FA-85131 / Fantasy sports scoring / Open access

Empty-net goals charged to the pulled goalie · case 01

A goalie who allowed one goal is charged three after two late empty-netters.

Verified by executionVariant 1 · 7 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

Goalie goals against are taken from the team total including empty-net goals.

VERIFIED REPAIR

Subtract every empty-net goal from the team total.

Unsuccessful approach: Subtracting at most one empty-net goal still charges the second.

Case contract

Score a hockey goalie in tenths. Goals against charged to the goalie exclude empty-net goals scored on his team. shots_faced counts only shots the goalie faced, so saves = shots_faced - goalie goals against. Save 0.2, goal against -2, win 4, overtime loss 1, regulation loss 0. A shutout (+3) needs zero goalie goals against and at least 3600 seconds on ice; the decision does not matter.

Why this case matters

Goalie scoring depends on attribution of empty-net goals and on playing-time conditions for shutouts.

1 / The failure

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(g):
    ga = g['team_ga']
    saves = g['shots_faced'] - ga
    pts = saves * 2 - ga * 20
    pts += {'W': 40, 'OTL': 10}.get(g['decision'], 0)
    if ga == 0 and g['toi_sec'] >= 3600:
        pts += 30
    return {'saves': saves, 'points': pts}
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
fixtures = [[('regression: empty-net attribution',
   [{'decision': 'OTL', 'empty_net_ga': 1, 'shots_faced': 26, 'team_ga': 1, 'toi_sec': 3540}],
   {'points': 62, 'saves': 26}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'W', 'empty_net_ga': 2, 'shots_faced': 23, 'team_ga': 5, 'toi_sec': 2400}],
   {'points': 20, 'saves': 20}),
  ('second regression',
   [{'decision': None, 'empty_net_ga': 1, 'shots_faced': 16, 'team_ga': 1, 'toi_sec': 2400}],
   {'points': 32, 'saves': 16}),
  ('normal control 1',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 7, 'team_ga': 2, 'toi_sec': 3900}],
   {'points': 10, 'saves': 5}),
  ('normal control 2',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 13, 'team_ga': 0, 'toi_sec': 2675}],
   {'points': 36, 'saves': 13}),
  ('normal control 3',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 26, 'team_ga': 3, 'toi_sec': 3900}],
   {'points': -4, 'saves': 23}),
  ('normal control 4',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 32, 'team_ga': 1, 'toi_sec': 3540}],
   {'points': 42, 'saves': 31})],
 [('regression: empty-net attribution',
   [{'decision': 'W', 'empty_net_ga': 1, 'shots_faced': 34, 'team_ga': 2, 'toi_sec': 3540}],
   {'points': 86, 'saves': 33}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'OTL', 'empty_net_ga': 2, 'shots_faced': 29, 'team_ga': 4, 'toi_sec': 3900}],
   {'points': 24, 'saves': 27}),
  ('second regression',
   [{'decision': 'L', 'empty_net_ga': 1, 'shots_faced': 18, 'team_ga': 3, 'toi_sec': 3900}],
   {'points': -8, 'saves': 16}),
  ('normal control 1',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 2, 'team_ga': 1, 'toi_sec': 3600}],
   {'points': -18, 'saves': 1}),
  ('normal control 2',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 8, 'team_ga': 1, 'toi_sec': 3599}],
   {'points': 4, 'saves': 7}),
  ('normal control 3',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 16, 'team_ga': 2, 'toi_sec': 3600}],
   {'points': -2, 'saves': 14}),
  ('normal control 4',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 3, 'team_ga': 1, 'toi_sec': 3540}],
   {'points': -6, 'saves': 2})],
 [('regression: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 1, 'shots_faced': 38, 'team_ga': 3, 'toi_sec': 2400}],
