EQL·Parse
esenfrdept

What it measures, and what it does not

A combat meter is a count over lines of text. What decides whether it is any good is what gets counted, what gets inferred and what cannot be known — and that every figure says which of the three it is.

Measured, inferred or declared — never mixed

Every figure carries where it came from. "Measured" is what the log counts; "inferred" is what follows from it by a written rule; "declared" is what you told it. A missing datum dressed as a measured one is worse than not having the column.

The shape of a hit, not its min and max

A move’s minimum and maximum are the only two figures in its row that a single freak hit can move. Measured on a real log, an attack with 557 samples gives 3 / 25 / 60: the minimum is 0.12× the median. What is shown is the median with p10 and p90, which answer the same question and survive a freak hit.

Idle time is measured against your weapon, not against zero

Counting every second without damage as idle blames your gear for something you did not do: with a three-second weapon, two seconds in three carry no damage by construction. And missing is attacking — 31.7% of the seconds with an attack are misses only. It is measured against your own cadence, inferred from your own gaps.

What the log does not say is not said

The EQL log never writes anyone’s health. So there are no pressure bars, no "you nearly died", no percentages that look like a health bar: 3,000 taken is a tragedy with 3,100 health and a stroll with 12,000, and the app does not know which.

And order is data, spacing is not

Within a second, lines are in the order they happened: checked with a prediction that could have failed — the killing blow must appear before the death line — and it came out 952 of 952. The spacing inside that second does not exist in the log, so the replay does not draw it.