Every fund has a stated process. The process gets written down, framed, and put on a deck. What separates funds that survive their first drawdown from funds that don't is a much smaller, much less glamorous artifact: a written list of the trades the process forbids. The forbidden list is shorter than the allowed list and harder to write — but it is the list that holds when conviction wavers.
One: single-signal trades
The temptation is constant. The insider bought ten thousand shares last week, you can read the Form 4 yourself, and the stock is down four percent today. It looks like a gift. Our process says no. One witness is a story. We need eight, ten, twelve. If only insider activity is firing, the composite score will be low, and we will not take the position — even if the insider is the CFO and even if the buy is the largest in five years.
Why? Because the times we have stretched to take a single-signal trade are the times we have learned, expensively, that the witnesses we ignored were saying something important. Quality was deteriorating. The drift was negative. The chart was broken. The insider was wrong. The chorus was a soloist for a reason.
Two: story stocks without earnings
A name that has not produced positive earnings in the last eight quarters cannot be scored on our valuation or persistence layers. Two of twelve layers are silent. The composite score, on a name like that, is structurally capped — it cannot reach the threshold even if all ten remaining layers fire. The math is the discipline. We do not trade names where the math forbids us.
This rule rules out a lot of biotech, most early-stage tech, and many of the names that show up in the Reddit-tier conversation. We are aware of the opportunity cost. We accept it. Our process is not about being everywhere; it is about being right where we are.
Three: IPOs in lock-up
For the first 180 days after an IPO, insider Form 4 activity and 13F holdings are mechanically distorted. Insiders cannot sell. Allocators have not had time to fully reflect new positions. Our chorus is forced into silence on two layers. We do not trade IPOs in lock-up — not even the ones that look interesting on the technicals, because the technicals on a sub-180-day name are dominated by underwriter support and first-mover speculation, not by the durable price action that our layers were calibrated on.
Four: low-ADV names
Below five million dollars of average daily volume, our entry is the price. Slippage eats Sharpe. We have run the numbers on names where the median daily volume is below the threshold — the in-sample edge is real, but the realised edge after costs is marginal at best. The strategy needs to be invisible. If we cannot be invisible, we do not show up.
Five: macro on conviction alone
We do not short the dollar because the deficit looks bad. We do not buy gold because inflation looks structural. We trade markets, not narratives. Macro shows up in the process only as a regime filter — when the VIX is above a threshold, we cut size; when the term structure inverts, we tighten gates. Macro never appears as a thesis on its own.
The forbidden list is not exhaustive. It is the spine. When the chorus is loud and the signal is clean, the rules above hold us back from many trades that would have worked. They also hold us back from many trades that would have worked the first time and taught us a habit that would have killed us the second.