Goal Prevention

A team's goals-against is the product of three things, so we split it into three - the framework is Alan Ryder's 2004 Shot Quality paper, run on our own expected goals:

• Shot quality - how dangerous the shots they allow are (expected goals per shot), vs the league.
• Volume - how many shots they allow, vs the league.
• Goaltending - goals allowed vs expected; below 1 means the goalies stop more than the chances deserve.

Multiply the three and you get the GA index - goals-against vs the league average. Shot quality × volume alone is the defense (everything but the goalie). The driver tag flags whichever side - defense or goaltending - is furthest from average.

1.00 = league average for every index; lower is better. Thresholds: below 0.97 (better) · near average · above 1.03 (worse). A 0.90 means ~10% fewer goals (or shots, or chance) than average.

Expand any team to see its goalies. GSAx= goals saved above expected (a goalie's xG faced minus goals allowed); positive is good. /100 puts it per 100 shots so a backup and a starter compare fairly (≈ +1.0/100 is elite, −1.0 poor). Regular season, 2010-11 on.

Goaltending is famously noisy: our own research finds a single season of save % or GSAx is mostly luck - a goalie needs ~5,000 shots (about three starter seasons) before the number is mostly signal. So True /100 is the honest best estimate - small samples regressed toward the era average - and the confidence dot (● low · ● moderate · ● high) plus the hover band (95% confidence on SV%) show how much to trust each line.