Skip to content
Montex Alpha Rail System
$497

Coefficient Calibration · Why Not Eyeballing

Live ATR alone lacks structure.

Baseline percentiles

P20–P80

Session baselines

4

Asset-class ladders

FX 7 · Idx/Au/Oil 3

Reliability grades

5

The key idea

Worked example

ENTRYSTOP = 1.2 × ATRBE EARNEDenvelope = price ± coefficient × ATR(authority timeframe)

Price inside its volatility envelope. The stop is quoted in ATR, and break-even is earned at a volatility-defined distance - not felt.

Four structured inputs

What the calibration reads that the chart doesn't show.

A coefficient decision in MARS is conditioned on four inputs, none of which are visible in the candle the eye is judging:

  • Trigger timeframe — the frame the trade executes and trails on.
  • Authority timeframe — the higher frame whose structure governs how much breathing room is justified.
  • Asset class — FX reads against a seven-tier ladder; indices, gold, and oil against broader three-tier ladders because they are structurally more volatile.
  • Engine mode — Conservative, Balanced, or Aggressive strictness, chosen before the reading, from gate state and trend quality.

The reliability grade

02

The eye cannot tell you how much to trust itself.

Every baseline the calibration reads against carries a sample-status grade — No Data, Insufficient, Provisional, Developing, Hardened — earned by logged sample count for that exact asset, timeframe, and session. That grade is the input eyeballing cannot produce: an operator's impression of 'normal volatility' has no attached count of the observations behind it, no session separation, and no flag saying the baseline is eight samples old. The calibration's verdicts arrive pre-qualified; the eye's arrive naked.

What a wrong guess costs

The two failure directions are priced differently.

Guess too tight and a valid trend trade dies to noise — on Trend No-Partial, the branch built to capture outliers, a choked runner caps exactly the fat-tail R the branch exists for. Guess too wide and every stop-out costs more than the structure required, giveback expands, and per-trade EV quietly erodes across the whole sample. Both errors are invisible on the trade where they happen and expensive across a hundred trades — which is why the answer has to be a parameter, not a feeling.

The consistency argument

Eyeballed distances can be right. They cannot be consistent.

A skilled eye sometimes beats the coefficient on a single trade — the problem is that eyeballing produces distances that vary with confidence, fatigue, and the last outcome. Consistency is what the evidence base needs: a hundred trades under one coefficient are analyzable; a hundred trades under a hundred moods are noise.

Reference

The eye versus the calibration stack, input by input

InputWhat the eye usesWhat the calibration uses
BaselineYesterday's impressionP20–P80 percentile cuts of logged samples per asset/TF/session
SessionUsually ignoredAll-Sessions, London, New York, and Asian baselines compared
Asset classOne intuition for everythingFX seven-tier ladder; Index/Gold/Oil three-tier ladders
Higher-timeframe contextThe chart being watchedTrigger → authority map with a rank-based add-on
ReliabilityUnknown and unknowableSample-status grade attached to every verdict

Connected inside MARS

This module doesn't work alone.

Connected — 01

Coefficient Calibration (full page)

Open system page →

Connected — 02

Feedback Loop

Open system page →

Connected — 03

The 1.5 → Wide Spectrum

Open system page →

Connected — 04

Volatility Distance Matrix

Open system page →

Go deeper

Operator briefs on this territory.

Deep dive — 01

The cost of eyeballing is not wrong distances. It is that nothing can ever be graded.

Concede that the eye is often right. The problem is what it does to the evidence base.

Read the full brief →

Deep dive — 02

A raw ATR reading means nothing without a reference, and the eye's reference is about a day long.

You are always comparing against something. The question is whether the something is a day or a quarter.

Read the full brief →

Deep dive — 03

The calibration reads four things. The chart in front of you shows one of them.

Not a question of skill. Three of the four inputs are simply not in the window you are looking at.

Read the full brief →

Every module ships in the complete MARS package.

One price. Eleven workbooks, three TradingView indicators, and the full manual library — $497.

Get MARS Pro — $497Explore the Platform