Every dataset, organised by source.
48 datasets across 7 sources. Pick what you need — download a file, query with SQL, or wire up an AI agent.
HKJC Racing
17 datasetsHong Kong barrier trial results, times and running comments
Hong Kong barrier trial (practice race) results — every horse's run in each morning trial batch at Sha Tin, Conghua and Happy Valley. Includes jockey, trainer, draw, gear, finishing order, time, lengths behind winner (LBW), running positions and the stewards' note on each horse's run. Barrier trials are how unraced and returning horses qualify and show form before the racecard, so this is early-signal data for spotting fitness and intent that the main results don't yet reveal.
Stewards' running comments for every Hong Kong runner
Stewards' "comments on running" — a plain-language note on how each horse ran (start, position, trouble, finish), alongside placing, jockey, gear and the race's class/distance/going.
Hong Kong horse racing payouts and betting dividends
Historical dividend payouts for Hong Kong horse racing. One row per pool type per race, covering 34 different bet types including WIN, PLACE, QUINELLA, FORECAST, TIERCE, QUARTET, FIRST 4, and multi-race exotics. Contains 25 years of data (2001-present). Includes denormalized race attributes for easy filtering without JOINs.
Hong Kong racehorse profiles, pedigree and career statistics
Detailed profiles of every active and retired Hong Kong racehorse. Look up any horse's breeding (sire, dam, origin country), current HKJC rating, career win record, total prize money earned, and trainer/owner details. Track rating changes throughout the season and identify horses on the rise or decline. Updated daily with the latest statistics.
Hong Kong horse racing Six Up and Triple Trio jackpots
Jackpot pool information for multi-race pools like Six Up and Triple Trio. Track rolling jackpots and estimated dividends for snowball pools. These pools accumulate when not won, creating large prize pools. One row per jackpot pool per meeting per snapshot.
Hong Kong jockey standings, wins, and performance statistics
Comprehensive jockey rankings with performance statistics. Track which jockeys are riding in Hong Kong and compare their career records - wins, win rates, podium places, and current season form. Includes visiting international riders. One snapshot per race meeting day, captured at the close of the meeting. Snapshot history begins December 2025.
Hong Kong Mark Six lottery results and winning numbers history
Complete history of Hong Kong Mark Six lottery draws with winning numbers, extra ball, jackpot amounts, and prize payouts. Search past results, analyze number frequency patterns, and track snowball jackpots. Updated after every draw (Tuesday, Thursday, Saturday). Each draw includes all six winning numbers plus the extra number, total sales, and first division prize amounts.
Hong Kong horse racing live odds history and market movements
The Hong Kong Jockey Club's betting odds as they moved before each race: Win and Place for every runner, and the exotic pools (Quinella, Quinella Place, Forecast, Trio, First Four, Double, and HKJC's top-20 and banker lists for Tierce, Trio, First Four and Quartet). Track how prices changed from the first quote we captured to the close, see which runners attracted late money, and study market efficiency. One row per odds MOVEMENT per combination: each pool update HKJC publishes is stored as the full set of that pool's prices, timestamped with HKJC's own update time (snapshot_timestamp, UTC); repeated unchanged states are not stored, so hold a price flat until the next row. Our collector polls, so if HKJC moves a pool more than once between two polls only the later state is recorded. In the 30 minutes before each race it polled about every minute from 8 December 2021 to 27 September 2023, about every 5 minutes from 1 October 2023 to 28 January 2024, about every minute from 31 January 2024 to 14 July 2024, about every 5 minutes from 1 October 2024 to 27 May 2026 and about every minute since 28 May 2026; earlier in the day, about every 5 minutes until 27 May 2026 and about every minute since 28 May 2026. Post time here is the race card's where we hold it, and otherwise the last five-minute mark at least 30 seconds before betting closed. At 5 minutes before post time the median age of the latest Win price we hold is under a minute in every period; at 1 minute before post time it is under 1, 5, 1, 5 and 1 minutes in those five periods: from 1 October 2023 to 28 January 2024 and from 1 October 2024 to 27 May 2026 polls fell on the five-minute marks post times fall on, so the price 1 minute before post time is usually the one from 5 minutes before. Hong Kong meetings (Sha Tin and Happy Valley): Win and Place odds from 8 December 2021; Quinella, Quinella Place, Forecast, Trio, First Four and Double from 1 October 2024. 