Developers / Field registry

Field registry

Every field, FRED series, spreadsheet mnemonic and Excel function the API serves, generated from the code that serves it. Each row names its source, the doors that carry it and a live count of its universe. The same list is available as GET /v1/fields and as the MCP tool field_search.

74 matches in Caveat. 12 fields whose source terms do not permit redistribution are not listed.

Field	Description	Source	Doors
`btc_netflow_artifact` Caveat · caveat	BTC exchange netflow computed from these rows is an artifact of exchange-internal (self) transfers, so genuine deposit and withdrawal flow appears only in the ETH and stablecoin rows.	Kresmion	meta.caveats on /v1 responses
`calibration_historical_not_forecast` Caveat · caveat	Calibration, Brier, and accuracy figures describe past resolved outcomes and are not a forecast of future results.	Kresmion	meta.caveats on /v1 responses
`central_bank_bis_lag` Caveat · caveat	BIS publishes policy rates with a lag of several days, so a row whose rate_source is fred or probe carries the level in force from a faster official source while superseded_rate_pct carries what BIS still says.	Kresmion	meta.caveats on /v1 responses
`central_bank_cycle_window_cumulative` Caveat · caveat	A hiking or cutting stance is read off the CUMULATIVE change across the named window, so a bank that hiked and then cut back to where it started reads as unchanged over that window.	Kresmion	meta.caveats on /v1 responses
`central_bank_level_not_flow` Caveat · caveat	A policy rate is a level rather than a flow, so it stays true until the bank changes it, which is why a row whose source has stopped publishing is withheld rather than served.	Kresmion	meta.caveats on /v1 responses
`central_bank_month_precision` Caveat · caveat	A rate change marked with month precision is known only to the month it ended in, because the long history behind it is a month-end series.	Kresmion	meta.caveats on /v1 responses
`central_bank_moves_daily_window_only` Caveat · caveat	The move log is day-precise only inside the trailing daily window; outside it there is no daily series to date a change against and only month-end precision exists.	Kresmion	meta.caveats on /v1 responses
`confluence-family-coverage` Caveat · caveat	Signal families cover different symbols, so the absence of a family leg for a symbol reflects that family's coverage and not disagreement.	Kresmion	meta.caveats on /v1 responses
`confluence-not-additive-proof` Caveat · caveat	Independent signal families agreeing on one symbol is broader context and not a prediction, and every contributing leg is listed so the independence of the sources can be judged directly.	Kresmion	meta.caveats on /v1 responses
`cot_lag` Caveat · caveat	CFTC Commitments of Traders data is weekly and released on roughly a three day lag from its Tuesday reporting date.	Kresmion	meta.caveats on /v1 responses
`econ_attribution_travels_with_the_data` Caveat · caveat	Each row's attribution line is required by the data owner's licence (World Bank CC BY 4.0; the OECD citation) and must stay with the data, including when you pass it on to others.	Kresmion	meta.caveats on /v1 responses
`econ_mixed_periods` Caveat · caveat	Economies publish on different dates and at different frequencies (World Bank annual, OECD monthly or quarterly), so every row names its own period and rows do not share one date.	Kresmion	meta.caveats on /v1 responses
`econ_values_as_published` Caveat · caveat	Values are shown as the World Bank or the OECD published them. Official statistics are revised: Kresmion keeps every value it has seen and serves the newest.	Kresmion	meta.caveats on /v1 responses
`econ_yoy_is_a_kresmion_calculation` Caveat · caveat	The yoy field is a Kresmion calculation from the published values: the % change against the same period a year earlier for a level series, the change in percentage points for a series that is already a rate or a share of GDP (see yoy_unit).	Kresmion	meta.caveats on /v1 responses
`end_date_scheduled_not_actual` Caveat · caveat	A prediction market end_date is the scheduled close and can differ from the actual resolution date.	Kresmion	meta.caveats on /v1 responses
