Trade Performance Data Gaps

This note explains the current data gap behind Alpha execution-quality and forward-outcome metrics. It is written for operators reviewing the Performance page after Alpha trades have synced.

Current State

The Alpha feed is synced and the Performance page can score whether Alpha trades were directionally consistent with the dashboard's available market context at execution time. That does not mean every context group is complete. Some groups are available, some are partial, some are unavailable, and some are deliberately not sourced.

What remains limited is instrument-level valuation data for the actual traded MSTR and COIN options:

• Execution benchmarks need sourced bid/ask/mid quotes near the trade timestamp.
• Execution-quality liquidity needs traded-contract OI, volume, spread, and quote-age inputs. BTC proxy liquidity is market backdrop only.
• Forward outcomes need sourced option marks at 1d, 7d, and 30d horizons. If the requested horizon falls after option expiry, the backfill can attempt a terminal option quote near expiry instead of looking for a non-existent live quote at the later horizon.
• Realized P&L needs an authoritative accounting, settlement, or validated lifecycle source.

Unavailable economics must stay blank or labelled unavailable. They must not display as zero.

The source-authority policy for direct, proxy, estimated, stale, pending, unavailable, unsupported, and not-entitled states is maintained in the Source Authority Register.

The derivative-equivalent identity policy for exact listed, adjusted, fallback, SMA/OTC mapped, unsupported, and unmapped contracts is maintained in Derivative Identity And Proxy Mapping.

The preparatory plan for converting these gaps into reliable source diagnostics is maintained in Data Ingestion Readiness Plan. Future SMA-specific source fields and mapping questions are maintained in SMA Ingestion Requirements.

Current Production Snapshot

As of the 2026-05-10 production API check for source=alpha_sync, lookback_days=365, and limit=1000:

The adjusted-contract fallback is intentionally source-labelled as:

amberdata_tradfi_level1_standard_root_fallback

That label means the row used standard MSTR/COIN quote lookup for an adjusted Alpha root such as 2MSTR or 2COIN. It is an explicit temporary analysis decision, not proof that the adjusted contract deliverable is identical to the standard contract.

Expiry-aware outcome marks are separately source-labelled:

amberdata_tradfi_level1_expiry_quote
amberdata_tradfi_level1_standard_root_expiry_quote_fallback

These labels mean the requested outcome horizon was after option expiry, so the backfill tried to source a terminal option quote near the expiry timestamp. They are not modelled intrinsic value and they are not authoritative realized P&L. If no expiry quote is available, the row remains unavailable with expiry_mark in missing_fields.

Performance Backfill And Sourcing Plan

This table separates what already exists as dashboard context, what can be backfilled from stored history, and what still needs a real source for the traded MSTR/COIN contracts.

Outstanding Data Issues

These are the data issues that remain after the current backfills and adjusted-standard fallback.

Backfill Strategy

1. Keep BTC proxy context for opportunity analysis as-is.
2. Backfill trade_market_context from stored snapshot history for all Alpha rows.
3. Source and backfill execution benchmark rows for MSTR/COIN quotes where coverage exists.
4. Source and backfill forward outcome marks for 1d, 7d, and 30d horizons.
5. Keep realized P&L and trade-level smile richness explicitly unavailable until the product decides to source them.
6. Expose quality states on every aggregate so missing data cannot be mistaken for zero.

Execution Benchmarks

Execution quality answers whether a trade filled well versus the market for the exact traded option. For the current Alpha scope, the benchmark source must be MSTR or COIN TradFi option data, not BTC option data.

For each execution, the dashboard needs a quote matching:

• the same underlying, currently MSTR or COIN
• the same expiry
• the same strike
• the same call/put side
• a timestamp close to the Alpha execution time
• a freshness window tight enough to trust

When a fresh quote exists:

If no fresh sourced bid/ask/mid exists, execution quality is unavailable.

That is not the same as zero slippage:

• 0 slippage means the fill exactly matched mid.
• unavailable means there is no defensible mid quote.

The dashboard must not infer good or bad execution from unavailable quote data.

Execution-Quality Liquidity

The current Performance source-group table may show BTC Proxy Liquidity. That is not the same as Alpha execution liquidity.

BTC proxy liquidity can describe broad crypto-vol market backdrop. It should not be used to judge whether a specific MSTR/COIN option trade was easy to execute.

To score execution-quality liquidity, the dashboard needs MSTR/COIN traded-contract data near the Alpha execution timestamp:

• contract open interest
• contract volume
• bid
• ask
• mid and spread
• quote timestamp / quote age
• contract identifier match to the Alpha row

Current implementation stores traded-contract bid/ask, mid/spread, quote age, and bid/ask quote size when Amberdata returns a fresh MSTR/COIN level-1 quote. Sampled MSTR and COIN level-1 quote responses did not include 24h contract volume or open interest, so those fields remain null unless the provider returns explicit volume24h/contractVolume24h and openInterest-style fields. Missing volume/OI must not be shown as zero.

Execution-liquidity quality states:

Forward Outcome Marks

Forward outcomes answer what happened after the trade at a defined horizon. For each trade, the dashboard looks for a sourced mark for the same MSTR or COIN option around:

• trade_time + 1 day
• trade_time + 7 days
• trade_time + 30 days

If the requested horizon is after option expiry, the backfill attempts a terminal option quote around the expiry timestamp. A successful terminal quote is persisted with outcome_state = terminal_mark and an expiry quote source label. A missing terminal quote remains unavailable; the dashboard does not substitute intrinsic/model value as realized P&L.

When a fresh sourced mark exists, forward P&L can be computed:

When no fresh sourced mark exists, forward P&L remains null. When the target horizon is still in the future, the outcome is pending.