   {'points': 32, 'saves': 36}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 39, 'team_ga': 2, 'toi_sec': 3600}],
   {'points': 108, 'saves': 39}),
  ('second regression',
   [{'decision': 'L', 'empty_net_ga': 1, 'shots_faced': 39, 'team_ga': 2, 'toi_sec': 3599}],
   {'points': 56, 'saves': 38}),
  ('normal control 1',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 36, 'team_ga': 2, 'toi_sec': 2400}],
   {'points': 28, 'saves': 34}),
  ('normal control 2',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 20, 'team_ga': 5, 'toi_sec': 2400}],
   {'points': -30, 'saves': 15}),
  ('normal control 3',
   [{'decision': 'L', 'empty_net_ga': 0, 'shots_faced': 17, 'team_ga': 2, 'toi_sec': 3600}],
   {'points': -10, 'saves': 15}),
  ('normal control 4',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 40, 'team_ga': 0, 'toi_sec': 3600}],
   {'points': 150, 'saves': 40})],
 [('regression: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 37, 'team_ga': 2, 'toi_sec': 2400}],
   {'points': 74, 'saves': 37}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 1, 'team_ga': 2, 'toi_sec': 3540}],
   {'points': 2, 'saves': 1}),
  ('second regression',
   [{'decision': None, 'empty_net_ga': 2, 'shots_faced': 7, 'team_ga': 2, 'toi_sec': 3900}],
   {'points': 44, 'saves': 7}),
  ('normal control 1',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 1, 'team_ga': 1, 'toi_sec': 3600}],
   {'points': -20, 'saves': 0}),
  ('normal control 2',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 38, 'team_ga': 5, 'toi_sec': 3540}],
   {'points': 6, 'saves': 33}),
  ('normal control 3',
   [{'decision': 'L', 'empty_net_ga': 0, 'shots_faced': 15, 'team_ga': 0, 'toi_sec': 3540}],
   {'points': 30, 'saves': 15}),
  ('normal control 4',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 11, 'team_ga': 0, 'toi_sec': 2400}],
   {'points': 32, 'saves': 11})],
 [('regression: empty-net attribution',
   [{'decision': 'W', 'empty_net_ga': 1, 'shots_faced': 40, 'team_ga': 4, 'toi_sec': 3599}],
   {'points': 54, 'saves': 37}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 33, 'team_ga': 3, 'toi_sec': 2400}],
   {'points': 44, 'saves': 32}),
  ('second regression',
   [{'decision': 'OTL', 'empty_net_ga': 1, 'shots_faced': 6, 'team_ga': 3, 'toi_sec': 3600}],
   {'points': -22, 'saves': 4}),
  ('normal control 1',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 9, 'team_ga': 0, 'toi_sec': 3600}],
   {'points': 88, 'saves': 9}),
  ('normal control 2',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 25, 'team_ga': 0, 'toi_sec': 3599}],
   {'points': 50, 'saves': 25}),
  ('normal control 3',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 17, 'team_ga': 5, 'toi_sec': 2400}],
   {'points': -66, 'saves': 12}),
  ('normal control 4',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 6, 'team_ga': 1, 'toi_sec': 3599}],
   {'points': 30, 'saves': 5})]]
for label, args, expected in fixtures[N-1]:
    check(label, solve(*args), expected)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
regression: empty-net attribution	`{'points': 40, 'saves': 25}`	`{'points': 62, 'saves': 26}`	Failed
partial repair probe: empty-net attribution	`{'points': -24, 'saves': 18}`	`{'points': 20, 'saves': 20}`	Failed
second regression	`{'points': 10, 'saves': 15}`	`{'points': 32, 'saves': 16}`	Failed
normal control 1	`{'points': 10, 'saves': 5}`	`{'points': 10, 'saves': 5}`	Passed
normal control 2	`{'points': 36, 'saves': 13}`	`{'points': 36, 'saves': 13}`	Passed
normal control 3	`{'points': -4, 'saves': 23}`	`{'points': -4, 'saves': 23}`	Passed
normal control 4	`{'points': 42, 'saves': 31}`	`{'points': 42, 'saves': 31}`	Passed