`esi_event_count_is_the_sample` Caveat · caveat	event_count (n on a history point) is how many scored headline releases fall inside the trailing 90-day window behind that value, not the releases of that single day; every day carries the trailing-window value, so consecutive points share most of their releases. A small count (thin: true) is a thin reading rather than a quiet economy.	Kresmion	meta.caveats on /v1 responses
`esi_is_derived_not_the_calendar` Caveat · caveat	This is a weighted aggregate of how releases landed against consensus and is not the underlying event calendar; the per-event rows, their scheduled times and their consensus figures are not served through this API.	Kresmion	meta.caveats on /v1 responses
`esi_surprise_is_relative_to_forecast` Caveat · caveat	Each release's surprise (actual minus consensus, in the indicator's own unit) is standardised against that indicator's own prior surprises (at most 40 earlier releases; an indicator with fewer than 8 is not scored), clipped at plus or minus four, weighted by importance (high 1.0, medium 0.5, low 0.2) and averaged over a trailing 90-day window. Inflation releases are kept out of the headline. The index measures direction and magnitude against expectations, not the level of the economy.	Kresmion	meta.caveats on /v1 responses
`fedpath_ladder_is_the_deepest_event` Caveat · caveat	The ladder is the one Polymarket event whose outcome markets carry the most volume; every other market naming the same meeting is a side market, listed with its share of the ladder's volume.	Kresmion	meta.caveats on /v1 responses
`fedpath_prices_not_a_forecast` Caveat · caveat	Leg prices are the last traded prices of event contracts on Polymarket, in percent, reported as prices. They are not a forecast by Kresmion and the legs of one meeting need not sum to 100.	Kresmion	meta.caveats on /v1 responses
`fedpath_single_venue` Caveat · caveat	One venue only: Polymarket. No other exchange's prices and no economic calendar rows are included; the event log is Federal Reserve releases and FOMC decision days.	Kresmion	meta.caveats on /v1 responses
`fred_curated_series_only` Caveat · caveat	Only the series Kresmion itself fetches are served, because FRED terms prohibit an application that replicates the essential user experience of the FRED API, so this is never an arbitrary-id pass-through.	Kresmion	meta.caveats on /v1 responses
`fred_observation_date_not_fetch_date` Caveat · caveat	The date on an observation is the period the number describes and never the day it was fetched, which is a different fact carried by as_of.	Kresmion	meta.caveats on /v1 responses
`gamma_modeled_not_measured` Caveat · caveat	Options gamma exposure and risk-neutral implied distributions are modeled from the options chain under pricing assumptions and are not directly measured dealer positioning.	Kresmion	meta.caveats on /v1 responses
`heuristic_labels_do_not_suppress` Caveat · caveat	A wallet label inferred from sweep behaviour alone can identify a counterparty but cannot prove who owns the address, so a row whose two sides collapse to one entity on heuristic evidence only is reported as flow and marked `internal_transfer_unverified` rather than removed. Read `label_confidence` for the weakest tier used.	Kresmion	meta.caveats on /v1 responses
`internal_transfers_labelled` Caveat · caveat	Rows where one owner moved value between its own wallets are labelled rather than hidden: read `is_internal_transfer` and exclude those rows before computing any flow total, or the same exchange shuffling its own balances will read as deposits.	Kresmion	meta.caveats on /v1 responses
`liquidations_single_venue_okx` Caveat · caveat	Realized liquidation events are collected from a single venue (OKX) over a starter symbol set and are not aggregated across all exchanges.	Kresmion	meta.caveats on /v1 responses
`liquidity_credit_legs_licence_gated` Caveat · caveat	The high-yield and investment-grade option-adjusted spread legs are ICE BofA index series whose FRED copyright tier requires pre-approval, so they obey FRED_RESTRICTED_POLICY and come back null when the platform is set to block them.	Kresmion	meta.caveats on /v1 responses
`liquidity_net_is_derived` Caveat · caveat	Net liquidity is computed as the Fed balance sheet less the overnight reverse repo facility less the Treasury General Account, which is a convention rather than a published series.	Kresmion	meta.caveats on /v1 responses
`liquidity_score_renormalises` Caveat · caveat	The liquidity regime score drops any input that could not be read and renormalises the remaining weights, so inputs_missing names exactly which legs the published number was computed without.	Kresmion	meta.caveats on /v1 responses
