Branch Logic — Blends & Attribution
The blend can hide the truth. Attribution finds it.
Layer 01 — The blend math
Weighted expectancy by live profile.
The Weekly Scorecard blends branch EV by the active profile weights — default 45 / 33.75 / 11.25 / 10 across Normal, Trend Partial, Trend No-Partial, and Overflow. Layers of blend exist: Trend Blend EV, Normal+Trend EV, and All-Blended EV, each answering a different composition question.
Layer 02 — Attribution layers
Where did the R actually come from?
The five-layer attribution model from the variant matrix: CP3 main EV is the truth engine; variant attribution shows where EV came from; checkpoint hit rates explain why it behaved that way; execution quality shows whether it's sustainable; SDE/regime review decides whether it's improving or decaying.
- Scenario_Profiles compares the active weight profile against alternates — was another blend structurally superior?
- Profile changes require persistence over multiple weeks and months, never one flattering sample.
How MARS uses this
MARS never trusts a single blended figure. Weekly branch probabilities produce a separate EV per branch, weighted by the live profile (45 / 33.75 / 11.25 / 10 by default) into the blend. Reviews read the decomposition first: which branch produced the edge, which one leaked it, and whether the weights still match the evidence.
How it benefits you
You stop repairing the wrong thing. When results dip, the decomposition shows whether the whole system weakened or one branch is dragging - so you fix a branch instead of overhauling a working system, and you spot a single strong branch masking weakness everywhere else.
Branch-level EV decomposition with blend weights. The blended number alone would hide exactly this picture.
Layer 03 — Attribution cadence
Weights move on review cadence, never mid-week.
Variant attribution runs continuously, but blend weights only move at scheduled review — a deliberate lag that keeps allocation decisions out of reach of a hot Tuesday or a bruising Thursday. The rule costs a little responsiveness and buys something worth far more: weight changes that always have a full week of settled evidence behind them.
In practice
The MARS authority stack. Authority flows down; evidence flows up; nothing lower may override anything above it.
The governing idea
Connected inside MARS
This module doesn't work alone.
Go deeper
Operator briefs on this territory.
Deep dive — 01
When to trust the blend, and when the blend is the thing lying to you.
Every question has a level. Answers fetched from the wrong level are how a healthy aggregate hides a broken branch.
Read the full brief →
Deep dive — 02
How four branch EVs become one system EV — and why weights are the profile.
Branch EV → Trend Blend → All-Blended, weighted by the active profile. The weights are the strategy's mix, stated as numbers.
Read the full brief →
Deep dive — 03
Changing the weights: the slowest decision in the system, deliberately.
Persistence across weeks and months, never one flattering sample. And never mid-week.
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.