How the scan works

Methodology

Last updated: August 12, 2026

This page has no performance numbers on it. That's deliberate, not an oversight — see why we don't publish them below. What it does have is how the scan works: what the scanner checks, how a setup gets graded, when it gets suppressed, and how changes to any of it get tested before they reach a live issue. It describes the rules, not the thresholds — the specific parameter values are the product, and we don't publish those.

Two disciplines, one scan

Every candidate is checked against two independent frameworks, not one. A volatility-contraction pattern (VCP) reading on the chart, and a CANSLIM-style fundamentals read on the business. A name has to clear both to be graded — a tight base on a company with deteriorating fundamentals doesn't get the same grade as a tight base on one with improving earnings.

The 8-point trend template

Before anything else runs, every candidate is checked against Mark Minervini's 8-point trend template — the standard set of moving-average and 52-week-range conditions used to confirm a stock is in a genuine Stage 2 uptrend, not just consolidating inside a longer downtrend. Names that fail it don't proceed to grading, regardless of how tight the base looks.

The grade

Setups that pass are scored across four weighted factors — VCP pattern confidence, how tight the base contraction actually is, relative-strength percentile against the market, and the fundamentals score — and converted into a single letter grade, A through F. The trend template is deliberately not one of them: nothing reaches grading without already passing it, so scoring it again would only add a constant to every setup and compress the range the other factors have to work in. The grade is a classification, not a raw output: we publish the letter, not the underlying number, because a bounded grade is honest about the precision a heuristic like this actually has. A stock that scores 74 isn't meaningfully different from one that scores 76, and showing two decimal places of false precision would be a worse signal than a clean letter.

Regime gating

The scan doesn't run the same way in every market. A separate, market-wide check reads the tape before any individual setup gets surfaced, and it tightens the scan as conditions deteriorate: a breakout has to clear a larger volume surge to be confirmed, and the stop floor narrows. In a confirmed downtrend the sizing framework goes to zero — the issue still shows you what the scanner found, and it sizes nothing new. Distribution days feed that read rather than forming a state of their own: enough of them move the market out of a confirmed uptrend on the count alone.

The goal is a newsletter that tells you when to stand down, not one that manufactures a reason to trade every single issue. Cash held because the tape says so is a different thing from cash held because nothing came up — this is what keeps that distinction real instead of just a line in the compliance disclaimer.

Suppression flags

Individual names get suppressed too, and Signals issues show why. The most common reason is an earnings report landing inside a short window — a tight base can look ready to break out and then gap on news before anyone can act on it, so those candidates are flagged and held back rather than listed as if nothing were pending. The removal happens at every tier — it's a safety gate, not a tier feature. What Signals adds is the reason.

How rule changes get tested before they ship

The scan's parameters — confidence thresholds, volume-surge multipliers, how tight a contraction has to be — aren't tuned by feel. Changes run through a walk-forward backtester against historical price data before anything reaches the live scan, specifically to catch a rule that only looks good because it was tuned to fit one stretch of the past. Live signals are also tracked forward after the fact, so drift between how a rule performed historically and how it's performing now gets caught rather than assumed away.

Why we don't publish backtest results

The Coil Report is a financial publication, not an investment adviser. A backtested win rate is a number whose window we chose. Marketing one implies future results, so we don't publish one here or anywhere else on this site. That's a compliance line, not a marketing choice we'd reverse if the numbers looked good.

A forward-tracked record is a different object, and we do publish that. Every graded setup is logged on the day it's graded, before anyone knows how it resolves, and nothing is removed afterwards. That running record goes to Signals subscribers, and every signal that finishes also appears in the free Letter's Receipts section. It's young, and the issue says so plainly rather than implying more than the sample supports.

What everyone gets is the process above, honestly described. What each issue carries is the setups themselves and the levels for your tier — the pivot at every tier, and the invalidation level, profit targets and a trailing stop at Signals. Judge the scan by watching it work in real time, not by a backtested number.