Celeste

Patch 2026-09-29.

Summary

Win rate 50.1% ± 0.3 (n=83,656). Grade B. vs average +0.1 pp.

Kills / deaths / assists per game: 6.2 / 6.0 / 12.5. Net worth per minute: 1,149.3. Average match length: 36.4 min.

Lane highlights: this patch, all ranks, clean matches, all purchases. Ranked by observed match win rate among assigned-lane opponents with enough games; differences use this hero's overall rate in the same population.

Best observed lane opponent: Paradox — 57.6% (likely range 56.1% to 59.1%), +7.6 pp vs hero overall, 3,988 games.

Worst observed lane opponent: Paige — 43.5% (likely range 42.1% to 44.9%), -6.6 pp vs hero overall, 4,700 games.

Top items and bottom items by phase

Bought before 10 min

Top items: Suppressor +2.9 pp, Mystic Burst +1.4 pp, Mystic Expansion +1.2 pp

Bottom items: Opening Rounds -3.3 pp, Extra Stamina -3.8 pp, Extra Regen -0.8 pp

10–20 min

Top items: Spirit Shielding +1.0 pp, Torment Pulse +0.8 pp, Restorative Locket +0.6 pp

Bottom items: Radiant Regeneration -4.0 pp, Opening Rounds -1.7 pp, Extra Stamina -1.1 pp

20 min or later

Top items: Spellslinger +1.1 pp, Witchmail +0.5 pp, Greater Expansion +0.3 pp

Bottom items: Transcendent Cooldown -3.6 pp, Boundless Spirit -1.3 pp, Torment Pulse -1.2 pp

Ranked by the low end of each likely range, from items with enough games. See Win rate impact below for the full tables.

View the most popular and highest win rate builds for this hero.

By rank tier

Show the rank table

Tier	Games	Win rate	Pick share	Ban rate
All tiers	83,656	50.1% ± 0.3 (n=83,656)	1.8% (n=4,643,618)	3.3% (n=387,129)
1 Initiate	7,067	45.6% ± 1.2 (n=7,067)	2.5% (n=286,633)	2.0% (n=26,191)
2 Seeker	11,091	48.4% ± 0.9 (n=11,091)	2.0% (n=567,145)	1.9% (n=53,624)
3 Acolyte	9,470	49.5% ± 1.0 (n=9,470)	1.8% (n=538,601)	2.0% (n=49,318)
4 Sentinel	7,936	50.6% ± 1.1 (n=7,936)	1.7% (n=454,663)	2.0% (n=40,935)
5 Mystic	7,090	50.7% ± 1.2 (n=7,090)	1.7% (n=412,948)	1.7% (n=37,362)
6 Ritualist	6,729	50.7% ± 1.2 (n=6,729)	1.6% (n=417,480)	1.6% (n=36,413)
7 Emissary	6,011	51.9% ± 1.3 (n=6,011)	1.5% (n=403,465)	1.8% (n=36,580)
8 Oracle	8,524	51.0% ± 1.1 (n=8,524)	1.6% (n=546,368)	5.3% (n=42,649)
9 Phantom	7,162	52.6% ± 1.2 (n=7,162)	1.7% (n=432,077)	7.9% (n=40,417)
10 Ascendant	4,353	54.3% ± 1.5 (n=4,353)	2.0% (n=213,670)	7.0% (n=16,821)
11 Eternus	1,707	54.8% ± 2.4 (n=1,707)	2.2% (n=78,077)	13.7% (n=6,819)

Win rate by match length

Show the match length table

Match length	Win rate
under 30 min	51.5% ± 0.8 (n=14,604)
30–40 min	50.4% ± 0.5 (n=46,025)
40 min or more	48.4% ± 0.6 (n=23,027)

Win rate impact

Show the win rate impact tables

Computed 2026-10-23 00:36 UTC. All ranks, clean matches.