18 Hong Kong meetings in that span have no odds at all: 10 September 2023; 3, 6, 10 and 13 March 2024; 8, 11, 15, 18, 22, 25 and 28 September 2024; 3, 6 and 9 November 2024; and 7, 11 and 14 January 2026. 2 more Hong Kong meetings were captured too thinly to time anything before the off, with fewer than 10 Win states in at least half their races: 13 September 2023 and 29 December 2024. Collection stopped while betting was still open at Happy Valley races 3-10 on 26 December 2024 and Sha Tin races 2-10 on 8 June 2025, so those races hold no odds from the last minutes of betting. The final Win odds of every finisher at these 20 meetings and in these 17 races are in our race results dataset (column odds). Elsewhere, missed polls leave 102 other races whose latest Win price at 5 minutes before post time is more than 10 minutes old, 18 of them more than an hour old; the meetings with five or more such races are Happy Valley races 1, 2 and 4-8 on 4 January 2023; Sha Tin races 4-8 on 29 January 2023; Sha Tin races 2 and 7-10 on 19 February 2023; Sha Tin races 1-7 on 28 June 2025; Happy Valley races 1-6 and 8 on 28 January 2026; Happy Valley races 2-4, 6 and 7 on 18 March 2026; and Sha Tin races 2-7 on 29 March 2026. Tierce and Quartet: the Hong Kong Jockey Club publishes no complete odds matrix for either pool, only the 20 lowest-priced combinations (TCETop, QTTTop) and, for each runner taken as banker, the 10 lowest-priced combinations with that runner first (TCEBanker, QTTBanker); Trio and First Four carry the same two lists (TRITop, FFTop, TRIBanker, FFBanker) beside their complete matrices. The top-20 lists run from 1 October 2024 but were not collected for meetings from 1 January 2026 to 24 May 2026; the banker lists exist for 8 meetings between 28 September 2025 and 27 December 2025 and for every meeting from 27 May 2026. Overseas races simulcast in Hong Kong (venues S1-S5) are included from 29 September 2024. Delivered hourly with the rest of the pipeline, not streamed.
Hong Kong horse racing closing odds and market data
Closing odds for Hong Kong races (Sha Tin and Happy Valley): for every pool and combination, the last price HKJC published before the race's post time, taken from the same movement series as HKJC Odds Movements, so the two never disagree. Win and Place from 8 December 2021; the exotic pools (Quinella, Quinella Place, Forecast, Trio, First Four, Double, and the top-20 and banker lists) from 1 October 2024, with the same gaps as HKJC Odds Movements. Compare closing odds with results to backtest, value runners or study favourites. A race appears once it has started. A pool has closing odds only where its last quote before the off was published within 60 minutes of it: the Win pools of 32 races have none because we hold no quote from that last hour (Happy Valley races 6-7 on 4 January 2023; Sha Tin races 5-7 on 29 January 2023; Happy Valley races 1-4 on 13 September 2023; Sha Tin races 1-10 on 29 December 2024; Sha Tin races 1-7 on 28 June 2025; and Happy Valley races 1-6 on 28 January 2026), and every other pool that trades up to the off is held to the same rule; presales pools (*Pre) and the PWIN snapshots carry their last quote whatever its age. 18 Hong Kong meetings since 8 December 2021 have no closing odds at all: 10 September 2023; 3, 6, 10 and 13 March 2024; 8, 11, 15, 18, 22, 25 and 28 September 2024; 3, 6 and 9 November 2024; and 7, 11 and 14 January 2026. Post times before 2026 are mostly derived from pool closure rather than a published race card (see race_start_time).
Hong Kong horse racing betting pool investment data
Betting pool investment snapshots showing how pools accumulate before races. Time-series data for analyzing betting patterns and market movements. Covers all pool types including WIN, PLACE, QUINELLA, FORECAST, and exotics. One row per pool per race per snapshot time.