SHA-256 / 27d4d82f117197741bfd8fc75020d8053d8549e244536b8a1bc449c52fd17889

2 / The unsuccessful fix

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(g):
    ga = g['team_ga'] - (1 if g['empty_net_ga'] else 0)
    saves = g['shots_faced'] - ga
    pts = saves * 2 - ga * 20
    pts += {'W': 40, 'OTL': 10}.get(g['decision'], 0)
    if ga == 0 and g['toi_sec'] >= 3600:
        pts += 30
    return {'saves': saves, 'points': pts}
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
fixtures = [[('regression: empty-net attribution',
   [{'decision': 'OTL', 'empty_net_ga': 1, 'shots_faced': 26, 'team_ga': 1, 'toi_sec': 3540}],
   {'points': 62, 'saves': 26}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'W', 'empty_net_ga': 2, 'shots_faced': 23, 'team_ga': 5, 'toi_sec': 2400}],
   {'points': 20, 'saves': 20}),
  ('second regression',
   [{'decision': None, 'empty_net_ga': 1, 'shots_faced': 16, 'team_ga': 1, 'toi_sec': 2400}],
   {'points': 32, 'saves': 16}),
  ('normal control 1',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 7, 'team_ga': 2, 'toi_sec': 3900}],
   {'points': 10, 'saves': 5}),
  ('normal control 2',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 13, 'team_ga': 0, 'toi_sec': 2675}],
   {'points': 36, 'saves': 13}),
  ('normal control 3',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 26, 'team_ga': 3, 'toi_sec': 3900}],
   {'points': -4, 'saves': 23}),
  ('normal control 4',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 32, 'team_ga': 1, 'toi_sec': 3540}],
   {'points': 42, 'saves': 31})],
 [('regression: empty-net attribution',
   [{'decision': 'W', 'empty_net_ga': 1, 'shots_faced': 34, 'team_ga': 2, 'toi_sec': 3540}],
   {'points': 86, 'saves': 33}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'OTL', 'empty_net_ga': 2, 'shots_faced': 29, 'team_ga': 4, 'toi_sec': 3900}],
   {'points': 24, 'saves': 27}),
  ('second regression',
   [{'decision': 'L', 'empty_net_ga': 1, 'shots_faced': 18, 'team_ga': 3, 'toi_sec': 3900}],
   {'points': -8, 'saves': 16}),
  ('normal control 1',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 2, 'team_ga': 1, 'toi_sec': 3600}],
   {'points': -18, 'saves': 1}),
  ('normal control 2',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 8, 'team_ga': 1, 'toi_sec': 3599}],
   {'points': 4, 'saves': 7}),
  ('normal control 3',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 16, 'team_ga': 2, 'toi_sec': 3600}],
   {'points': -2, 'saves': 14}),
  ('normal control 4',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 3, 'team_ga': 1, 'toi_sec': 3540}],
   {'points': -6, 'saves': 2})],
 [('regression: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 1, 'shots_faced': 38, 'team_ga': 3, 'toi_sec': 2400}],
   {'points': 32, 'saves': 36}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 39, 'team_ga': 2, 'toi_sec': 3600}],
   {'points': 108, 'saves': 39}),
  ('second regression',
   [{'decision': 'L', 'empty_net_ga': 1, 'shots_faced': 39, 'team_ga': 2, 'toi_sec': 3599}],
   {'points': 56, 'saves': 38}),
  ('normal control 1',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 36, 'team_ga': 2, 'toi_sec': 2400}],
   {'points': 28, 'saves': 34}),