Bought before 10 min

Show 27 rows with too little data

Comparison: Bought before 10 min. Up to 10 rows per tier, eligible alternatives first. Purchase shares: 83,656 hero appearances.
Item	Comparison tier	Bought in	Win rate when bought	Win rate impact	Likely range	Distinct matches	Data covered	Evidence
Mystic Burst	Tier 1 / 7 alternatives (group 1)	28.1%	52.2% ± 0.6 (n=23,516)	+1.4 pp	+0.8 to +1.9 pp	83,412	99.8%	likely helps
Mystic Expansion	Tier 1 / 7 alternatives (group 1)	46.8%	51.7% ± 0.5 (n=39,113)	+1.2 pp	+0.5 to +1.9 pp	83,412	99.8%	likely helps
Grit	Tier 1 / 7 alternatives (group 1)	14.9%	53.0% ± 0.9 (n=12,488)	+1.3 pp	+0.4 to +2.2 pp	83,412	99.8%	likely helps
Mystic Regeneration	Tier 1 / 7 alternatives (group 1)	85.7%	51.1% ± 0.4 (n=71,718)	+0.2 pp	-0.0 to +0.5 pp	83,412	99.8%	unclear
High-Velocity Rounds	Tier 1 / 7 alternatives (group 1)	92.8%	50.4% ± 0.4 (n=77,620)	-0.7 pp	-1.3 to -0.2 pp	83,412	99.8%	likely hurts
Extra Regen	Tier 1 / 7 alternatives (group 1)	85.2%	50.9% ± 0.4 (n=71,275)	-0.8 pp	-1.4 to -0.2 pp	83,412	99.8%	likely hurts
Extra Stamina	Tier 1 / 7 alternatives (group 1)	12.0%	47.1% ± 1.0 (n=10,069)	-3.8 pp	-5.8 to -1.7 pp	83,412	99.8%	likely hurts
Suppressor	Tier 2 / 3 alternatives (group 2)	14.5%	54.7% ± 0.9 (n=12,107)	+2.9 pp	+1.4 to +4.4 pp	43,312	100.0%	likely helps
Healing Booster	Tier 2 / 3 alternatives (group 2)	22.4%	54.9% ± 0.7 (n=18,724)	+0.5 pp	-0.3 to +1.3 pp	43,312	100.0%	unclear
Opening Rounds	Tier 2 / 3 alternatives (group 2)	26.4%	46.9% ± 0.7 (n=22,091)	-3.3 pp	-4.0 to -2.6 pp	43,312	100.0%	likely hurts
Spirit Strike	Tier 1 / 7 alternatives (group 1)	0.8%	48.6% ± 3.7 (n=710)	—	—	83,412	99.8%	not enough data
Rapid Rounds	Tier 1 / 7 alternatives (group 1)	3.7%	45.6% ± 1.8 (n=3,082)	—	—	83,412	99.8%	not enough data
Extra Spirit	Tier 1 / 7 alternatives (group 1)	7.8%	46.6% ± 1.2 (n=6,552)	—	—	83,412	99.8%	not enough data
Improved Spirit	Tier 2 / 3 alternatives (group 2)	4.1%	45.4% ± 1.7 (n=3,422)	—	—	43,312	100.0%	not enough data
Melee Charge	Tier 2 / 3 alternatives (group 2)	0.1%	insufficient data (n=58)	—	—	43,312	100.0%	not enough data
Quicksilver Reload	Tier 2 / 3 alternatives (group 2)	1.0%	47.5% ± 3.4 (n=833)	—	—	43,312	100.0%	not enough data
Stalker	Tier 2 / 3 alternatives (group 2)	0.0%	insufficient data (n=41)	—	—	43,312	100.0%	not enough data
Spirit Shielding	Tier 2 / 3 alternatives (group 2)	3.1%	57.6% ± 1.9 (n=2,552)	—	—	43,312	100.0%	not enough data
Compress Cooldown	Tier 2 / 3 alternatives (group 2)	0.4%	44.4% ± 5.0 (n=374)	—	—	43,312	100.0%	not enough data
Active Reload	Tier 2 / 3 alternatives (group 2)	0.1%	insufficient data (n=79)	—	—	43,312	100.0%	not enough data
Radiant Regeneration	Tier 3 / 1 alternatives (group 3)	54.6%	54.1% ± 0.5 (n=45,694)	—	—	45,682	100.0%	not enough data
Stamina Mastery	Tier 3 / 1 alternatives (group 3)	0.2%	49.1% ± 7.6 (n=161)	—	—	45,682	100.0%	not enough data
Torment Pulse	Tier 3 / 1 alternatives (group 3)	1.1%	54.5% ± 3.2 (n=949)	—	—	45,682	100.0%	not enough data
Majestic Leap	Tier 3 / 1 alternatives (group 3)	0.0%	insufficient data (n=1)	—	—	45,682	100.0%	not enough data
Silence Wave	Tier 3 / 1 alternatives (group 3)	0.0%	insufficient data (n=3)	—	—	45,682	100.0%	not enough data
Express Shot	Tier 3 / 1 alternatives (group 3)	0.0%	insufficient data (n=8)	—	—	45,682	100.0%	not enough data
Cultist Sacrifice	Tier 3 / 1 alternatives (group 3)	0.1%	insufficient data (n=85)	—	—	45,682	100.0%	not enough data
Rapid Recharge	Tier 3 / 1 alternatives (group 3)	0.2%	41.2% ± 8.2 (n=136)	—	—	45,682	100.0%	not enough data
Greater Expansion	Tier 3 / 1 alternatives (group 3)	0.1%	50.0% ± 9.4 (n=104)	—	—	45,682	100.0%	not enough data
Lifestrike	Tier 3 / 1 alternatives (group 3)	0.0%	insufficient data (n=6)	—	—	45,682	100.0%	not enough data
Transcendent Cooldown	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=1)	—	—	0	0.0%	not enough data
Spiritual Overflow	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=1)	—	—	0	0.0%	not enough data
Ricochet	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=1)	—	—	0	0.0%	not enough data
Boundless Spirit	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=2)	—	—	0	0.0%	not enough data
Magic Carpet	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=1)	—	—	0	0.0%	not enough data
Escalating Exposure	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=1)	—	—	0	0.0%	not enough data
Mercurial Magnum	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=3)	—	—	0	0.0%	not enough data

