I benchmarked 25 models against doing nothing. None of them won

The project benchmarks 25 models against a "do nothing" baseline, revealing that none outperform it, which is a useful sanity check for model evaluation. To ado