Five million backtests.
Two thousand calibrated candidate edges.
Eighty survivors that cleared the full validation gauntlet.
A small set under forward observation in live market conditions.
That is the filter behind this newsletter. Not the wins. Not the screenshots. The graveyard.
You almost never see the graveyard. You see the clean setup after the fact, the equity curve after the filtering, the signal after the dead versions have been buried. That is where false confidence breeds.
The mortality rate is the point. If everything I tested survived, the gauntlet would be theater. The fact that most candidates die is what gives the survivors any weight.
What "survived" actually means
A candidate edge is a trading idea with enough structure to test.
Calibration is the cheap part. Most ideas can be made to look convincing on a backtested chart. Pretty equity curves are easy to generate when you let yourself peek at the future.
Survival, in this work, means something narrower:
The edge held up across regimes I had not tuned it for.
It survived broad asset and timeframe pressure, not just the one chart where it looked tidy.
Execution costs (slippage, commission, the gap between the print and the fill) did not eat the math.
It survived out-of-sample windows the calibration never saw.
Forward observation, with real time passing and the market not knowing the test was running, did not kill it.
In this context, validated does not mean guaranteed. It means the idea survived the gates I use before I let it near publication, deployment, or larger risk decisions.
A small subset earns the word deployed.
Almost nothing earns the word certain.
What the survivors have in common
After watching roughly two thousand candidates go in and eighty come out, a pattern shows up.
The survivors are rarely the prettiest. They are rarely the ones that looked obvious on the initial chart. The obvious ones usually had a hidden dependency on a regime I had not tested properly. They worked in one market, then stopped working in another. The calibration window flattered them.
The survivors share three traits. First, they name their regime. A survivor does not claim to work everywhere; it says, I work when X is true. That makes it auditable, and it makes the kill switch obvious.
Second, they tolerate execution friction. A signal that needs perfect fills is not an edge. It is a chart pattern. Survivors preserve a margin between paper math and real math.
The third matters most: they lose cleanly. Survivors have a defined wrong-state. They do not drift into rationalization. The exit rule is part of the edge.
Most candidates die not because the entry was bad, but because the exit was vague. A clean loss is a feature.
What killed the rest
The mortality curve has a shape. Most candidates die in one of four ways.
Regime-coupled. Worked in one decade, not the next. Calibration window flattered them. This is the most common cause of death.
Execution-fragile. The edge existed on paper. It did not survive contact with slippage, partial fills, or the spread.
Overfit. Three parameters tuned on the same window, and the fourth window said no. The chart was a postcard from a place that does not exist.
No kill condition. The entry was clear. The exit was a vibe. When the trade went against the operator, the rule became negotiable.
None of these are exotic. All of them are common.
Each future issue studies one survivor, one failure, or one part of the validation process that decides the difference.
The format rotates through a small set of columns:
Pivot Point. A regime read. What the market is mechanically doing, where rotation pressure sits, and what posture survives the most plausible counter-moves.
Validation Rigor. A candidate edge walked through the gauntlet in public. What cleared, what nearly failed, what was rejected.
Edge of the Week. A field note from the live forward-observation set. Not a recommendation. A report.
Prop Firm Lab. Notes from the prop-firm research lane when something useful transfers beyond that context.
The cadence is weekly. The rule is simple: no filler.
The contract
Most trading content sells confidence.
The PrimeFold Press publishes the filter.
You do not have to take my word that the work is rigorous. The validation procedure is the product. The survival rate is the receipt.
Each issue picks one survivor, one death, or one gate in the process and shows the reasoning end-to-end. If a gate seems wrong, push back. If the regime call looks off, push back. The work gets better when serious readers can argue with it.
Eighty survivors out of two thousand candidates is not a brag. It is a description of how narrow the door is.
Most of the work is throwing things out.
— Collin
The PrimeFold Press
One question to carry into Issue #2:
What was the moment you knew one of your own trading ideas was dead?
Methodology note. Each candidate edge runs through six gates: in-sample calibration, an out-of-sample window untouched by tuning, cross-asset basket testing, slippage and execution stress, BH-FDR correction to control false discoveries across the candidate population, and forward observation in live market conditions. Surviving a single gate is cheap. Surviving all six is rare. The 80/2,000 ratio reflects the full chain, not any single filter.
In The PrimeFold Press, validated means the idea survived the procedure described above through the date of publication. It does not mean guaranteed, profitable in your account, or future-proof. Live capital is deployed selectively as conviction and capacity allow; specific positions are disclosed in field-note issues when relevant.
Validated = survived the procedure through date of publication. Not financial advice. Not a guarantee. Not future-proof.
Push back: Reply to this email or post in the discussion thread. The work gets better when serious readers can argue with it.