10–20 min

Show 26 rows with too little data Show all 14 rows

Comparison: 10–20 min. Up to 10 rows per tier, eligible alternatives first. Purchase shares: 83,656 hero appearances.
Item	Comparison tier	Bought in	Win rate when bought	Win rate impact	Likely range	Distinct matches	Data covered	Evidence
Grit	Tier 1 / 3 alternatives (group 1)	22.2%	52.7% ± 0.7 (n=18,610)	+0.4 pp	-0.3 to +1.1 pp	44,148	100.0%	unclear
Mystic Expansion	Tier 1 / 3 alternatives (group 1)	20.5%	51.6% ± 0.7 (n=17,146)	-0.3 pp	-0.9 to +0.3 pp	44,148	100.0%	unclear
Extra Stamina	Tier 1 / 3 alternatives (group 1)	22.3%	50.1% ± 0.7 (n=18,652)	-1.1 pp	-1.7 to -0.6 pp	44,148	100.0%	likely hurts
Spirit Shielding	Tier 2 / 7 alternatives (group 2)	34.5%	53.0% ± 0.6 (n=28,861)	+1.0 pp	+0.6 to +1.4 pp	76,155	100.0%	likely helps
Restorative Locket	Tier 2 / 7 alternatives (group 2)	51.2%	51.8% ± 0.5 (n=42,864)	+0.6 pp	+0.3 to +0.9 pp	76,155	100.0%	likely helps
Healbane	Tier 2 / 7 alternatives (group 2)	21.8%	50.9% ± 0.7 (n=18,247)	-0.2 pp	-0.8 to +0.4 pp	76,155	100.0%	unclear
Healing Booster	Tier 2 / 7 alternatives (group 2)	58.8%	49.9% ± 0.4 (n=49,213)	-0.6 pp	-0.9 to -0.3 pp	76,155	100.0%	likely hurts
Suppressor	Tier 2 / 7 alternatives (group 2)	27.0%	48.6% ± 0.7 (n=22,559)	-0.4 pp	-1.0 to +0.1 pp	76,155	100.0%	unclear
Mystic Vulnerability	Tier 2 / 7 alternatives (group 2)	13.4%	50.5% ± 0.9 (n=11,249)	-1.3 pp	-2.5 to -0.0 pp	76,155	100.0%	likely hurts
Opening Rounds	Tier 2 / 7 alternatives (group 2)	16.5%	52.1% ± 0.8 (n=13,823)	-1.7 pp	-2.5 to -0.9 pp	76,155	100.0%	likely hurts
Torment Pulse	Tier 3 / 4 alternatives (group 3)	69.2%	53.1% ± 0.4 (n=57,871)	+0.8 pp	+0.6 to +1.0 pp	76,377	100.0%	likely helps
Greater Expansion	Tier 3 / 4 alternatives (group 3)	30.5%	57.0% ± 0.6 (n=25,498)	+0.9 pp	+0.2 to +1.6 pp	76,377	100.0%	likely helps
Stamina Mastery	Tier 3 / 4 alternatives (group 3)	20.3%	51.7% ± 0.8 (n=17,010)	-1.2 pp	-1.9 to -0.6 pp	76,377	100.0%	likely hurts
Radiant Regeneration	Tier 3 / 4 alternatives (group 3)	36.0%	46.0% ± 0.6 (n=30,086)	-4.0 pp	-5.3 to -2.7 pp	76,377	100.0%	likely hurts
Spirit Strike	Tier 1 / 3 alternatives (group 1)	3.0%	52.3% ± 2.0 (n=2,510)	—	—	44,148	100.0%	not enough data
Rapid Rounds	Tier 1 / 3 alternatives (group 1)	3.0%	46.2% ± 1.9 (n=2,512)	—	—	44,148	100.0%	not enough data
Extra Spirit	Tier 1 / 3 alternatives (group 1)	5.3%	46.7% ± 1.5 (n=4,427)	—	—	44,148	100.0%	not enough data
