Your clear-time inputs decide how useful the farming optimizer becomes. A route that looks perfect on paper can be the wrong choice if your build clears it slowly, misses moving enemies, pauses at bosses, or fails one run out of every few attempts. When you calibrate the route, you turn a static database view into a player-specific plan.

Record three clean 1-1 clears and three to five clears on the highest stage you can repeat. Use the median, then rerun the numbers after a weapon, rune, or pet change. The optimizer needs the account you are playing today, not the screenshot from one lucky clear.

Why 1-1 and a high stable route both matter

Run Stage 1-1 to set a low-pressure baseline. You can see how quickly your current hero setup moves through a simple route when enemies do not resist your damage or interrupt your pattern. Use that number so the optimizer does not treat every account as if it has the same animation speed, pet support, rune setup, and loading rhythm.

Use your highest reliable route to test the other side of the account. You need to know how your build behaves when enemies survive longer and the route exposes friction. If your 1-1 time is fast but your highest clear is slow, a middle stage may beat the late route. If both numbers are strong, later stages can rank higher because reward gains survive the clear-time drag.

Use median samples, not record runs

A good calibration sample is boring. Run the stage at least three times with the same build, ignore a run with obvious input mistakes, and use the median of the remaining clean clears. The fastest run is tempting, but it usually describes the best possible loop, not the loop you will repeat for thirty minutes while checking drops, swapping menus, or recovering from a bad spawn.

If a stage has uneven boss behavior, record more than three runs. Five samples are usually enough for route planning because the optimizer is not trying to prove a laboratory speedrun time. It is trying to help you decide whether a route is obviously better, roughly tied, or clearly not worth the friction.

  • Keep hero, gear, pet, rune, and skill setup unchanged during the sample.
  • Do not mix manual active play with semi-idle clears in the same calibration set.
  • Rerun calibration after a major weapon upgrade, rune swap, or pet change.

How to read close rankings

When two routes are separated by only a small hourly difference, the ranking should not be treated as a command. A 5 percent advantage can disappear if one stage has more movement, a boss that stalls your build, or a reward pattern that forces you to check boxes more often. In that case, choose the route you can repeat cleanly and review the ranking again after you have more live samples.

Use the optimizer to cut bad routes first. A late route with higher rewards loses if it takes twice as long. When the top three routes sit close together, treat them as a shortlist. Run each one, note failure rate and side drops, then keep the route you can repeat.

For example, if one route is projected to earn slightly more Gold per hour but forces you to pause for boss timing, and another route is a few percent lower but can be repeated smoothly, the smoother route may be the better real farm. Treat the close-ranking zone as a testing queue. Run each candidate with the same build, record a small batch of clears, and update the optimizer when your live median no longer matches the old input.

Update and correction notes

After a patch, rerun the two timing checks before you trust an old route. Stage rewards, chest behavior, server timing, and skill interactions can change. Check the visible data version and page update date before you spend a long session on yesterday's route.

If a route appears wrong, send a useful correction report rather than only saying the table is bad. Include hero level, 1-1 time, high-route time, stage name, target mode, game version, and what you observed in the live game. That information lets the maintainer separate a data issue from a build-specific timing issue.

Keep one note for the numbers you last trusted. A simple date, hero, build, and two median clear times are enough. Without that note, it is easy to forget that the optimizer result was based on an older weapon, a different pet, or a patch that changed server timing.

Add device or network notes when loading time changes the sample.