Skip to content

Field guide / Method

Methodology · night field register

How the breeding results are computed

Tamedex does not ship a third-party pre-computed combination table. Every result is computed by our own script from combi ranks and the special-combination list; where two candidates tie, the order was calibrated against paldb.gg results (explained below). That keeps each result explainable, versioned and tested. Try any pairing below, then read how the number is produced.

01 · breeding notes

Pair the records.

Pair two records to read the route, or reverse-search a target’s cleanest lineage. New here? Read the setup guide →

Plan a chain ↗
Parent A
Parent B
loading…

Choose two records

Order does not affect the child

interface study · A + B and B + A give the same child · computed in-browser from Tamedex’s breeding table; see methodology for sample-check scope.

The formula, illustrated

Each Pal has a hidden breeding rank (its combi rank). For an ordinary pairing the child rank is the rounded midpoint of the two parents:

child_rank = floor((rankA + rankB + 1) / 2)

The child is the breeding-eligible Pal whose combi rank sits closest to that midpoint — not the average of the two Pals themselves:

lower rankhigher rankparent Aparent Bchild_rankclosest → child
Two parents define a midpoint on the hidden rank axis. Tamedex snaps that midpoint to the nearest breeding-eligible Pal — the moss marker — which becomes the child.

Worked example — a pairing from our breeding table. The calculator gives the same child in either parent order; see the sample-check scope below.

The pipeline

Every pairing runs through the same deterministic steps:

01

Read the ranks

Look up each parent’s hidden combi rank.

02

Rounded midpoint

Apply floor((A+B+1)/2) to get the target rank.

03

Nearest & tie-break

Pick the closest eligible Pal; break exact ties by a calibrated priority.

04

Special override

Unique combinations replace the formula result entirely.

Because the child is a deterministic function of two ranks, A + B and B + A always give the same child, and the calculator runs entirely in your browser from a small versioned snapshot.

Tie-break calibration

Published combi ranks are rounded to the nearest ten, so many pairings land exactly between two candidates. The real game resolves these with an internal priority order. We derive that order from paldb.gg’s game-derived child-to-parents data and encode it as a per-Pal tie priority. Because the same source is used afterwards for the check below, a match on those tie cases shows that our table is consistent with paldb.gg, not that it was confirmed independently in the game. Legendaries are never an ordinary result, but most (26 in this table) breed true with a second of the same Pal. Astralym and Panthalus have no breeding recipe in the data.

Verification

0 mismatchesin 3,487 reference parent pairs for 40 sample target Pals, checked on 2026-10-26 against paldb.gg. That is about 10.3% of the 33,957 rows in our computed table.

What this shows

  • For the 40 sample targets (ordinary results, variants, special combinations and legendary self-pairs), our table lists exactly the same parent pairs as the reference: no missing pairs, no extra pairs.
  • The deploy command runs this check on both the computed table and the calculator data file sent to your browser. If either file is missing or any sample target changes, the deploy stops.

What it does not show

  • It does not cover every combination. About nine in ten rows of the table have not been compared with any reference.
  • It is not independent of the calibration: the tie-break priority came from the same source (see above).
  • It is not an in-game test. We have not bred these pairs in the game, and a second independent source is not yet used.

Found a result that looks wrong? Tell us the two parent Pals and the game version, and we will check it. Game version covered: Palworld 1.0.0 (game build 2026-07-15; build reference 1.0.0.100427-2026-07-15.1); data version 1.0.0, data captured 2026-08-18.

How we source & verify

Breeding logic
Computed by our own scriptcombi rank + specials + calibrated tie-break
Reference check
paldb.gg sample40 Pals, 3,487 pairs, 0 mismatches (2026-10-26)
Where a fact is missing
Shown as unavailablenever invented
Artwork
Site-made illustrationsmost avatar records are drafts; review is ongoing

See the sources and IP policy pages for full provenance and artwork boundaries.

Tamedex is an unofficial field register. See the sources page for data provenance and the IP policy for the ongoing illustration review.