  ('normal control 2',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 20, 'team_ga': 5, 'toi_sec': 2400}],
   {'points': -30, 'saves': 15}),
  ('normal control 3',
   [{'decision': 'L', 'empty_net_ga': 0, 'shots_faced': 17, 'team_ga': 2, 'toi_sec': 3600}],
   {'points': -10, 'saves': 15}),
  ('normal control 4',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 40, 'team_ga': 0, 'toi_sec': 3600}],
   {'points': 150, 'saves': 40})],
 [('regression: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 37, 'team_ga': 2, 'toi_sec': 2400}],
   {'points': 74, 'saves': 37}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 1, 'team_ga': 2, 'toi_sec': 3540}],
   {'points': 2, 'saves': 1}),
  ('second regression',
   [{'decision': None, 'empty_net_ga': 2, 'shots_faced': 7, 'team_ga': 2, 'toi_sec': 3900}],
   {'points': 44, 'saves': 7}),
  ('normal control 1',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 1, 'team_ga': 1, 'toi_sec': 3600}],
   {'points': -20, 'saves': 0}),
  ('normal control 2',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 38, 'team_ga': 5, 'toi_sec': 3540}],
   {'points': 6, 'saves': 33}),
  ('normal control 3',
   [{'decision': 'L', 'empty_net_ga': 0, 'shots_faced': 15, 'team_ga': 0, 'toi_sec': 3540}],
   {'points': 30, 'saves': 15}),
  ('normal control 4',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 11, 'team_ga': 0, 'toi_sec': 2400}],
   {'points': 32, 'saves': 11})],
 [('regression: empty-net attribution',
   [{'decision': 'W', 'empty_net_ga': 1, 'shots_faced': 40, 'team_ga': 4, 'toi_sec': 3599}],
   {'points': 54, 'saves': 37}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 33, 'team_ga': 3, 'toi_sec': 2400}],
   {'points': 44, 'saves': 32}),
  ('second regression',
   [{'decision': 'OTL', 'empty_net_ga': 1, 'shots_faced': 6, 'team_ga': 3, 'toi_sec': 3600}],
   {'points': -22, 'saves': 4}),
  ('normal control 1',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 9, 'team_ga': 0, 'toi_sec': 3600}],
   {'points': 88, 'saves': 9}),
  ('normal control 2',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 25, 'team_ga': 0, 'toi_sec': 3599}],
   {'points': 50, 'saves': 25}),
  ('normal control 3',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 17, 'team_ga': 5, 'toi_sec': 2400}],
   {'points': -66, 'saves': 12}),
  ('normal control 4',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 6, 'team_ga': 1, 'toi_sec': 3599}],
   {'points': 30, 'saves': 5})]]
for label, args, expected in fixtures[N-1]:
    check(label, solve(*args), expected)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
regression: empty-net attribution	`{'points': 62, 'saves': 26}`	`{'points': 62, 'saves': 26}`	Passed
partial repair probe: empty-net attribution	`{'points': -2, 'saves': 19}`	`{'points': 20, 'saves': 20}`	Failed
second regression	`{'points': 32, 'saves': 16}`	`{'points': 32, 'saves': 16}`	Passed
normal control 1	`{'points': 10, 'saves': 5}`	`{'points': 10, 'saves': 5}`	Passed
normal control 2	`{'points': 36, 'saves': 13}`	`{'points': 36, 'saves': 13}`	Passed
normal control 3	`{'points': -4, 'saves': 23}`	`{'points': -4, 'saves': 23}`	Passed
normal control 4	`{'points': 42, 'saves': 31}`	`{'points': 42, 'saves': 31}`	Passed