Monster Rounds	Tier 1 / 3 alternatives (group 1)	0.1%	47.4% ± 9.7 (n=97)	—	—	44,148	100.0%	not enough data
Close Quarters	Tier 1 / 3 alternatives (group 1)	0.0%	insufficient data (n=33)	—	—	44,148	100.0%	not enough data
Melee Lifesteal	Tier 1 / 3 alternatives (group 1)	0.0%	insufficient data (n=22)	—	—	44,148	100.0%	not enough data
Mystic Regeneration	Tier 1 / 3 alternatives (group 1)	4.0%	43.0% ± 1.7 (n=3,326)	—	—	44,148	100.0%	not enough data
Improved Spirit	Tier 2 / 7 alternatives (group 2)	7.1%	46.7% ± 1.3 (n=5,922)	—	—	76,155	100.0%	not enough data
Melee Charge	Tier 2 / 7 alternatives (group 2)	0.1%	48.4% ± 9.9 (n=95)	—	—	76,155	100.0%	not enough data
Quicksilver Reload	Tier 2 / 7 alternatives (group 2)	3.3%	43.1% ± 1.8 (n=2,798)	—	—	76,155	100.0%	not enough data
Majestic Leap	Tier 3 / 4 alternatives (group 3)	0.0%	insufficient data (n=2)	—	—	76,377	100.0%	not enough data
Silence Wave	Tier 3 / 4 alternatives (group 3)	0.0%	insufficient data (n=27)	—	—	76,377	100.0%	not enough data
Express Shot	Tier 3 / 4 alternatives (group 3)	0.1%	insufficient data (n=66)	—	—	76,377	100.0%	not enough data
Cultist Sacrifice	Tier 3 / 4 alternatives (group 3)	0.2%	47.3% ± 8.5 (n=129)	—	—	76,377	100.0%	not enough data
Rapid Recharge	Tier 3 / 4 alternatives (group 3)	2.8%	46.0% ± 2.0 (n=2,343)	—	—	76,377	100.0%	not enough data
Tesla Bullets	Tier 3 / 4 alternatives (group 3)	0.0%	insufficient data (n=1)	—	—	76,377	100.0%	not enough data
Spirit Burn	Tier 4 / 0 alternatives (group 4)	0.1%	insufficient data (n=88)	—	—	0	0.0%	not enough data
Glass Cannon	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=2)	—	—	0	0.0%	not enough data
Lightning Scroll	Tier 4 / 0 alternatives (group 4)	0.1%	41.0% ± 9.2 (n=105)	—	—	0	0.0%	not enough data
Refresher	Tier 4 / 0 alternatives (group 4)	0.1%	62.7% ± 9.2 (n=102)	—	—	0	0.0%	not enough data
Crushing Fists	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=15)	—	—	0	0.0%	not enough data
Leech	Tier 4 / 0 alternatives (group 4)	0.1%	insufficient data (n=59)	—	—	0	0.0%	not enough data
Transcendent Cooldown	Tier 4 / 0 alternatives (group 4)	0.1%	insufficient data (n=84)	—	—	0	0.0%	not enough data
Indomitable	Tier 4 / 0 alternatives (group 4)	0.1%	insufficient data (n=87)	—	—	0	0.0%	not enough data
Silencer	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=2)	—	—	0	0.0%	not enough data
Vortex Web	Tier 4 / 0 alternatives (group 4)	0.0%	insufficient data (n=3)	—	—	0	0.0%	not enough data