`mm_administered_effective_dating` Caveat · caveat	An administered rate is a level that stays true until the Fed changes it, and FRED dates it by its EFFECTIVE day, so it runs ahead of the market rates. Every row here is read at `as_of`, which is the newest date a market rate carries.	Kresmion	meta.caveats on /v1 responses
`mm_bills_discount_basis` Caveat · caveat	The Treasury bill rows are quoted on a DISCOUNT basis, which is systematically below the investment-basis constant-maturity yield for the same bill. Do not subtract one from a policy rate without converting the basis.	Kresmion	meta.caveats on /v1 responses
`mm_corridor_position_unclamped` Caveat · caveat	corridor_position_pct is not clamped: a rate above the ceiling reports above 100 and a rate below the floor reports below 0, because that is the thing worth seeing.	Kresmion	meta.caveats on /v1 responses
`mm_spread_needs_both_legs` Caveat · caveat	A spread point exists only where both legs carry an observation on the same day, so a gap in the series is a day one leg did not print rather than a day the spread was zero.	Kresmion	meta.caveats on /v1 responses
`mm_stress_is_width_not_level` Caveat · caveat	The stress read is the WIDTH of the 1st to 99th percentile distribution behind a print, not the level of the rate, and it exists only for the overnight rates the New York Fed calculates from transactions.	Kresmion	meta.caveats on /v1 responses
`mm_volume_nyfed_only` Caveat · caveat	volume_bn and percentile_width_bps exist only for the five overnight rates the New York Fed calculates from transactions. They are null for every administered rate and every bill.	Kresmion	meta.caveats on /v1 responses
`no-advice` Caveat · caveat	This is a descriptive record of how past signals were followed by price movement and is not advice or a recommendation to trade.	Kresmion	meta.caveats on /v1 responses
`offerings_null_is_not_zero` Caveat · caveat	Every financial field is nullable and a null is a figure the document did not state rather than a zero; parse_confidence reports how sure the classifier was about the row it did read.	Kresmion	meta.caveats on /v1 responses
`offerings_one_row_one_filing` Caveat · caveat	One row is one FILING rather than one deal, so a single IPO appears as its registration, its amendments and its priced prospectus, tied together by ipo_group_key.	Kresmion	meta.caveats on /v1 responses
`offerings_range_not_midpoint` Caveat · caveat	A price range stated by an S-1 or S-1/A is reported as a range and is never collapsed to its midpoint, and a deal still in registration carries no priced figure at all.	Kresmion	meta.caveats on /v1 responses
`orderbook_top150_15min` Caveat · caveat	Prediction-market order-book snapshots cover roughly the top 150 markets by 24h volume plus the pinned 2026 midterm markets, sampled every 15 minutes, so other smaller markets and sub-15-minute moves are not captured.	Kresmion	meta.caveats on /v1 responses
`pm-options-descriptive` Caveat · caveat	The pm-vs-options spread is descriptive market data comparing two venues under stated pricing assumptions and does not contain buy or sell recommendations.	Kresmion	meta.caveats on /v1 responses
`pm-options-expiry-gap` Caveat · caveat	The Deribit option expiry used can differ from the prediction market close date by up to seven days, and the exact signed gap is reported per row as expiry_gap_days.	Kresmion	meta.caveats on /v1 responses
`pm-options-thin-books` Caveat · caveat	A prediction-market versus options spread is information about two separate order books and is not an executable arbitrage, and Polymarket books can be thin, so the execution-cost surface reports the cost of trading a given size.	Kresmion	meta.caveats on /v1 responses
`pm-options-touch-approx` Caveat · caveat	Touch-market implied probabilities use the exact first-passage probability under GBM with per-strike implied vol (driftless forward). Model assumptions still apply to touch rows: constant per-strike vol, no jumps, no vol term structure.	Kresmion	meta.caveats on /v1 responses
`rv_adr_multiples_refused` Caveat · caveat	Valuation multiples use the latest close and the share count the filer reported. A depositary filer (20-F, 40-F) reports an ordinary share count while its quote is an ADR price at a ratio this platform does not store, so its market cap and every multiple derived from it are refused rather than estimated.	Kresmion	meta.caveats on /v1 responses
