Most systems grade two-thirds of the game. KaNeXT grades the third phase like a phase.
Special teams is a real third phase, not a footnote. The kicking, return, and coverage units roll up into a Team Special Teams KR that combines with offense and defense to make the team number. The specialist is on the one universal scale, so a kicker and a left tackle are comparable numbers. And every position player carries a special-teams contribution term folded into his OVERALL, small for most starters and decisive for the coverage-and-return core the roster is built on.
A game has three phases. So the team number has three phases.
A composite team. Team KR is not two phases and a footnote. It rolls up three phase KRs at a research-grounded weight, and the third phase moves the team number like the other two.
The split comes from the DVOA-to-wins research, moderated per level, and reflects that offense is the most predictive phase, not that special teams does not matter. It is the third of the game on the same number as the other two. Grade the whole game, or grade two-thirds of it and call the difference luck.
Illustrative on the real three-phase Team KR roll-up (snap-weighted, positional-value-weighted, the split grounded in the research). Composite team, demonstration figures.
A kicker and a left tackle are comparable numbers.
A composite precision kicker, scored on the Specialist cluster and landing on the same universal scale as every position player. No additive special-teams term applies to him, because his entire value already lives in the phase.
The kicker, punter, and long snapper are the contained sub-engine, and the whole phase lands on the one ruler, so you can rank a punter against a guard and a returner against a slot when choosing who to keep. One scale, or the specialist is a number you can never compare to anything.
Illustrative on the real Specialist cluster and sub-engine (K/P/LS on the one scale, over-expected and operation bands, the returner routing). Composite specialist, demonstration figures.
Where the bottom of the roster earns its value.
Every position player carries a bounded special-teams contribution term (0 to +2.0 KR), folded into his OVERALL after the scheme reading and aggregated into Team Special Teams KR by his special-teams snap share. Small for most starters. Often the whole roster case for the coverage-and-return core.
A starter model rates the WR2 and forgets the ace, then cannot explain why the ace has a roster spot and a fourth-round pick does not. This term is the explanation, the coverage-and-return core is real team value folded into the player and rolled into the phase, and the engine flags the roster-bubble player for whom special teams is the whole case. The phase nobody rates is the phase that makes the roster.
Illustrative on the real special-teams contribution term (0 to +2.0, folded into OVERALL, aggregated into Team Special Teams KR by snap share, decisive for the coverage-and-return core). Composite players, demonstration figures.
Special teams is a third of the game and almost every rating system rounds it to zero, treats the specialist as an uncomparable curiosity, and lets the coverage-and-return core disappear because it never starts. KaNeXT does the opposite on all three counts: the phase rolls up into Team KR at a real research-grounded weight so a team is graded on the whole game and not two-thirds of it, the specialist lands on the one universal scale so a kicker and a left tackle are the same kind of number and a punter can be ranked against a guard, and every position player carries a special-teams contribution term small for the starter and decisive for the ace so the value that decides the bottom of the roster is priced and named instead of felt. A phase you refuse to grade is a phase you are guessing at, and in a close game the guess is the margin.
Grade the third phase. It is on the same scale as the other two.
Special Teams puts the whole third of the game on the one KR scale, the phase in Team KR, the specialist comparable to any position, and the coverage-and-return core priced where the roster is decided.