Hong Kong horse racing results and historical form guide
Complete historical record of every Hong Kong horse race from the Jockey Club. Track individual horse performances across thousands of races with detailed finishing positions, winning margins, race times, and running styles. Perfect for form analysis, identifying patterns, and researching historical performance trends. Includes 60+ data points per race entry covering horse details, jockey and trainer information, race conditions, and performance metrics. Download ready-to-use CSV and Parquet files for your own analysis.
Hong Kong horse racing results and form guide (core columns)
Hong Kong horse racing results and form guide — the Starter column set: identifiers, result, race specifications, connections, horse attributes, pedigree, performance flags, and the horse's career record. The premium race-grain analytics (in-running positions, sectional and final-800m 200m-split times, pace deltas, and the as-of-race rating) are in the Pro dataset, hkjc_race_results.
Hong Kong horse race specifications and historical archive
Complete catalog of every Hong Kong horse race from 1977 to present. Search and filter races by venue, distance, class, track conditions, and more. Each race includes specifications, video replay links, and pace analysis data. Perfect for historical research, race analysis, and identifying patterns by track conditions or distance. Contains more than 30,000 races across nearly 50 years of Hong Kong racing history.
Hong Kong race card scratches and jockey changes
Race card changes including scratches, jockey changes, gear changes, and barrier changes. Track late changes before race time for form analysis. Each change is timestamped showing when it was announced. One row per change event.
Hong Kong horse racing sectional times and pace analysis
Detailed sectional time data showing how each horse ran through every section of the race. Analyze running styles, identify front-runners vs closers, and study pace patterns. Each row contains times and positions at each 400m checkpoint plus derived indicators for quick filtering. Perfect for understanding how races unfolded and profiling horse running preferences.
Hong Kong racecourse penetrometer and hammer readings
Track condition readings (penetrometer and hammer) for Hong Kong racecourses. Use this to analyze ground conditions and their correlation with race outcomes. Penetrometer readings indicate turf softness; hammer readings measure firmness. One row per meeting date and venue (readings are taken once per meeting).
Hong Kong trainer standings, wins, and performance statistics
Comprehensive trainer rankings with performance statistics. Track which trainers are operating in Hong Kong and compare their career records - wins, win rates, podium places, and current season form. Includes visiting international trainers. One snapshot per race meeting day, captured at the close of the meeting. Snapshot history begins December 2025.
HK Companies
6 datasetsRegistered-office timeline for Hong Kong companies
A registered-office address timeline for Hong Kong companies. Read it as a history of where a company has been on record, not yet as a current-address lookup. Three of its arms are frozen legacy captures that stopped on 2024-04-18, when the modern e-services API withdrew the address field; a live arm reading the Companies Registry's open-data portal resumed observation on 2026-09-03 and is still filling in. RECENCY THEREFORE VARIES BY COMPANY — read last_observed_at per row rather than assuming one as-of date for the dataset. Where the live sweep has not yet reached a company, its most recent address may have been superseded by a move we have not seen. Use it to reconstruct where a company was and when it moved; verify against the registry before relying on it for where a company is today. One row per company per distinct registered-office address observed on record; effective_from/effective_to are the registry's own dates where given (else null) and reach back to 1902, first/last_observed_at are our observation bounds. Coverage: 68.4% of active (non-dissolved) Hong Kong companies have at least one address on record here.
Hong Kong Companies Registry company directory
Directory of every registered Hong Kong company from the Companies Registry: over 3.6 million companies, active and dissolved, dating back decades. Search by company name or business registration number; check incorporation and dissolution dates, active/dissolved status, company type, and winding-up mode. Includes enrichment from the modern ICRIS electronic filing system. The registered-office address is the most recent one we have observed, and its recency varies by company: address_as_of gives the date we observed it, and the full address history is in crhk_address_history.