SHA-256 / 34d8511ea417b24cdbe3741c1a0762984f4bb4e606cb019e3e1d41e986ca5cdd

3 / The verified repair

Exit 0

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(g):
    ga = g['team_ga'] - g['empty_net_ga']
    saves = g['shots_faced'] - ga
    pts = saves * 2 - ga * 20
    pts += {'W': 40, 'OTL': 10}.get(g['decision'], 0)
    if ga == 0 and g['toi_sec'] >= 3600:
        pts += 30
    return {'saves': saves, 'points': pts}
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
fixtures = [[('regression: empty-net attribution',
   [{'decision': 'OTL', 'empty_net_ga': 1, 'shots_faced': 26, 'team_ga': 1, 'toi_sec': 3540}],
   {'points': 62, 'saves': 26}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'W', 'empty_net_ga': 2, 'shots_faced': 23, 'team_ga': 5, 'toi_sec': 2400}],
   {'points': 20, 'saves': 20}),
  ('second regression',
   [{'decision': None, 'empty_net_ga': 1, 'shots_faced': 16, 'team_ga': 1, 'toi_sec': 2400}],
   {'points': 32, 'saves': 16}),
  ('normal control 1',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 7, 'team_ga': 2, 'toi_sec': 3900}],
   {'points': 10, 'saves': 5}),
  ('normal control 2',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 13, 'team_ga': 0, 'toi_sec': 2675}],
   {'points': 36, 'saves': 13}),
  ('normal control 3',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 26, 'team_ga': 3, 'toi_sec': 3900}],
   {'points': -4, 'saves': 23}),
  ('normal control 4',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 32, 'team_ga': 1, 'toi_sec': 3540}],
   {'points': 42, 'saves': 31})],
 [('regression: empty-net attribution',
   [{'decision': 'W', 'empty_net_ga': 1, 'shots_faced': 34, 'team_ga': 2, 'toi_sec': 3540}],
   {'points': 86, 'saves': 33}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'OTL', 'empty_net_ga': 2, 'shots_faced': 29, 'team_ga': 4, 'toi_sec': 3900}],
   {'points': 24, 'saves': 27}),
  ('second regression',
   [{'decision': 'L', 'empty_net_ga': 1, 'shots_faced': 18, 'team_ga': 3, 'toi_sec': 3900}],
   {'points': -8, 'saves': 16}),
  ('normal control 1',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 2, 'team_ga': 1, 'toi_sec': 3600}],
   {'points': -18, 'saves': 1}),
  ('normal control 2',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 8, 'team_ga': 1, 'toi_sec': 3599}],
   {'points': 4, 'saves': 7}),
  ('normal control 3',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 16, 'team_ga': 2, 'toi_sec': 3600}],
   {'points': -2, 'saves': 14}),
  ('normal control 4',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 3, 'team_ga': 1, 'toi_sec': 3540}],
   {'points': -6, 'saves': 2})],
 [('regression: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 1, 'shots_faced': 38, 'team_ga': 3, 'toi_sec': 2400}],
   {'points': 32, 'saves': 36}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 39, 'team_ga': 2, 'toi_sec': 3600}],
   {'points': 108, 'saves': 39}),
  ('second regression',
   [{'decision': 'L', 'empty_net_ga': 1, 'shots_faced': 39, 'team_ga': 2, 'toi_sec': 3599}],
   {'points': 56, 'saves': 38}),
  ('normal control 1',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 36, 'team_ga': 2, 'toi_sec': 2400}],
   {'points': 28, 'saves': 34}),
  ('normal control 2',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 20, 'team_ga': 5, 'toi_sec': 2400}],
   {'points': -30, 'saves': 15}),
  ('normal control 3',
   [{'decision': 'L', 'empty_net_ga': 0, 'shots_faced': 17, 'team_ga': 2, 'toi_sec': 3600}],
   {'points': -10, 'saves': 15}),
  ('normal control 4',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 40, 'team_ga': 0, 'toi_sec': 3600}],
