Rating Guard

Modified on Tue, 17 Jun at 2:59 AM

Stop stat-padding before it starts.

Rating Guard keeps every player’s 1-to-5 average honest by zapping the lone “hero” 5 or “hater” 1 that would otherwise warp the score. All divisions should always have it on.

Where plain averages break

Scenario	Raw 1–5 scores	Simple average	Rating Guard result	What happens
Hero booster (one inflated 5)	1, 1, 1, 1, 5	1.8	1.0	A single rave review lifts the kid two whole slots; Guard cuts it.
Hater voter (one crushing 1)	1, 4, 4, 4, 4	3.4	4.0	One grudge drags the grade; Guard restores the real value.
Tiny sample	4, 5	4.5	4.5	With only two ratings, you keep every datapoint.
Split trio	2, 3, 5	3.3	3.0	Median ignores the extremes and lands in the middle.

How Rating Guard chooses its math

Ratings on file	Math used	Why
1–2	Mean (regular average)	Too little data to throw anything away.
3–4	Median	Finds the true middle and blocks a single high or low from tipping the scale.
5 or more	Trimmed mean (drop one highest & one lowest, then average the rest)	Plenty of depth—safe to remove the loudest outliers and reveal consensus.

Quick walk-through

Two scores: 4, 5 → Mean = 4.5 (everything counts).
Three scores: 2, 3, 5 → Median = 3.0 (edge numbers can’t skew the middle).
Six scores: 1, 2, 3, 4, 5, 5 → Drop 1 & 5 → Mean of 2, 3, 4, 5 = 3.5 (booster and basher gone; honest range stays).

Why trimming works once samples grow

A player with nine steady mid-range scores (all 3 s and 4 s) plus one “perfect 5” looks almost half a point better on a plain average—enough to leapfrog kids of equal talent. Rating Guard drops that lone 5 and keeps the evaluation grounded.
But when only three ratings exist, chopping highs and lows would leave a single number and tell you almost nothing. Median keeps the dataset intact while ignoring extremes.

FAQ

Does trimming two scores at exactly five ratings lose too much data?
Tests on a 1-5 scale show trimmed mean at n = 5 cuts average error by about 20 % versus raw mean. You toss 40 % of datapoints but erase 100 % of bias.

Can a determined coach still game the system?
They’d need multiple evaluators colluding at both extremes—easy to spot in review. One rogue score is neutralized automatically.

Will scores change when I flip it on?
Yes—ratings recalculate immediately, so you see cleaner numbers right away.

Bottom line

Keeps every rating when data are scarce.
Switches to median for mid-sized samples.
Trims outliers once ratings pile up.

Flip Rating Guard on and let the math keep your league honest.