The Momentum Agreement Archetype: when timescales must agree
One monitored market series is read at several timescales, each reduced to a direction vote, and a position is taken only when a supermajority of them agree, standing flat whenever they do not.
Method archetype
Many direction methods a reader meets reduce the market to one number and trade that number. A momentum-agreement model does something deliberately different. It reads the direction of one monitored market series at several different timescales, turns each timescale into a plain up-or-down vote, and takes a position only when a supermajority of those timescales agree. When the timescales disagree, the method stands aside.
The whole construct is a count of votes against a fixed bar, and the discipline is in what it refuses to do. The rest of this explainer follows that idea to its consequences: what the method reads, how a horizon becomes a vote, how the votes are counted, why every horizon counts equally, and why standing flat is the behavior rather than a footnote.
What this method reads: one series, several timescales
The method watches a single monitored market series and asks one question of it at several speeds: which way is this series moving. It measures the series' directional change over a small set of different lookback windows, from fast to slow. The important point, and the easiest one to get wrong, is that these horizons are not separate indicators. They are the same series viewed at different timescales.
Picture one line being judged by a fast eye, a steadier eye, and a patient eye. At the family level that is all that is exposed: a chosen market read, looked at over several windows. The series itself, the windows, and how many of them there are, stay inside the model.
From change to a vote: each horizon becomes a bare sign
Each horizon's change is then stripped down to a single fact: is it up or is it down. That is the whole of what survives. The size of the move is discarded. A horizon that is barely positive and a horizon that is strongly positive both cast the same up vote, and the same is true on the downside. This throwing-away of magnitude is intentional.
The method is not measuring how hard the series moves at any one speed; it is asking whether the speeds agree on direction, and reducing each horizon to a bare sign is what makes that question answerable by a simple count.
The agreement score: counting signs, not blending a score
The votes are then added up into a small whole number: the agreement score, a tally of how many horizons point up against how many point down. There is no blended score, no weighted average, no continuous conviction reading hiding underneath. If most horizons point one way the tally leans that way; if they are split it sits near the middle. Everything the method decides flows from this integer count.
The supermajority bar: long, short, or flat
The count is checked against a fixed bar. When the votes reach a supermajority pointing the same way, the method reads a directional state in that direction. A supermajority of up votes reads long. A supermajority of down votes reads short. Anything short of a supermajority in either direction reads flat. There is no in-between conviction.
| Vote tally | What it means | State the method reads |
|---|---|---|
| A supermajority of up votes | The timescales agree on rising direction | Long |
| A supermajority of down votes | The timescales agree on falling direction | Short |
| Short of a supermajority either way | The timescales do not agree | Flat |
A count either clears the integer bar or it does not, which is why the method has crisp states rather than a dial: the output is one of three plain readings.
Why every horizon counts equally
Because the decision is a count, every horizon carries exactly one vote and no horizon's vote is worth more than another's. A single horizon moving hard in one direction cannot drag the method into a position on its own: its strength was already discarded, and one vote is never a supermajority.
The selectivity comes from the arithmetic of the vote rather than from a tuned strength filter that a strong reading could overpower: one loud horizon is outvoted by the quiet agreement of the rest, or by their disagreement.
Symmetric by design
The long and the short conditions are exact mirrors. The same supermajority of up votes that earns a long would, as a supermajority of down votes, earn an equally valid short, and a position that fails to reach the bar in either direction is flat. There is no built-in lean toward going long or toward going short.
An agreed rise and an agreed fall are the same kind of event seen from opposite sides, with the neutral, no-supermajority case sitting symmetrically between them.
The quality mask: a second gate that can keep the method flat
A directional vote is necessary but not always sufficient. A separate quality mask sits on top of the count and can hold the method flat even when a supermajority is present. The mask is a guard: it lets the method decline a technically valid vote when conditions are not suitable for acting on it.
Flat is the behavior, by construction
Standing flat is not a gap in this method; by construction it is the method's most common state. When the timescales disagree, the method does not guess and it does not lean on its strongest horizon. It simply does not trade, treating disagreement as a clear instruction to stay out rather than as a weak or missing signal.
A reader watching this method should expect long stretches of deliberate inactivity, broken by the narrower windows in which the timescales line up. This flat-heaviness is a structural consequence of the agreement bar, not a measured calendar frequency.
Why agreement is the selectivity mechanism
Requiring a supermajority is strictly harder to satisfy than any single timescale acting alone, and that difficulty is the point. By insisting on agreement, the method concentrates its activity into the periods where the read is corroborated across speeds, opens fewer positions, rests more often, and sets an entry condition that one timescale by itself can never trigger.