20 min or later

Show 24 rows with too little data Show all 16 rows

Comparison: 20 min or later. Up to 10 rows per tier, eligible alternatives first. Purchase shares: 83,656 hero appearances.
Item	Comparison tier	Bought in	Win rate when bought	Win rate impact	Likely range	Distinct matches	Data covered	Evidence
Mystic Vulnerability	Tier 2 / 4 alternatives (group 1)	34.5%	53.1% ± 0.6 (n=28,899)	+0.2 pp	-0.2 to +0.5 pp	52,392	99.5%	unclear
Compress Cooldown	Tier 2 / 4 alternatives (group 1)	12.4%	52.8% ± 1.0 (n=10,402)	+0.4 pp	-0.3 to +1.2 pp	52,392	99.5%	unclear
Spirit Lifesteal	Tier 2 / 4 alternatives (group 1)	24.9%	52.9% ± 0.7 (n=20,838)	+0.1 pp	-0.4 to +0.5 pp	52,392	99.5%	unclear
Opening Rounds	Tier 2 / 4 alternatives (group 1)	18.3%	52.5% ± 0.8 (n=15,279)	-0.5 pp	-1.1 to +0.0 pp	52,392	99.5%	unclear
Greater Expansion	Tier 3 / 5 alternatives (group 2)	48.7%	49.2% ± 0.5 (n=40,710)	+0.3 pp	+0.1 to +0.6 pp	61,846	98.2%	likely helps
Superior Cooldown	Tier 3 / 5 alternatives (group 2)	25.5%	52.0% ± 0.7 (n=21,322)	+0.4 pp	-0.1 to +0.9 pp	61,846	98.2%	unclear
Stamina Mastery	Tier 3 / 5 alternatives (group 2)	22.8%	49.8% ± 0.7 (n=19,040)	-0.2 pp	-0.7 to +0.3 pp	61,846	98.2%	unclear
Tankbuster	Tier 3 / 5 alternatives (group 2)	19.0%	48.8% ± 0.8 (n=15,924)	-0.6 pp	-1.2 to -0.0 pp	61,846	98.2%	likely hurts
Torment Pulse	Tier 3 / 5 alternatives (group 2)	15.6%	44.3% ± 0.9 (n=13,046)	-1.2 pp	-1.9 to -0.4 pp	61,846	98.2%	likely hurts
Spellslinger	Tier 4 / 7 alternatives (group 3)	20.2%	57.9% ± 0.7 (n=16,939)	+1.1 pp	+0.5 to +1.6 pp	71,338	98.8%	likely helps
Witchmail	Tier 4 / 7 alternatives (group 3)	48.8%	55.3% ± 0.5 (n=40,865)	+0.5 pp	+0.1 to +0.8 pp	71,338	98.8%	likely helps
Escalating Exposure	Tier 4 / 7 alternatives (group 3)	47.9%	55.1% ± 0.5 (n=40,050)	+0.0 pp	-0.3 to +0.3 pp	71,338	98.8%	unclear
Spiritual Overflow	Tier 4 / 7 alternatives (group 3)	27.6%	56.0% ± 0.6 (n=23,096)	+0.2 pp	-0.3 to +0.6 pp	71,338	98.8%	unclear
Refresher	Tier 4 / 7 alternatives (group 3)	17.0%	54.1% ± 0.8 (n=14,260)	-0.4 pp	-1.2 to +0.5 pp	71,338	98.8%	unclear
Boundless Spirit	Tier 4 / 7 alternatives (group 3)	18.8%	54.4% ± 0.8 (n=15,755)	-1.3 pp	-2.0 to -0.7 pp	71,338	98.8%	likely hurts
Transcendent Cooldown	Tier 4 / 7 alternatives (group 3)	12.3%	53.9% ± 1.0 (n=10,290)	-3.6 pp	-4.6 to -2.5 pp	71,338	98.8%	likely hurts
Spirit Strike	Tier 1 / 0 alternatives (group 4)	0.2%	48.1% ± 7.1 (n=185)	—	—	0	0.0%	not enough data
