Subject: Overfitting gate — possible per-scientist state affecting my submissions since 21 Aug

My first 8 submissions all passed validation (e.g. signal_2215f6777613f88f, accepted 21/08 01:38:48, currently Q14). My 8 submissions since then were ALL rejected as "Overfitted submission" — including one file that is byte-identical to my accepted signal_2215f6777613f88f except a single label-independent position-scaling line (submitted 23/08), and one architecturally new per-asset model with out-of-sample validation stronger than in-sample (submitted 24/08). Meanwhile the leaderboard shows dozens of new signals from other scientists accepted in the same window.

Could you confirm whether the overfitting gate includes any per-scientist adjustment (submission count, recent-rejection streak, or similar)? If it does not, the one-line-diff pair above may be worth checking on your side. Happy to share both files. — Quant_a