That is not the same as zero P&L:

• 0 P&L means the trade broke even.
• unavailable means the mark needed to compute P&L is missing.
• pending means the horizon has not matured.

Hit rates and average P&L must exclude unavailable and pending rows.

Why The Gap Exists

The likely causes are source and mapping constraints rather than dashboard calculation logic:

• Amberdata TradFi coverage does not return rows for every historical MSTR/COIN option contract. Direct probes for representative missing standard contracts returned HTTP 200 with empty data.
• Amberdata TradFi level-1 quote requests reject single windows wider than one hour, so broader nearest-session recovery uses bounded chunked polling or another endpoint/source.
• Adjusted Alpha roots such as 2MSTR and 2COIN are not the same contract identifier as standard MSTR or COIN. The current fallback is explicitly labelled and temporary.
• Alpha trade contract fields may not map exactly to every market-data provider's option identifier format.
• Some target timestamps fall outside regular market hours, on weekends, on holidays, or after expiry.
• Some horizons are legitimately pending because the target timestamp has not happened yet.
• Expiry or realized P&L needs settlement, accounting, or another authoritative source; the dashboard does not substitute intrinsic value as realized P&L.

How To Fix It

There are three defensible paths.

1. Fix source coverage. Confirm Amberdata entitlement and coverage for historical MSTR and COIN option level-1 quotes. If coverage exists, fix symbol or contract mapping and rerun backfills. If wider search is needed, enable chunked quote recovery within Amberdata's one-hour request limit. For expired-before-horizon rows, use the expiry-aware terminal quote path.

2. Use another authoritative source. Source execution quotes, forward marks, terminal marks, and realized outcomes from Alpha, broker/OMS, OPRA, OptionMetrics, OCC/settlement, accounting, or PMS data.

3. Add an explicitly estimated path. If the business accepts model marks, stale quotes, or nearest-session marks, label them estimated or stale. Do not populate bid/ask/mid from estimates, and do not call model-derived values realized P&L.

Where Operators Should Look

Trade Book is the source-row and sync control surface. It is where operators confirm that Alpha rows are loaded and reconciled.

Performance is the measurement surface. It is where operators check:

• opportunity alignment
• source-group quality
• execution benchmark availability
• forward outcome availability
• pending/unavailable states
• whether P&L is sourced enough to read

Backfill routes are admin-only production operations. They are not currently ordinary Trade Book buttons.

Status Check Command

Use this command before and after backfills:

curl -sS \
  'https://vol-backend-244916812493.us-east4.run.app/api/internal/trades/analytics/status?source=alpha_sync'

Key fields:

Backfill Commands

Set the admin secret without printing it:

ADMIN_SECRET="$(gcloud secrets versions access latest --secret=vol-admin-secret)"

Optional nearest-session recovery controls:

Quotes recovered outside the initial tight window are source-labelled as amberdata_tradfi_level1_chunked_recovery. Adjusted 2MSTR / 2COIN rows still require allow_adjusted_standard_fallback=true and remain source-labelled separately.

Execution benchmark dry run:

curl -sS -X POST \
  'https://vol-backend-244916812493.us-east4.run.app/api/internal/trades/execution-benchmarks/backfill' \
  -H "x-admin-secret: ${ADMIN_SECRET}" \
  -H 'Content-Type: application/json' \
  -d '{"source":"alpha_sync","limit":100,"dry_run":true,"window_minutes":30,"max_staleness_minutes":15,"allow_adjusted_standard_fallback":true,"quote_recovery_enabled":true,"quote_recovery_window_minutes":180,"quote_recovery_chunk_minutes":60,"quote_recovery_max_requests":8}'

Execution benchmark write:

curl -sS -X POST \
  'https://vol-backend-244916812493.us-east4.run.app/api/internal/trades/execution-benchmarks/backfill' \
  -H "x-admin-secret: ${ADMIN_SECRET}" \
  -H 'Content-Type: application/json' \
  -d '{"source":"alpha_sync","limit":100,"dry_run":false,"window_minutes":30,"max_staleness_minutes":15,"allow_adjusted_standard_fallback":true,"quote_recovery_enabled":true,"quote_recovery_window_minutes":180,"quote_recovery_chunk_minutes":60,"quote_recovery_max_requests":8}'

Forward outcome marks dry run:

curl -sS -X POST \
  'https://vol-backend-244916812493.us-east4.run.app/api/internal/trades/outcomes/backfill' \
  -H "x-admin-secret: ${ADMIN_SECRET}" \
  -H 'Content-Type: application/json' \
  -d '{"source":"alpha_sync","limit":100,"lookback_days":365,"horizons":[1,7,30],"dry_run":true,"quote_window_minutes":30,"max_mark_staleness_minutes":60,"allow_adjusted_standard_fallback":true,"quote_recovery_enabled":true,"quote_recovery_window_minutes":180,"quote_recovery_chunk_minutes":60,"quote_recovery_max_requests":8}'

Forward outcome marks write:

curl -sS -X POST \
  'https://vol-backend-244916812493.us-east4.run.app/api/internal/trades/outcomes/backfill' \
  -H "x-admin-secret: ${ADMIN_SECRET}" \
  -H 'Content-Type: application/json' \
  -d '{"source":"alpha_sync","limit":100,"lookback_days":365,"horizons":[1,7,30],"dry_run":false,"quote_window_minutes":30,"max_mark_staleness_minutes":60,"allow_adjusted_standard_fallback":true,"quote_recovery_enabled":true,"quote_recovery_window_minutes":180,"quote_recovery_chunk_minutes":60,"quote_recovery_max_requests":8}'

Run dry-runs first. Only persist when the output is understood. Persisted unavailable records are valid quality-state evidence; they should not be interpreted as failed trades.