refresh card (as_of=2026-07-16)

README.md

@@ -1,3 +1,43 @@
 ---
-license: mit
+license: cc-by-4.0
+pretty_name: cascade held-out eval pool (24h-lagged reveal)
+tags:
+  - time-series
+  - forecasting
+  - benchmark
 ---
+
+# cascade eval pool — lagged public reveal
+
+Daily snapshots of the **held-out evaluation pool** used by the
+[cascade](https://github.com/TensorLink-AI/cascade) subnet, built from the
+[tsbench-forge](https://github.com/tensorlink-dev/TSBench-Forge) live catalog of
+real public time series across the 7 GIFT-Eval domains.
+
+**Each snapshot is revealed 24h AFTER it was used for scoring** (`as_of` is a
+closed UTC day). Because the pool rotates daily, a released snapshot is always
+for a round that is already scored — it lets anyone **reproduce the leaderboard**
+without exposing the round currently in play.
+
+## Layout
+
+```
+snapshots/<YYYY-MM-DD>/          one folder per revealed day (as_of)
+  <series_id>.npy                float32, freshest context_length + horizon points
+  metadata.json                  {series_id: {freq, seasonal_period, domain, source}}
+  provenance.json                build config (context_length, horizon, builder_version)
+```
+
+Scoring is identity-agnostic: MASE uses `seasonal_period`, CRPS/WQL use the
+values + a model's quantiles, and the KOTH cluster-bootstrap groups windows by
+`source`. Re-cut windows of `context_length + horizon`, forecast the horizon,
+and the metrics reproduce byte-for-byte.
+
+## Latest snapshot — `2026-07-16`
+
+- **series:** 228
+- **domains:** nature=86, sales=63, econ_fin=33, energy=17, web_cloudops=15, healthcare=13, transport=1
+- **cadences:** D=96, 15min=48, min=21, 6min=16, 10min=13, 30min=11, H=11, 5min=6, 240S=6
+
+> This is an **evaluation** set, not training data. Publishing it to train on
+> would contaminate the benchmark it exists to measure.