{
  "hypothesis": "H1 NEW-LISTING DRIFT: newly listed USDT perps underperform the market over their first month (supply unlock / airdrop / MM inventory selling into listing hype).",
  "registered_utc": "2026-11-04T22:59:17Z",
  "data": "pit.db (Binance USDT-M, 683 crypto perps incl. delisted, 4h, 2020-01..2026-10)",
  "event": "first 4h bar of a symbol, only if after 2020-02-01 (earlier = archive start, not a listing)",
  "entry": "close of the bar 24h after the first bar (skip listing-day chaos)",
  "position": "SHORT the new coin, hedged LONG an equal-weight basket of all other perps trading at entry (market-neutral excess return)",
  "hold": "30 days; if the coin stops trading earlier, mark at its last close",
  "liquidity": "median 4h turnover over the first 24h >= $1.67M (live screen)",
  "costs": "0.055%/side taker on both the short and the basket (4 x 0.055%), funding over the hold on the short leg (short receives positive funding, pays negative)",
  "metric": "mean excess return per event, net",
  "clustering": "t-stat clustered by listing calendar month",
  "split": "TRAIN events before 2023-09-01, TEST on/after",
  "confirm_bar": "TEST mean > 0 with clustered t >= 2.0, TRAIN same sign, >= 40 TEST events",
  "fail": "TEST t < 1.5 -> dead; between -> inconclusive, no tuning on this data",
  "secondary_reported_not_decisive": ["hold 7d and 14d", "no-hedge raw short", "by year"],
  "known_contamination": "ticker renames (MATIC->POL, RNDR->RENDER etc.) appear as new listings; reported count, not removed"
}