Changelog

This page documents all notable changes to the Blocklens platform and API.

Cohort Spending Analysis (Enterprise dashboard)

Data correction: for ranges longer than two years, the week or month at each internal loading boundary was counted twice (e.g. the week of 2019-12-30 showed ₿2.3M instead of ₿1.14M), and yearly SOPR was approximated (LTH SOPR for 2018 showed 1.10 instead of 1.33). Both now match the source data exactly. Dates no longer shift by a day for users east of UTC.
Incomplete first/last periods are marked as partial; the default cohorts now start with 1d (age exactly 1 day) and 1d–1w (2–6 days).
Faster loading for long ranges, instant BTC/USD switching, and the dashboard state (range, aggregation, unit, tab, dates, custom cohorts) is kept in a shareable link.
Market X-Ray opens with a default comparison (latest period vs ~1M/3M/6M/1Y earlier), periods are picked by year, columns are labelled by cohort.
Cohort Breakdown adds a % of volume view, a cohort SOPR overlay and the share spent at a loss; SOPR Pulse uses a log scale by default.
Cohort editor validates names and ranges before applying; dates and numbers follow your language; exports (PNG/CSV/JSON) contain exactly what the active tab shows.

API

POST /v1/enterprise/sopr: buckets and cohorts now also return btc_volume, usd_volume and cost_basis_usd; each period returns first_date, last_date and days. Cohort names may use any script; duplicate names and overlapping age ranges are rejected with 422.

Data correction

Net Liquidity (macro_net_liquidity): the Reverse Repo component was subtracted without a unit conversion, which overstated the level by up to ~$2T in 2021–2023 (e.g. 2022-06-01: $6.16T, previously shown as $8.12T). The full history has been recomputed. The series is now published for every US business day instead of weekly.
Bitcoin treasuries (/v1/dat/*): holdings for a part of the tracked companies and governments were stuck at their 2026-08-10 values. History for 2026-08-11 → 2026-10-16 has been corrected and aggregates recomputed.
ETF totals (/v1/etf/aggregate): Grayscale Mini was counted twice for 2024-07-31 → 2026-07-18, overstating total ETF holdings by up to ~48K BTC (e.g. 2025-06-02: 1,301,052 BTC, previously 1,344,415). Total holdings, AUM, BTC dominance, ETF count, market shares and the ETF realized price were recomputed.
WisdomTree Bitcoin Fund (BTCW) is now counted as a US spot ETF: us_btc_holdings, us_aum and the US spot ETF realized price include it since 2024-01-11.
The descriptions of etf_realized_price and us_spot_etf_realized_price now describe the actual calculation (sum of per-fund weighted-average cost / BTC held); values are unchanged.

ETF flows

Daily flows are now aligned to the trading date for every fund.
New fields: flow_usd (/v1/etf/ticker) and us_spot_net_flow_usd (/v1/etf/aggregate) — daily net flow in USD.
Corrected historical flows for FBTC and IBIT, and filled missing days (e.g. ARKB 2026-09-21).
Added MSBT (Morgan Stanley Bitcoin Trust).
flow_usd and us_spot_net_flow_usd now cover the full history since 2024-01-11; the newest day is provisional until every fund's figure is published.
Cumulative USD flows recalculated from the exact daily USD flow: us_spot_cumulative_money_flow is now the sum of us_spot_net_flow_usd since 2024-01-11 ($56.9B on 2026-10-16, previously estimated at $83.3B). etf_cumulative_money_flow uses the exact flow for US spot ETFs; non-US ETPs and GBTC before 2024 remain an estimate (BTC change × BTC price).
The ETF Net Flow and ETF Net Flow Comparison charts now show the USD flow next to the BTC flow.

Derivatives

BitMEX removed from open interest and funding after the exchange shut down (data up to 2026-09-16 is kept).
Added Kraken and Bitfinex funding rates.

Data correction

Corrected historical BTC daily price OHLC for 2026-04-17 → 2026-07-11 to match our reference source. Price-derived series — realized volatility (1w–1y) and price drawdown — were recomputed for the affected range.
Restated Realized Cap by Age (<24h) and the Realized Cap by Age (%) breakdown for 2026-07-06 and 2026-07-07, which were previously understated.
No other metrics were affected: realized cap, MVRV, NUPL and holder series are unchanged.

Snapshot API

SVG output is now a first-class citizen alongside PNG. The POST /v1/chart/snapshot format field accepts "png", "svg", or "json"; both GET and POST also honor Accept: image/svg+xml.
New route: GET /v1/chart/renders/{render_id}.svg — rendered SVGs are persisted for 1 hour and re-fetchable by URL, mirroring the existing .png behavior.
SVG responses now include X-Render-Id so the persistent URL can be reconstructed from a single render call.
PNG and SVG variants of the same chart are cached independently.
Cycle performance overlays (x_axis: "day_offset") now render X-axis tick labels as calendar dates from the latest cycle of the chosen family (e.g. Oct '25, Jan '26, … for ATH overlays anchored at cycle_ath_5 → 2025-10-06) instead of Day 0, Day 90, … . Earlier cycles remain aligned on the same axis positions.

MCP Server

render_chart: the format parameter enum is now "png" | "svg" | "json" (was "png" | "json"). When format: "svg" is passed, the tool returns an image/svg+xml content block and a /chart/renders/{id}.svg URL in the summary.

Metrics & API

Metric count expanded from 127 to 210+ active metrics (exchanges, derivatives, DAT, and more).
New endpoint: GET /v1/cycle-boundaries — Bitcoin halving cycle start/end dates and durations.
Metrics API now uses a registry-driven query builder for consistent response formatting.

Snapshot API

Heatmap rendering: render distribution-based charts (e.g., Cost Basis Distribution) with configurable color scales (viridis, plasma, etc.) and Y-axis scales. Includes color scale legend.
Cycle performance overlay charts: x_axis: "day_offset" aligns all halving cycles to Day 0 for comparison.
New per-metric parameter: show_in_legend — control whether individual metrics appear in the chart legend.

MCP Server

New tool: get_cycle_boundaries — query halving cycle dates directly from AI agents.
render_chart now supports heatmap rendering and cycle performance overlays.
Unified TypeScript codebase (tools.ts) for all MCP tools.
Tool count: 18 → 19.