Hong Kong Companies Registry company directory
The core Hong Kong company directory (the Starter-floor twin of crhk_companies, without its resolved current-address columns). Directory of every registered Hong Kong company from the Companies Registry: over 3.6 million companies, active and dissolved, dating back decades. Search by company name or business registration number; check incorporation and dissolution dates, active/dissolved status, company type, and winding-up mode. Includes enrichment from the modern ICRIS electronic filing system. The registered-office address is the most recent one we have observed, and its recency varies by company: address_as_of gives the date we observed it, and the full address history is in crhk_address_history.
Hong Kong company name changes and historical records
Historical record of company names for Hong Kong registered companies. Each row is one distinct name in one language (English, Chinese, or Other) for one company, with the registry's effective date range where available. Exactly one name per company and language is flagged as the current name. Reconciled across legacy Companies Registry data, the modern ICRIS API, company profiles, and the crhkall database. Useful for identifying predecessor companies and tracing corporate history.
Hong Kong Companies Registry document filings and annual returns
Document filings from the Hong Kong Companies Registry including annual returns, financial statements, director changes, and other regulatory filings. Search by company, document type, or filing date. Contains filing metadata, document pricing, and availability for online viewing. Useful for corporate research and compliance monitoring.
Licensed money lenders, applications and expirations in Hong Kong
Hong Kong's official money lender register under the Money Lenders Ordinance (Cap. 163), published by the Companies Registry: current licensees with licence numbers and expiry dates, pending licence applications, expired licences, and dismissed or withdrawn applications — one row per register entry, bilingual names, refreshed weekly from the registry's data files.
HKEX Disclosures
4 datasetsHong Kong Exchange listed company directors and board members
Directors and board members of Hong Kong Exchange listed companies. Look up who sits on which boards, the company's industry, their appointment dates, tenure length, and whether they are currently serving or have resigned. Holds every director of every HKEX-listed issuer. Useful for corporate governance analysis and director network mapping.
Hong Kong Exchange disclosure reason codes reference
Reference table mapping HKEX disclosure reason codes to their descriptions. Use this alongside the HKEX Insider Disclosures dataset to decode form types, event codes, and capacity codes found in disclosure filings. Each code belongs to a category (form type, event reason, capacity, etc.) with a human-readable description.
Concert parties, controlled corporations and capacity behind each HKEX SDI filing
Detail companion to the HKEX Insider Disclosures dataset: the named parties behind each Part-XV (SDI) filing. Where HKEX Insider Disclosures carries only presence flags for the nested interest blocks, this dataset lists them — persons acting in concert, other parties to a s.317 agreement, controlled corporations, joint shareholders, trusts, the capacity in which shares are held, and derivative interests — each with its party names, codes and long/short share counts. One row per disclosure form; join to HKEX Insider Disclosures on form_serial_number. Multi-valued blocks are flattened to single, pipe-delimited (" | ") cells, aligned by element index across a block's columns. Dependent (child/spouse) deemed interests are included, as HKEX's own mandatory SDI filing already discloses them publicly.
Hong Kong Exchange insider dealing and shareholding disclosures
Insider dealing and substantial shareholding disclosures filed with the Hong Kong Exchange under Part XV of the Securities and Futures Ordinance. Track share transactions by directors, chief executives, and substantial shareholders of listed companies. Each record includes the discloser's name, stock code, event date, and extracted details such as event type, number of shares involved, and transaction prices. Covers all forms filed through the HKEX electronic disclosure system.
SFC Licences
7 datasetsPrincipal address, email and website for Hong Kong SFC-licensed firms
Principal business address (English and Chinese), email, and website for every Hong Kong SFC-licensed corporation. One row per firm, taken from the SFC public register's address page. Useful for outreach, due diligence, and entity resolution against other corporate datasets.
Complaints-handling officer contacts for Hong Kong SFC-licensed firms
The designated complaints-handling officer for each Hong Kong SFC-licensed corporation — the phone, fax, email, and address to use when raising a complaint against the firm. One row per complaints-officer contact (most firms have one; a few have two). Includes contacts the live register has since dropped (retained from the archive). Useful for compliance, dispute handling, and entity resolution.