Rapid Rounds	Tier 1 / 0 alternatives (group 4)	0.6%	39.8% ± 4.3 (n=498)	—	—	0	0.0%	not enough data
Mystic Expansion	Tier 1 / 0 alternatives (group 4)	7.9%	49.0% ± 1.2 (n=6,567)	—	—	0	0.0%	not enough data
Extra Spirit	Tier 1 / 0 alternatives (group 4)	6.3%	56.6% ± 1.3 (n=5,310)	—	—	0	0.0%	not enough data
Monster Rounds	Tier 1 / 0 alternatives (group 4)	0.0%	insufficient data (n=14)	—	—	0	0.0%	not enough data
Close Quarters	Tier 1 / 0 alternatives (group 4)	0.0%	insufficient data (n=21)	—	—	0	0.0%	not enough data
Melee Lifesteal	Tier 1 / 0 alternatives (group 4)	0.0%	insufficient data (n=12)	—	—	0	0.0%	not enough data
Mystic Regeneration	Tier 1 / 0 alternatives (group 4)	0.3%	41.7% ± 5.8 (n=278)	—	—	0	0.0%	not enough data
Extended Magazine	Tier 1 / 0 alternatives (group 4)	0.1%	insufficient data (n=64)	—	—	0	0.0%	not enough data
Grit	Tier 1 / 0 alternatives (group 4)	1.4%	50.4% ± 2.8 (n=1,192)	—	—	0	0.0%	not enough data
Improved Spirit	Tier 2 / 4 alternatives (group 1)	9.6%	53.4% ± 1.1 (n=8,017)	—	—	52,392	99.5%	not enough data
Melee Charge	Tier 2 / 4 alternatives (group 1)	0.0%	insufficient data (n=30)	—	—	52,392	99.5%	not enough data
Quicksilver Reload	Tier 2 / 4 alternatives (group 1)	0.6%	49.6% ± 4.2 (n=534)	—	—	52,392	99.5%	not enough data
Stalker	Tier 2 / 4 alternatives (group 1)	0.0%	insufficient data (n=5)	—	—	52,392	99.5%	not enough data
Spirit Shielding	Tier 2 / 4 alternatives (group 1)	2.3%	43.9% ± 2.2 (n=1,890)	—	—	52,392	99.5%	not enough data
Active Reload	Tier 2 / 4 alternatives (group 1)	0.0%	insufficient data (n=37)	—	—	52,392	99.5%	not enough data
Majestic Leap	Tier 3 / 5 alternatives (group 2)	0.0%	insufficient data (n=11)	—	—	61,846	98.2%	not enough data
Silence Wave	Tier 3 / 5 alternatives (group 2)	0.2%	39.8% ± 8.4 (n=128)	—	—	61,846	98.2%	not enough data
Express Shot	Tier 3 / 5 alternatives (group 2)	0.2%	49.0% ± 6.8 (n=202)	—	—	61,846	98.2%	not enough data
Cultist Sacrifice	Tier 3 / 5 alternatives (group 2)	0.0%	insufficient data (n=25)	—	—	61,846	98.2%	not enough data
Rapid Recharge	Tier 3 / 5 alternatives (group 2)	6.4%	45.5% ± 1.3 (n=5,365)	—	—	61,846	98.2%	not enough data
Frenzy	Tier 4 / 7 alternatives (group 3)	0.1%	insufficient data (n=48)	—	—	71,338	98.8%	not enough data
Spirit Burn	Tier 4 / 7 alternatives (group 3)	6.9%	55.5% ± 1.3 (n=5,743)	—	—	71,338	98.8%	not enough data
Glass Cannon	Tier 4 / 7 alternatives (group 3)	0.0%	insufficient data (n=18)	—	—	71,338	98.8%	not enough data

