The distance between gross EV and net EV is itself a reading.

Two numbers, one switch

The toggle changes which question the scorecard is answering.

Gross expectancy is computed from hit probabilities and payoff structure alone — it describes the strategy as a mathematical object, independent of who is trading it or where. Net expectancy subtracts recorded friction, which makes it specific to this account, this broker, this instrument set and this holding behaviour. Both are legitimate; neither is a more accurate version of the other. The distinction matters because the toggle silently changes what a comparison means. Comparing this quarter's blended expectancy against last quarter's is only valid if both were read in the same mode, and the failure is easy to commit because the numbers look like the same number. The standing checklist item exists for exactly this reason: confirm the net/gross toggle is intentionally set before the review, not merely wherever it was left. A review that begins without checking the toggle is a review whose first figure has an unknown definition.

Which one decides

Serious review runs on net, because compounding runs on net.

The instruction is unambiguous — use net expectancy for serious review, because friction affects compounding. The reasoning is arithmetic rather than preference. Compounding acts on what actually lands in the account, so a gross figure describes a growth path that does not exist. The distortion is also not uniform: friction is roughly proportional to trade count and holding duration, while gross expectancy is proportional to edge per trade, so a high-frequency branch and a low-frequency branch with identical gross expectancy do not have identical net expectancy, and the difference is not small. This is the mechanism by which a branch can look like the best performer on the tab and be the worst contributor to the curve. Every ranking, every allocation argument, and every profile comparison that is going to inform a decision belongs in net.

The gap as diagnostic

Net close to gross is a health reading; net far below gross is a finding.

Because the two numbers are computed from the same trades, their separation is a clean measurement of what friction is costing this specific structure. The good sign is net remaining close to gross, which says the edge survives the cost of harvesting it. The warning sign is gross positive with net marginal or negative, which says the strategy is real and the implementation is eating it — a genuinely different problem from having no edge, and one with a genuinely different repair. An operator who only ever reads net will see a marginal number and reasonably conclude the strategy is weak, then spend a quarter adjusting entry logic that was never the problem. An operator who only ever reads gross will see a healthy number and be unable to explain why the account is flat. The diagnostic value lives in holding both and reading the distance, which is why the toggle exists at all rather than the scorecard simply being built in net.

Where it changes decisions

A branch ranking computed in gross can invert under the toggle.

This is the practical consequence and the reason the toggle is not cosmetic. Suppose two branches show comparable gross expectancy, and one of them achieves it across roughly three times the trade count. In gross they are peers and the allocation argument is a coin flip decided by preference. In net they are not peers at all, because the high-count branch has paid three times the friction to arrive at the same headline. Any weighting decision, any profile comparison, any argument about which branch has earned an increase, is being made on a ranking that the toggle can reorder. The same reasoning applies to duration: a branch that holds through financing costs carries a friction load a fast branch does not, and that load is invisible in gross. The rule that follows is narrow and firm — comparisons that will change an allocation are made in net, and if a decision would flip depending on the toggle, it is not yet a decision.

When gross is the right read

Gross has one honest job, and it is not decision-making.

Gross expectancy is the correct read for theoretical comparison, which is a real and useful activity as long as it is labelled. Judging whether a structural change to the payoff — a different partial rule, a different target convention — improved the underlying mathematics is a question about the strategy, and mixing friction into it introduces a variable that has nothing to do with the change being tested. The same holds for comparing this account's structure against a published result or a simulation output, neither of which carries this broker's costs. The boundary is simply stated: gross answers questions about the strategy, net answers questions about the account, and the failure is always the same one — a gross figure quoted in a context where the reader will act on it as though it were net.

• Serious review, allocation, and profile comparison: net, without exception.
• Structural change tests and external comparison: gross, and say so out loud.
• The toggle is confirmed at the start of a review, never assumed from last time.

The key idea

Two numbers exist because one number would hide the more useful finding.

A scorecard built only in net would be defensible and would still be worse, because it would collapse two distinguishable failures — no edge, and an edge consumed by implementation — into one indistinguishable reading. A scorecard built only in gross would be a marketing document. Keeping both, and treating their separation as a first-class output rather than a rounding difference, is what allows the operator to answer the question that actually determines what to do next: is this structure failing to produce edge, or is it producing edge and failing to keep it? Those two situations look identical in the account balance and have almost nothing in common as repairs.