This selectivity is structural, not measured: it follows from the construction of the vote rather than from a tuned strength filter or any observed result. The selectivity such a method seeks is not in being early or loud but in being selective by construction, and that is a property of the rule, not a claim about returns.
Where this method can fail
A consensus method has characteristic ways of being wrong, and they should be stated plainly. Agreement is not correctness: a supermajority of timescales can line up on the same side and the series can still move the other way. The vote raises the bar for acting; it does not guarantee the direction is right.
- Consensus can arrive late. Because the method waits for several timescales to agree, it can enter after a move is already mature, when the fast and slow reads have only just converged.
- A choppy series can starve it. If the timescales rarely reach a supermajority, the method stays flat for long stretches and does little, which is correct behavior by its own rules but can read as inactivity.
- The quality mask can hold it out of otherwise valid reads. The same guard that declines acting in poor conditions will, by design, veto some supermajority votes that would have worked; that selectivity has a cost as well as a benefit.
Honesty about the family's lifecycle belongs here too. Models built this way are judged through the same staged process as every other family: an idea, a backtest, a held-back walk-forward, and a human review, with the signal pattern replayed on unseen data and scored against an acceptance-gate scorecard. A family-level explainer should not be read as proof that every model in the family has cleared every gate.
How to read its state on the Commentary Desk
On a Stonewell One dashboard a model built this way reads as a sparse three-state lane. For most of the calendar the lane sits at flat, because the timescales are not agreeing or the quality mask is holding the method out. The lane opens to a directional state only in the narrow windows where a supermajority of timescales share a sign and the mask allows it, and it closes back to flat when that supermajority dissolves, when one timescale flips, or when the mask trips.
Informational
Context, no action implied: routine flat stretches where the timescales are not agreeing, a minority of horizons drifting toward one direction while the rest disagree, or the mask holding the method out while the votes are split. None of these is a state change worth flagging; agreement forming is not yet a signal.
Warning
A state change worth flagging: a supermajority of horizons reaching a shared up or down sign with the mask open, so a fresh directional state is active; a held read closing back to flat because the supermajority dissolved or a contributing horizon flipped; or the mask tripping an otherwise valid vote toward flat.
Elevated caution
Flag, but interpret carefully: a directional read that has stayed live across a long stretch, since consensus can arrive after a move is already mature; an unusually choppy period where the timescales rarely agree and the method does little; or any spell where the monitored series appears to be re-basing, which can distort how the horizons line up.
A reader scanning the lane should expect long quiet runs and short active windows, and should read the active windows as moments of corroborated direction rather than as a continuously varying conviction. Every desk note stays a neutral, family-level risk observation, never trading advice.
What this archetype is not
It is not a blended-score method. Nothing here is averaged into a continuous conviction number; the decision is a count of votes against an integer bar. It is not a percentile or positioning read: nothing is ranked into a band and nothing about crowd positioning is consulted, the inputs are direction signs of one series. It is not a reversion or an options method: it does not wait for a stretched extreme to snap back and it does not read an options-market quantity, it bets that an agreed direction persists.
The distinction most worth holding onto
It is not several independent signals agreeing. The horizons are different lookbacks of the same monitored series, not separate instruments. Describing it as several independent indicators agreeing, or as a blended conviction score, would quietly turn it into a different archetype.
What we can and cannot claim
This explainer describes the family mechanism, not the wiring of any one model. Which series is watched, exactly how many horizons are read, how long each lookback is, how large a supermajority is required, and how the quality mask is built are model internals. They are not stated here because they are estimated and re-checked for each model rather than fixed constants the article is withholding.
A method is not a recommendation, and a model built this way is not a guarantee of future results. It is a disciplined way of acting only when several timescales of one series corroborate a direction and standing aside when they do not, and it earns its place only by surviving an idea, a backtest, a held-back walk-forward, and a staged review before it is relied on.
Sources
Method family: multi-horizon momentum consensus, also described as cross-timescale momentum agreement. The family reads the direction of one chosen market series over several lookback horizons, reduces each horizon to a bare up-or-down sign, and takes a position only when a supermajority of those signs agree on direction.Stonewell One documented model-family taxonomy, the catalogue of distinct model archetypes and how each one forms its directional read.Stonewell One model lifecycle and staged review process: how a candidate signal moves from an idea, through a backtest and a held-back walk-forward, to a staged review before it is relied on.