Observed builds

View Most popular / Highest win rate builds for these patch, rank, scope and purchase-view settings.

Item effects

Show the item effect tables

Computed 2026-10-23 00:36 UTC from 387,129 clean matches across all ranks. Expected win chance at purchase, Win rate vs expected and Edge over same-tier items come from a model of the game before each purchase; they are separate from the win rate impact comparison.

Bought before 10 min

0 items below the selected purchase-share filter. Purchase shares use all hero games, including for phase purchases.

Purchase shares: 83,656 hero appearances. Computed 2026-10-23 00:36 UTC from 387,129 clean matches, all ranks.
Item	Tier	Bought in	Win rate when bought	Expected win chance at purchase	Win rate vs expected	vs hero	Edge over same-tier items
Spirit Shielding	2	3.1%	57.6% ± 1.9 (n=2,552)	51.8%	+5.8 pp	+7.5 pp	+4.8 pp
Restorative Locket	2	2.8%	58.4% ± 2.0 (n=2,341)	52.8%	+5.6 pp	+8.3 pp	+4.5 pp
Suppressor	2	14.5%	54.7% ± 0.9 (n=12,107)	49.5%	+5.2 pp	+4.6 pp	+4.2 pp
Healbane	2	2.3%	56.1% ± 2.2 (n=1,888)	50.9%	+5.2 pp	+6.0 pp	+4.2 pp
Healing Booster	2	22.4%	54.9% ± 0.7 (n=18,724)	51.3%	+3.6 pp	+4.8 pp	+2.6 pp
Mystic Slow	2	0.4%	50.5% ± 5.1 (n=364)	47.7%	+2.8 pp	+0.5 pp	+1.8 pp
Mystic Burst	1	28.1%	52.2% ± 0.6 (n=23,516)	48.9%	+3.4 pp	+2.2 pp	+1.8 pp
Mystic Vulnerability	2	1.5%	52.1% ± 2.7 (n=1,267)	49.3%	+2.8 pp	+2.0 pp	+1.8 pp
Cold Front	2	3.4%	50.3% ± 1.8 (n=2,868)	47.5%	+2.8 pp	+0.2 pp	+1.8 pp
Grit	1	14.9%	53.0% ± 0.9 (n=12,488)	49.7%	+3.2 pp	+2.9 pp	+1.7 pp

10–20 min

0 items below the selected purchase-share filter. Purchase shares use all hero games, including for phase purchases.