`rv_annual_filings_only` Caveat · caveat	Every fundamental is the full fiscal year reported on one annual filing, so nothing is trailing twelve months and nothing is annualised from a quarter; a null value always carries a refusal code and never means zero.	Kresmion	meta.caveats on /v1 responses
`rv_peer_level_sub_industry_then_sector` Caveat · caveat	Peers are the company's GICS sub-industry where it is classified, falling back to its sector; `meta.peer_level` says which, and a sector median is context rather than a comparison.	Kresmion	meta.caveats on /v1 responses
`rv_percentile_is_population_not_rating` Caveat · caveat	A percentile is where this company sits in a distribution of its peers. It is an observation about a population, not a rating, and no metric here is scored as good or bad.	Kresmion	meta.caveats on /v1 responses
`rv_price_return_split_refused` Caveat · caveat	A price return is refused outright when the window contains a consecutive-close jump of `meta.rules.split_jump_ratio` or more: the stored price series is unadjusted and is never restated after a corporate action, so a split there is a raw discontinuity, not a return.	Kresmion	meta.caveats on /v1 responses
`short_pct_float_is_a_lower_bound` Caveat · caveat	Short interest as a percentage of float divides by a float derived by subtracting Section 16 affiliate holdings and affiliate 13D/G blocks from shares outstanding, and because any affiliate not yet observed stays in the denominator, the published percentage is a lower bound rather than a point estimate.	Kresmion	meta.caveats on /v1 responses
`short_pct_float_partial_coverage` Caveat · caveat	A float percentage is published only for the minority of shorted tickers whose Section 16 insider holdings are known, because many issuers have not filed since collection began, and the remaining names carry no float percentage rather than an estimated one; the measured coverage for this response is in meta.coverage.	Kresmion	meta.caveats on /v1 responses
`short_pct_shares_outstanding_is_not_float` Caveat · caveat	Short interest as a percentage of shares outstanding is a different measure from percentage of float, is always the smaller of the two because float is a subset of shares outstanding, and is published wherever a share count is available.	Kresmion	meta.caveats on /v1 responses
`short_volume_not_short_interest` Caveat · caveat	This is FINRA daily short-sale volume, a flow, and is not the bi-monthly exchange short interest, a stock; the two are different measures and are not interchangeable.	Kresmion	meta.caveats on /v1 responses
`short_volume_off_exchange_hedging` Caveat · caveat	FINRA short volume counts only off-exchange trades reported to a FINRA facility (TRF/ADF), it includes market-maker hedging, and it is not a measure of directional positioning.	Kresmion	meta.caveats on /v1 responses
`smart-money-descriptive` Caveat · caveat	This score is a deterministic summary of public insider, congressional, institutional, and short-volume filings and is descriptive positioning context, not advice.	Kresmion	meta.caveats on /v1 responses
`smart-money-lag` Caveat · caveat	The 13F institutional leg reports quarterly on roughly a 45 day lag and congressional disclosures lag the trade by up to 45 days, so each leg carries its own as_of timestamp for the date its evidence reflects.	Kresmion	meta.caveats on /v1 responses
`smart-money-short-coverage` Caveat · caveat	The FINRA short-volume leg covers only about 29 symbols and can be several days stale, so when it is absent its weight is redistributed proportionally across the legs that are present.	Kresmion	meta.caveats on /v1 responses
`sovereign_spread_ecb_free_of_charge` Caveat · caveat	The ECB yields behind the euro-area rows are available free of charge at data.ecb.europa.eu, and the spreads are modified data computed by Kresmion from the yields as published.	Kresmion	meta.caveats on /v1 responses
`sovereign_spread_is_a_calculation_not_a_rating` Caveat · caveat	A spread here is a Kresmion calculation, one published 10-year government bond yield minus another, and is market-implied context, not a credit rating: no rating agency is involved.	Kresmion	meta.caveats on /v1 responses
