Three composites, three questions — and why one would not do.

The three questions

Total system, core plan, trend engine.

All Blended asks what the account experienced. It combines all four primitive branches and is the only view that corresponds to the equity curve, which makes it the honest total and the weakest diagnostic — it reports the outcome of the mix without separating the mix from the components. Normal + Trend asks how the core plan behaved, combining Normal, Trend Partial, and Trend No-Partial while deliberately excluding Overflow. Trend Blend asks how the trend engine behaved, combining the two trend branches and excluding Normal. Each is a legitimate reading of 'the system', and an operator who does not state which one they mean has produced a sentence rather than a finding.

The Overflow exclusion

Normal + Trend exists because supplemental flow distorts.

Overflow is supplemental higher-timeframe opportunity flow, and the manual is direct that it must be controlled because its volume and cost characteristics can distort expectancy. That distortion is not a flaw in Overflow — it is a property of what supplemental flow is. It runs at a different frequency, carries a different cost profile, and is deliberately capped at a small share of deployment with the tightest behavioural scrutiny in the stack. Folding it into every composite would mean that any reading of the plan is really a reading of the plan plus a supplement that was never intended to be judged on the same terms. Normal + Trend removes it so that the question 'is the core plan working' can actually be asked. When Normal + Trend and All Blended diverge, the divergence is itself the finding: it locates the difference in Overflow, which is precisely the branch whose contribution most needs watching.

The Normal exclusion

Trend Blend isolates the engine with the fat tail.

Trend Partial and Trend No-Partial share an exit architecture family and a dependence on continuation; Normal does not. Blending them together isolates the part of the system whose returns are geometric and whose variance is highest, and that isolation matters because trend behaviour degrades differently from baseline behaviour. A trend engine losing efficiency looks like a run of trades that were structurally correct and failed to extend — a pattern that Normal's steadier profile will substantially mute inside a wider composite. Trend Blend is the view in which that degradation is visible early enough to be a diagnosis rather than a post-mortem.

• Trend Blend is the right entity when the question involves trailing behaviour, continuation, or fat-tail capture.
• Normal is the right entity when the question is about baseline edge quality — it is the simulation centre for a reason.
• Reading Trend Blend against Normal separates 'the market changed' from 'this exit architecture stopped working'.

How the three are read together

The differences between composites carry more information than any one of them.

The practical technique is not to pick a composite but to read the gaps. All Blended constructive while Normal + Trend is weak means the total is being carried by Overflow, which is supplemental flow doing structural work it was never sized to do. Normal + Trend healthy while Trend Blend degrades means the core plan is intact and the trend engine specifically is not — a branch-level problem wearing a composite's clothes. All three moving together means the finding is systemic rather than local, which is both the clearest and the most serious of the readings. None of these are available from a single composite, which is the entire argument for having three.

The key idea

A view without an exclusion is a summary, not a diagnostic.

Every useful analytical view is defined by what it refuses to include, because the exclusion is what isolates a question. All Blended excludes nothing and therefore isolates nothing — it is the number the account already knows. The other two composites earn their place by leaving something out on purpose, and the thing they leave out is the thing they are asking about.