Experimental predictions: the model is reference statistics, not validated against the market yet. The track record shows real settled picks when they exist, otherwise clearly-labelled sample rows. No accuracy or profit guarantee.