`sovereign_spread_local_currency` Caveat · caveat	Outside the euro area the bonds are in the local currency, so a spread also carries currency and inflation expectations, not only credit.	Kresmion	meta.caveats on /v1 responses
`sovereign_spread_monthly_both_legs` Caveat · caveat	Every spread is monthly: both legs are averages for the same calendar month (the US leg is the mean of the daily H.15 DGS10 prints, and China's yield the mean of the Ministry of Finance curve's daily 10-year points, over complete months), so no daily spread is served. Where a row's y10_leg.construction is set, that 10-year yield is not an average of benchmark bond yields (Colombia: a fitted zero-coupon rate; China: an interpolated curve point).	Kresmion	meta.caveats on /v1 responses
`supply_customs_value_not_a_price` Caveat · caveat	A corridor figure is the customs value the reporting country declared for one HS basket in one month, in US dollars, never a price. Values published in euros or yen are converted at one published ECB monthly average rate, and a quantity divided into a value is a declared unit value, not a market price.	Kresmion	meta.caveats on /v1 responses
`supply_licence_withheld` Caveat · caveat	A corridor whose source licence is not confirmed for reuse is listed with its status and no figures (latest is null and withheld_reason says why); absent figures are never zero. Figures are served to Developer plan keys only: a free-plan key receives every corridor in the same shape, with the plan named in withheld_reason.	Kresmion	meta.caveats on /v1 responses
`supply_mirror_gap_expected` Caveat · caveat	The same trade seen from the two sides (an export corridor and its mirror import corridor) differs by valuation basis (FOB against CIF), timing and origin attribution; the gap is expected and is not an error in either series.	Kresmion	meta.caveats on /v1 responses
`supply_mom_masked_jan_mar` Caveat · caveat	Month-on-month changes for January to March carry mom_masked true because the Lunar New Year moves trade between those months; read yoy_pct and yoy_3m_pct for those periods.	Kresmion	meta.caveats on /v1 responses
`supply_z36_on_levels` Caveat · caveat	z36 scores the latest month against the trailing 36 months, excluding the latest, on the value level of a trending series, so high readings are far more common than low ones. rank_same_month is the rank of this value among every stored month with the same calendar month (1 is the highest).	Kresmion	meta.caveats on /v1 responses
`thirteenf_ticker_resolution` Caveat · caveat	13F holdings resolve a ticker for only about 50% to 93% of rows because no free CUSIP-to-ticker map is complete, so aggregate by CUSIP rather than ticker to avoid undercounting a position.	Kresmion	meta.caveats on /v1 responses
`tic_lag` Caveat · caveat	Treasury TIC data is published on roughly a six week lag, so the most recent reported month reflects cross-border holdings from about six weeks earlier.	Kresmion	meta.caveats on /v1 responses
`track-record-in-sample` Caveat · caveat	Detector thresholds evolved over time, so these statistics are the live production history of the signals as they actually fired and are not a frozen out of sample backtest.	Kresmion	meta.caveats on /v1 responses
`track-record-price-coverage` Caveat · caveat	Forward returns use daily closing prices where the equity price feed can lag by a session or two and the crypto price feed is current, so the most recent equity signals may not yet have a completed 30 day horizon.	Kresmion	meta.caveats on /v1 responses
`track-record-young` Caveat · caveat	Most Kresmion detectors came online in April and May 2026, so the occurrence count for many signal types is small and these figures are descriptive of the live history rather than predictive.	Kresmion	meta.caveats on /v1 responses
`unlocks_hand_curated` Caveat · caveat	Token unlock schedules are hand-curated from public vesting disclosures and are not an exhaustive or automatically verified feed.	Kresmion	meta.caveats on /v1 responses
`yield_curve_partial_is_partial` Caveat · caveat	A partial curve is served as partial: the eleven tenors do not all exist for all time, nothing is interpolated, and no tenor is carried forward from a neighbouring day.	Kresmion	meta.caveats on /v1 responses

Universe counts are computed against the table each field is read from and refresh every 15 minutes. Bulk row counts are the database planner's estimate. Metadata only; information, not investment advice.