Enforcement and disciplinary history for Hong Kong SFC-licensed firms
Regulatory enforcement history for Hong Kong's SFC-licensed corporations: every disciplinary action — fines, public reprimands, suspensions, revocations — with the date of action, the action type (English and Chinese), and the press-release document reference. One row per action. Includes historical actions that the live SFC register has since removed (retained from the archive), so the record is more complete than the public register's current view. Essential for compliance due diligence and counterparty risk screening.
Executive officers of Hong Kong banks conducting SFC-regulated activities
Executive Officers of Hong Kong's registered institutions (banks) that conduct SFC-regulated activities under the AMLO regime: which individual is the EO responsible for which bank's regulated activities (dealing, advising, asset management, etc.), and from when. One row per (executive officer, institution) edge. This is the bank/RI arm of the SFC people↔firm graph — complementary to the responsible-officer and representative datasets (which cover licensed corporations) — and the only SFC source for the executive officers who hold no other licence.
Hong Kong SFC licensed financial intermediaries directory
Directory of persons and entities licensed by Hong Kong's Securities and Futures Commission (SFC). Look up any financial intermediary by their Central Entity Reference (CE Ref) and find their name, whether they are an individual or a corporation, the regulated activities they are licensed for (Types 1-12: dealing, advising, asset management, etc.), whether they currently hold an active licence, and whether they hold an Anti-Money Laundering Ordinance (AMLO) licence, and the validity period of their licensing (when it began and, for past licensees, when it ended). Covers all entities on the live SFC public register plus deregistered entities retained from the historical archive. Essential for verifying whether a financial service provider is properly licensed in Hong Kong.
Licensed-representative relationships for Hong Kong SFC-licensed firms
The licensed-representative graph for Hong Kong's SFC-licensed corporations: which licensed individual is a representative at which firm, and the regulated activities (Types 1-12 / AMLO) they cover there. One row per (corporation, representative) edge. Built from the SFC public register's per-firm representative pages. Useful for compliance due diligence and relationship mapping across the SFC register.
Responsible-officer relationships for Hong Kong SFC-licensed firms
The responsible-officer graph for Hong Kong's SFC-licensed corporations: which licensed individual is a Responsible Officer at which firm, and the regulated activities (Types 1-12 / AMLO) they cover there. One row per (corporation, officer) edge. Built from the SFC public register's per-firm RO pages. Useful for compliance due diligence, relationship mapping, and entity resolution across the SFC register.
ICAC
1 datasetHK Public
1 datasetHK Property
12 datasetsHK building directory with geo, district and completion
Master list of Hong Kong residential buildings from Centaline, a tier of the estate→phase→building→unit hierarchy. English and Chinese estate, phase, building and address names, the parent estate and phase codes, district and county, latitude/longitude, completion date, unit and floor counts, flats per floor, the assigned school net, and a facilities summary. Reconciled across several Centaline sources, newest observation per field.
Per-district building counts and latest median $/sqft
Hong Kong district browse/landing dimension: one row per OFFICIAL Hong Kong district (the 18 District Council districts, keyed on Centaline's stable district_code) with bilingual district and region names, its building count and the latest monthly median saleable $/sqft, for top-level navigation and at-a-glance market context. Centaline's two sub-areas (Tseung Kwan O, Lai King / Lai Wan) are folded into their parent districts, so every figure covers the whole district.
HK multi-phase estates with location, price summary and new launches
Estate dimension: the TOP of the estate→phase→building→unit hierarchy — the large multi-phase Centaline development (e.g. "Ocean Shores", "Yoho Town"). Bilingual name, the stable numeric id, the building / phase footprint, and the estate-level extras: completion-date range, address, HK1980 grid centre, MTR proximity, the Centaline recent unit-price summary with its active-listing & transaction counts, neighbourhood geography, the Chinese developer legal name, and the primary-market new-launch block (new developments only). Join hk_property_phases / hk_property_buildings / hk_property_transactions on estate_code (or the numeric centa_estate_id). Standalone single-phase developments are intentionally absent — reach those through hk_property_phases.
