A trading endpoint becomes more useful when you can ask the same question about an earlier session and inspect the evidence behind its interpretation. The historical trader-context additions cover expiry removal, strike dependency and income diagnostics, with explicit timestamps, source status and strategy-family research references.
How much concentration belongs to the next expiry?
/v1/exposure/expiry-impact/SPY?at=2026-10-12T15:55:00 compares current GEX with the same snapshot excluding one expiry. It reports current, excluded and remaining totals, removed gross share, sign changes and the largest strike reductions. The new gamma-context routes support SPY and QQQ.
Remaining GEX is current minus excluded GEX. Spot, Greeks and other maturities stay fixed. This helps identify whether a concentration depends on one maturity, but it is not a forecast of tomorrow's dealer hedging, prices or post-expiry returns. Calls-positive/puts-negative is a sign convention, not measured dealer inventory.
Does a level survive beyond one maturity?
/v1/exposure/strike-dependency/{symbol}?strike={strike}&at=... breaks an exact observed strike into expiry contributions. A dominant-expiry share describes concentration at that strike. It does not establish a support/resistance probability. The term-structure response also carries a gamma_concentration block for SPY/QQQ.
What does an income candidate actually pay?
/v1/strategies/yield-enhancement/SPY?at=2026-10-12T15:55:00&structure=cash_secured_put&feesPerContract=1 replays candidate selection and cash diagnostics. For one assumed standard put contract, gross credit is quoted bid × 100 and pre-credit collateral is strike × 100. Net credit subtracts the supplied opening fee. Covered-call returns require your stock cost basis.
The legacy selector ranks using mid; the new diagnostics show bid-based cash arithmetic. Missing fees leave net fields unknown. Absolute delta is a proxy, not a calibrated assignment probability. Annualized premium yield extrapolates one period and does not include all stock downside, repeated entry opportunities or the realized cost of assignment.
Contract adjustment status is also a limitation. An empty archived option root and a multiplier of 100 do not prove that a contract is an unadjusted standard series; a special distribution can adjust strikes while retaining a 100-share deliverable. Income replay can select such a series, so the assumed standard-contract arithmetic does not certify its deliverable.
SPY/QQQ have no company earnings event to exclude. For other equities, an unknown historical earnings calendar causes an insufficient-data decision when the default exclusion is requested. Historical earnings reconstruction is outside this addition.
Which evidence can a historical decision use?
The income response's metrics.research links versioned packs. A pack must have publication, definition-freeze and data-cutoff timestamps no later than the replay instant before its ID can attach to that response. A study published later is discoverable through an explicitly separate retrospective catalog link, not silently imported into the earlier decision.
The first pack's published_at_utc is October 16, 2026, when the frozen artifact was built; it is not an API deployment timestamp. Earlier replays therefore return no_pack_available_as_of_request and an explicit retrospective catalog link. Eligibility uses the recorded timestamps and does not prove that the API or pack was publicly served at that earlier instant.
The first SPY/QQQ income pack includes modeled returns, benchmarks, daily drawdown, ledgers, cost sensitivity and reproduction inputs. Its fixed monthly 28–35 DTE strategy differs from the endpoint's default full-chain target-delta selection. It establishes neither that selection policy's exact historical performance nor predictive usefulness for gamma signals.
Check the source before using the number
Read quote timestamps, calculation basis, OI vintage status and unavailable reasons. The SPY/QQQ raw snapshot paths bypass daily option materialized views; the daily reconstructed series remain disabled until unhealthy sources are repaired and verified. A cached number cannot establish source health.
Existing persisted-volatility references must precede the replay date and must have been created by the replay instant. That excludes later-created backfills, while unversioned in-place revisions remain a limitation. Historical timestamp filtering alone does not prove original data availability.
Build a reviewable experiment
- Fix the timestamp, symbol and strategy parameters; retain the complete response and data-quality fields.
- Separate snapshot arithmetic from a testable future-outcome hypothesis.
- Define entries, exits, benchmark, costs, collateral and missing-data handling before evaluating results.
- Keep later retrospective research separate from evidence available to the historical decision.
Read the route and field documentation and research-pack availability and downloads. The additions require the corresponding Historical API release; local validation does not establish live deployment.
Historical replay makes an endpoint's calculations inspectable. A research pack adds an auditable experiment around a defined strategy. Trading usefulness still depends on source quality, execution assumptions and evidence that survives a separate validation period.
