30-44`	`weak_alignment`	Opposing
`0-29`	`misaligned`	Opposing
null	`insufficient_data`	Not scored

Performance page headline percentage:

supportive among decisive rows = supportive / (supportive + opposing)

Mixed/no-edge rows remain visible but are excluded from the decisive denominator.

Current 365d production read:

supportive = 322 rows
mixed      = 225 rows
opposing   = 202 rows
decisive   = 524 rows

headline = 322 / 524 = 61.5%

This is not a performance grade. It is a strategy-read percentage.

Signal Point Rules

DVol Percentile

Trade assumption	Condition	Impact	Reason
Long-vol buy	DVol percentile `<= 25`	`+20`	Implied vol is cheap versus its own recent range.
Long-vol buy	DVol percentile `>= 75`	`-20`	Implied vol is rich, weakening a long-vol entry.
Short-vol sell	DVol percentile `>= 75`	`+20`	Implied vol is rich, supporting short-vol.
Short-vol sell	DVol percentile `<= 25`	`-20`	Implied vol is cheap, weakening short-vol.

VRP

Trade assumption	Condition	Impact	Reason
Long-vol buy	VRP `<= 0`	`+15`	Implied vol is low versus realized vol.
Long-vol buy	VRP `>= 5`	`-15`	Implied vol is rich versus realized vol.
Short-vol sell	VRP `>= 5`	`+15`	Positive/rich VRP supports selling vol.
Short-vol sell	VRP `<= 0`	`-15`	Low/negative VRP weakens selling vol.

Smile Rich/Cheap

Trade assumption	Condition	Impact	Reason
Long-vol buy	Smile `cheap`	`+15`	Buying optionality looks more attractive.
Long-vol buy	Smile `rich`	`-15`	Buying optionality looks less attractive.
Short-vol sell	Smile `rich`	`+15`	Selling rich optionality looks more attractive.
Short-vol sell	Smile `cheap`	`-15`	Selling cheap optionality looks less attractive.

Current Alpha boundary:

Smile is intentionally not sourced for MSTR/COIN Alpha proxy rows.

BTC smile may be useful market context, but it is not a clean trade-level smile proxy for MSTR/COIN options.

Gamma Regime

Trade assumption	Condition	Impact	Reason
Long-vol buy	`short_gamma`	`+5`	Short-gamma regimes can amplify moves.
Long-vol buy	`long_gamma`	`-5`	Long-gamma regimes can damp moves.
Short-vol sell	`long_gamma`	`+5`	Long-gamma regimes can support vol selling.
Short-vol sell	`short_gamma`	`-5`	Short-gamma regimes can hurt short-vol exposure.

Stationarity

Condition	Impact	Reason
`stable`	`+5`	Historical gamma-regime behavior is behaving enough like the current sample to trust more.
`unstable`	`-5`	Historical regime relationships are less reliable.
`insufficient`	no point impact	Not enough evidence to add or subtract points; confidence is still reduced.

Term Structure

Trade assumption	Condition	Impact	Reason
Long-vol buy	`front_rich` or `backwardated`	`+5`	Near-term risk concentration can support long-vol exposure.
Short-vol sell	`front_cheap` or `contango`	`+5`	Benign front-end structure can support short-vol carry.

Current implementation note:

Term structure currently adds supportive points in favorable cases, but it does not subtract points in the opposite cases.

This asymmetry should be reviewed with the business.

Confidence Penalties

The alignment score and confidence are separate.

The score says whether observed signals supported the assumed trade direction.
Confidence says how much trust to place in that score.

Confidence starts at 1.0 and is multiplied down.

Context Quality

Context quality	Penalty
`unavailable` or `unmapped`	`x0.15`
`stale` with no observed signals	`x0.25`
`proxy_mapped`	`x0.90`

Proxy Mapping

Mapping quality	Penalty
`proxy_mapped`	multiply by mapping confidence
`unmapped`	`x0.10`

Current proxy mapping:

Underlying	Market proxy	Mapping confidence
`MSTR`	`BTC`	about `0.80`
`COIN`	`BTC`	about `0.75`

Signal Coverage

The model expects six scoring signals:

dvol_percentile
vrp
gamma_regime
stationarity_status
smile_rich_cheap
term_structure_state

Coverage penalty:

confidence *= 0.4 + 0.6 * (observed_signals / 6)

Stationarity Confidence

Stationarity state	Penalty
`unstable`	`x0.70`
`insufficient`	`x0.80`

Current 365d production average confidence:

48.4%

Business Sign-Off Questions

Should buys always be treated as long-vol and sells as short-vol until strategy intent exists?
Are the DVol thresholds of 25 and 75 the right cheap/rich cutoffs?
Is VRP <= 0 the right long-vol support threshold?
Is VRP >= 5 the right short-vol support threshold?
Should gamma regime only carry +/-5, or should it be more important?
Should stationarity add/subtract points, reduce confidence only, or both?
Should term structure have negative points in unfavorable cases?
Should mixed rows be excluded from the headline denominator, or should the headline show total supportive share?
Should proxy-mapped MSTR/COIN context reduce confidence by the current mapping confidence?
Should these thresholds become configurable instead of hard-coded?

Recommended Demo Explanation

Use this wording when walking the Performance page:

The opportunity number is not a P&L score and not a trading grade. Each Alpha row is classified from the context we had at entry. Aligned and partially aligned rows count as supportive; weak and misaligned rows count as opposing; mixed rows stay visible but do not enter the decisive denominator. Current production is 322 supportive, 225 mixed, and 202 opposing rows, so the headline is 322 divided by 524 decisive rows, or 61.5%. Because the mixed bucket is still material and confidence is below 50%, the right read is not "good" or "bad"; it is "segment before judging."