Monthly HK price medians and sales volumes by district and building
Actual monthly property-market aggregates computed directly from registered sales (no modelling or prediction) at three grains: whole Hong Kong, by district, and by building. Median price, median/quartile saleable and median gross $/sqft, and transaction count per month. Known non-residential sales (car parks, commercial, industrial) are excluded; older rows whose class is unknown are kept. Medians are mix-sensitive: they reflect the composition of what sold each month, not constant-quality price movement — subsidised-sale (HOS) registration waves at roughly half private-market $/sqft can move a single month's pooled median by more than 10%, so month-over-month deltas must not be read as market appreciation. p25/p75 ship alongside so the spread is visible. The in-progress current calendar month is kept but flagged month_is_complete = false; even a complete month keeps gaining late-registered sales for a few weeks.
Per-neighbourhood (HMA) volumes, buildings and geo
Hong Kong neighbourhood (HMA / 中原樓市片區) browse dimension below the district: one row per Centaline neighbourhood with bilingual names, its parent district code, region, registered-sale volume, building count and a representative coordinate.
Development phases with completion span, building count and location
Development phases: the middle level of the estate→phase→building hierarchy. A phase groups buildings within an estate (e.g. "PHASE 2 CROWN OF ST BARTHS"). The parent estate code and numeric ids, bilingual phase names, the completion-date span of its buildings, a building count, and the phase's centre coordinates and neighbourhood. Join hk_property_buildings.phase_code (members) and, via estate_code, hk_property_estates (parent estate).
Hong Kong amenities with category, bilingual names and coordinates
Hong Kong points of interest from several sources, one row per amenity record: the Centaline property portal's amenity graph, and Hong Kong government open data — Hospital Authority hospitals, LCSD sports centres and parks, public libraries, Transport Department ferry piers, the Lands Department's iGeoCom GeoCommunity Database (convenience stores, supermarkets, markets, cooked-food stalls, clinics, bus termini, MTR entrances, kindergartens and more) and the FEHD register of licensed restaurants. Near-duplicates across sources are removed, and iGeoCom's several records for one premises collapse to one amenity. Each point carries a category (Transport, Parking, Daily convenience, Food & drink, Health, Education, Recreation), a finer sub_type, bilingual names, WGS84 coordinates and, where the source gives one, a bilingual street address. Pair it with the coordinates in hk_property_buildings to measure what is near a building.
Pending-registration HK property sales (sales pipeline)
Provisional (pending-registration) Hong Kong property sales from Centaline — sales agreed but not yet registered with the Land Registry. A leading market-velocity / sales-in-progress signal: these typically register weeks later and reappear in hk_property_transactions with a memorial. Kept separate from registered sales so the two are never double-counted.
HK residential rental listings with monthly rent and rent/sqft
Hong Kong residential rental records from Centaline (Centadata). One row per rental record: date, monthly rent, saleable and gross floor areas, rent per square foot on both bases, plus bedroom count and aspect. The rental counterpart to hk_property_transactions (sales). Rentals never register with the Land Registry, so this is a recent series rather than a deep historical archive.
Registered HK property sales with price and $/sqft
Registered Hong Kong residential property transactions as recorded by Centaline (Centadata). One row per Land Registry memorial: registration date, consideration (price), saleable and gross floor areas, and price per square foot on both bases. The core fact table for property analytics.
Individual flats with block, floor, flat and floor areas
Individual flats/units, the lowest tier of the property hierarchy, built from each unit's most recent registered sale (Centaline publishes no separate unit list). Block, floor, flat and latest known saleable and gross areas, plus model-ready fields: a parsed integer floor_number, the unit's position in its building's grid (unit_grid_x/unit_grid_y, 0-based flat-column and floor ordinals — for per-building flat-grid heatmaps and as hedonic-model features), and ratio-imputed areas flagged by the *_is_imputed columns.
Centaline property-type codes with bilingual labels
Property-type / usage reference dimension: Centaline's scheme code for every transacted unit (RE = private residential, RH = Home Ownership Scheme, the R* subsidised-sale schemes, CP/SP/CM = unit-usage-only codes incl. car parks) with its bilingual label. Join on hk_property_transactions.usage_code or unit_usage_code to label a transaction's property type.