Select Data · Institutional market microdata

Exclusive, self-recorded market microdata.

Recorded live, tick by tick — not scraped, never backfilled. Point-by-point sports and crypto microstructure, captured forward-only and frozen at the moment it happened. Built for quants and researchers who need history they can trust.

Self-recorded CSV + Parquet Delivered on request

Independent data vendor. Datasets contain publicly observable market data, recorded independently.

01 · Live

Recorded, not scraped

Every record is a live capture, stamped with its ingestion time — not reassembled from an endpoint that only shows the present.

02 · Frozen

Point-in-time integrity

Forward-only capture. Nothing is revised, relabeled or “corrected” after the fact, so what you test on is what was really there.

03 · Honest

Gaps logged, never faked

Disconnects are written as explicit gap markers, never interpolated into a false continuity. Voided sessions stay in the archive.

01 · Datasets

Three datasets at launch

More on request

Each dataset ships as CSV and Parquet with a documented, versioned schema. Coverage is ongoing — the field lists below are sample schemas; contact us for current record and session counts.

Tennis · point-by-point

Tennis Match Ticks + Live Odds

New · now recording

Point-by-point match state paired with live moneyline and per-set odds, captured tick by tick for the full match. Our newest recording line.

SamplingPer point + every odds change (tick-level)
FieldsMatch state + odds book (see schema)
CoverageRecording since 2026 · ongoing
FormatCSV + Parquet
DeliveryOn request (signed link or direct transfer)
Sample schema
state (per point)ts_ms, match_id, tour, round, server, point_no, point_winner, game_score, set_games, sets, is_break_point
odds (per tick)ts_ms, match_id, market{ml|set}, selection, bid, ask, mid, implied_prob

ATP / WTA / ITF · recording ongoing · counts on request.

MLB · pitch / play-level

MLB Pitch & Play Events + Odds

Recording

Pitch-by-pitch and play-level events aligned with live moneyline, run-line and total odds across the game.

SamplingPer pitch / per play + odds ticks
FieldsPitch/play events + odds (see schema)
CoverageRecording since 2026 · ongoing
FormatCSV + Parquet
DeliveryOn request (signed link or direct transfer)
Sample schema
events (per pitch/play)ts_ms, game_id, inning, half, balls, strikes, outs, base_state, batter_id, pitcher_id, pitch_type, release_speed, pitch_result, play_event, home_score, away_score
odds (per tick)ts_ms, game_id, market{ml|total|runline}, selection, price, implied_prob

Regular-season games · recording ongoing · counts on request.

BTC · 5-minute

BTC 5-Minute Bars + Order-Book Microstructure

Recording

5-minute OHLCV bars alongside forward-only order-book snapshots — depth, spread and imbalance — for microstructure research.

Sampling5-minute bars + order-book snapshots
FieldsOHLCV bars + book snapshots (see schema)
CoverageRecording since 2026 · ongoing
FormatCSV + Parquet
DeliveryOn request (signed link or direct transfer)
Sample schema
bars (5-minute)bucket_ts_ms, open, high, low, close, volume, vwap, trade_count
book (snapshot)ts_ms, mid, spread, bid_px[1..N], bid_sz[1..N], ask_px[1..N], ask_sz[1..N], depth_imbalance

Spot BTC · recording ongoing · counts on request.

02 · Why recorded

Why recorded beats scraped

Scraped history is assembled after the fact from endpoints that only ever show the present. Recording captures the past as it actually happened — and that difference is the whole product.

Fidelity

Timestamp fidelity

Every record carries the capture time at ingestion (epoch milliseconds), not a re-derived or vendor-normalized clock. You see the book exactly as it stood, at the instant it stood there — the arrival order and latency structure a scrape flattens away.

Integrity

No backfill, ever

Recordings are forward-only. Nothing is reconstructed, revised or restated after capture. Scraped datasets silently inherit later corrections and re-labelings; a recording is frozen the moment it lands, so a study runs on the real past, not a cleaned-up version of it.

Population

Survivorship-bias immunity

Because capture is live, sessions later voided, abandoned or delisted stay in the archive exactly as they occurred. Scrapes of “current” endpoints quietly drop whatever no longer exists — the classic hidden bias. We keep the full population, disconnect gaps included and clearly marked.

03 · Custom

Custom recording — name the source, we record it

If it streams, we can capture it forward-only into the same canonical schema — sports, crypto, or any public real-time market feed. Tell us the venue, the markets and the fields you need; we scope the recording, start capturing, and deliver on your cadence in CSV and Parquet.

Recordings begin at commission. We do not sell history we did not capture, so a custom set accrues from day one — and it is yours to evaluate on a sample before you commit.

04 · Licensing

Licensing

Three tiers, priced on scope and redistribution rights. All pricing is by contact — tell us the datasets and the use, and we’ll quote.

Research

Contactfor pricing

Individual & academic research

  • One dataset of your choice
  • CSV + Parquet snapshots
  • Documented, versioned schema
  • Internal / non-redistribution use
  • Email support
Contact

Fund

Most requested
Contactfor pricing

Desks & quant teams

  • All datasets
  • Ongoing updates as we record
  • Multiple seats
  • Priority custom-recording queue
  • Sample-first evaluation
Contact

Enterprise

Contactfor pricing

Firms & data platforms

  • Everything in Fund
  • Bespoke recording pipelines
  • Programmatic / bulk delivery
  • Redistribution terms available
  • Onboarding & delivery SLA
Contact
Get started

Request a sample or a quote

Onboarding is manual during early access. Tell us which datasets you want and how you’d use them — we’ll send a free sample and open a private channel to scope the rest.

Free samples · limited early-access slots · open a 🎁 ticket in Discord #support. Coverage and pricing are indicative and may change.