   {'points': 150, 'saves': 40})],
 [('regression: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 37, 'team_ga': 2, 'toi_sec': 2400}],
   {'points': 74, 'saves': 37}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 1, 'team_ga': 2, 'toi_sec': 3540}],
   {'points': 2, 'saves': 1}),
  ('second regression',
   [{'decision': None, 'empty_net_ga': 2, 'shots_faced': 7, 'team_ga': 2, 'toi_sec': 3900}],
   {'points': 44, 'saves': 7}),
  ('normal control 1',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 1, 'team_ga': 1, 'toi_sec': 3600}],
   {'points': -20, 'saves': 0}),
  ('normal control 2',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 38, 'team_ga': 5, 'toi_sec': 3540}],
   {'points': 6, 'saves': 33}),
  ('normal control 3',
   [{'decision': 'L', 'empty_net_ga': 0, 'shots_faced': 15, 'team_ga': 0, 'toi_sec': 3540}],
   {'points': 30, 'saves': 15}),
  ('normal control 4',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 11, 'team_ga': 0, 'toi_sec': 2400}],
   {'points': 32, 'saves': 11})],
 [('regression: empty-net attribution',
   [{'decision': 'W', 'empty_net_ga': 1, 'shots_faced': 40, 'team_ga': 4, 'toi_sec': 3599}],
   {'points': 54, 'saves': 37}),
  ('partial repair probe: empty-net attribution',
   [{'decision': 'L', 'empty_net_ga': 2, 'shots_faced': 33, 'team_ga': 3, 'toi_sec': 2400}],
   {'points': 44, 'saves': 32}),
  ('second regression',
   [{'decision': 'OTL', 'empty_net_ga': 1, 'shots_faced': 6, 'team_ga': 3, 'toi_sec': 3600}],
   {'points': -22, 'saves': 4}),
  ('normal control 1',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 9, 'team_ga': 0, 'toi_sec': 3600}],
   {'points': 88, 'saves': 9}),
  ('normal control 2',
   [{'decision': None, 'empty_net_ga': 0, 'shots_faced': 25, 'team_ga': 0, 'toi_sec': 3599}],
   {'points': 50, 'saves': 25}),
  ('normal control 3',
   [{'decision': 'OTL', 'empty_net_ga': 0, 'shots_faced': 17, 'team_ga': 5, 'toi_sec': 2400}],
   {'points': -66, 'saves': 12}),
  ('normal control 4',
   [{'decision': 'W', 'empty_net_ga': 0, 'shots_faced': 6, 'team_ga': 1, 'toi_sec': 3599}],
   {'points': 30, 'saves': 5})]]
for label, args, expected in fixtures[N-1]:
    check(label, solve(*args), expected)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
regression: empty-net attribution	`{'points': 62, 'saves': 26}`	`{'points': 62, 'saves': 26}`	Passed
partial repair probe: empty-net attribution	`{'points': 20, 'saves': 20}`	`{'points': 20, 'saves': 20}`	Passed
second regression	`{'points': 32, 'saves': 16}`	`{'points': 32, 'saves': 16}`	Passed
normal control 1	`{'points': 10, 'saves': 5}`	`{'points': 10, 'saves': 5}`	Passed
normal control 2	`{'points': 36, 'saves': 13}`	`{'points': 36, 'saves': 13}`	Passed
normal control 3	`{'points': -4, 'saves': 23}`	`{'points': -4, 'saves': 23}`	Passed
normal control 4	`{'points': 42, 'saves': 31}`	`{'points': 42, 'saves': 31}`	Passed

SHA-256 / 10cf0cd3546609aa248ddbdb96cd829a507c72b6e43abbd77ca6041d52453a8c

Verification & scope

A deterministic toy scoring contract stipulated for this example; it is not the rulebook of any real fantasy platform. This reproducer isolates one failure mechanism. Results cover the supplied fixtures. Variants within a family share a test contract and should remain grouped when constructing evaluation splits. Related mechanisms with a shared evaluation_group must also remain together; these controlled models are not independent production incidents.

Observations recorded using Python 3.12.14 at 2026-09-29T14:50:37.664234+00:00.

Case digest / e3d3f96b7cd3dbc5e6263a990ac08acf5803619df6dd4f4f160b92deb707f477