What makes this especially interesting is the separation between the extraordinary story and the signal itself. A 1,650% rise, forced liquidation and index inclusion are almost impossible for a human investor to view without emotion—yet the model simply ranks what is in front of it. That discipline is valuable. The real test now, however, is whether the ranking captures continuing fundamental and price strength or merely rewards an already extreme move. I would be particularly interested in how valuation, reversal risk and position sizing enter the process here. A fascinating pick—and probably the clearest possible demonstration of both the strength and the psychological difficulty of following a quantitative system.
This is a deeply insightful observation. That tension is exactly it — the model doesn't distinguish "still strong" from "already extreme." It just scores the same factors across 450+ stocks every month, no valuation check or reversal logic layered on top.
Position sizing (1-3%) and the SPY market signal are the only risk controls. Backtested across 95,000+ stock-months with that in/out SPY overlay, it's held up reasonably well over time — but no illusions that it's making the distinction you're describing in real time.
What makes this especially interesting is the separation between the extraordinary story and the signal itself. A 1,650% rise, forced liquidation and index inclusion are almost impossible for a human investor to view without emotion—yet the model simply ranks what is in front of it. That discipline is valuable. The real test now, however, is whether the ranking captures continuing fundamental and price strength or merely rewards an already extreme move. I would be particularly interested in how valuation, reversal risk and position sizing enter the process here. A fascinating pick—and probably the clearest possible demonstration of both the strength and the psychological difficulty of following a quantitative system.
This is a deeply insightful observation. That tension is exactly it — the model doesn't distinguish "still strong" from "already extreme." It just scores the same factors across 450+ stocks every month, no valuation check or reversal logic layered on top.
Position sizing (1-3%) and the SPY market signal are the only risk controls. Backtested across 95,000+ stock-months with that in/out SPY overlay, it's held up reasonably well over time — but no illusions that it's making the distinction you're describing in real time.
Appreciate you digging into it this deeply.