Purchase shares: 83,656 hero appearances. Computed 2026-10-23 00:36 UTC from 387,129 clean matches, all ranks.
Item	Tier	Bought in	Win rate when bought	Expected win chance at purchase	Win rate vs expected	vs hero	Edge over same-tier items
Scourge	4	0.1%	76.1% ± 8.8 (n=88)	58.0%	+18.2 pp	+26.1 pp	+9.1 pp
Refresher	4	0.1%	62.7% ± 9.2 (n=102)	47.1%	+15.6 pp	+12.7 pp	+6.6 pp
Guardian Ward	2	0.2%	57.7% ± 7.7 (n=156)	51.3%	+6.4 pp	+7.6 pp	+4.4 pp
Monster Rounds	1	0.1%	47.4% ± 9.7 (n=97)	43.5%	+4.0 pp	-2.6 pp	+3.0 pp
Witchmail	4	3.7%	66.0% ± 1.7 (n=3,093)	54.1%	+11.9 pp	+16.0 pp	+2.9 pp
Grit	1	22.2%	52.7% ± 0.7 (n=18,610)	49.0%	+3.7 pp	+2.6 pp	+2.7 pp
Spellslinger	4	0.8%	63.3% ± 3.6 (n=683)	51.9%	+11.3 pp	+13.2 pp	+2.3 pp
Spirit Snatch	3	1.9%	58.9% ± 2.4 (n=1,604)	52.4%	+6.4 pp	+8.8 pp	+2.0 pp
Spirit Shielding	2	34.5%	53.0% ± 0.6 (n=28,861)	49.0%	+4.0 pp	+2.9 pp	+2.0 pp
Dispel Magic	3	3.4%	53.5% ± 1.8 (n=2,812)	47.2%	+6.3 pp	+3.5 pp	+1.9 pp

20 min or later

0 items below the selected purchase-share filter. Purchase shares use all hero games, including for phase purchases.

Purchase shares: 83,656 hero appearances. Computed 2026-10-23 00:36 UTC from 387,129 clean matches, all ranks.
Item	Tier	Bought in	Win rate when bought	Expected win chance at purchase	Win rate vs expected	vs hero	Edge over same-tier items
Guardian Ward	2	0.4%	56.4% ± 5.5 (n=312)	51.1%	+5.3 pp	+6.3 pp	+4.7 pp
Trophy Collector	2	0.1%	51.5% ± 9.7 (n=99)	47.9%	+3.6 pp	+1.5 pp	+3.0 pp
Witchmail	4	48.8%	55.3% ± 0.5 (n=40,865)	47.7%	+7.6 pp	+5.2 pp	+2.4 pp
Extra Spirit	1	6.3%	56.6% ± 1.3 (n=5,310)	54.3%	+2.3 pp	+6.5 pp	+2.4 pp
Scourge	4	3.3%	58.4% ± 1.8 (n=2,789)	51.4%	+7.0 pp	+8.3 pp	+1.8 pp
Divine Barrier	4	0.6%	58.7% ± 4.1 (n=542)	51.8%	+6.9 pp	+8.6 pp	+1.7 pp
Compress Cooldown	2	12.4%	52.8% ± 1.0 (n=10,402)	50.6%	+2.3 pp	+2.8 pp	+1.7 pp
Blood Tribute	3	1.1%	60.5% ± 3.1 (n=962)	57.7%	+2.8 pp	+10.4 pp	+1.5 pp
Extra Stamina	1	7.3%	52.4% ± 1.3 (n=6,069)	50.9%	+1.4 pp	+2.3 pp	+1.5 pp
Mystic Vulnerability	2	34.5%	53.1% ± 0.6 (n=28,899)	51.1%	+2.0 pp	+3.0 pp	+1.4 pp

How we calculate this

In observed win rates, ± shows half the width of the likely range; n is the number of games or appearances counted. Percentage-point differences are labeled pp.

Grades use win rates adjusted toward the overall average to reduce small-sample swings. Difference from the corresponding population average, in percentage points: S at or above +3; A at or above +1 and < +3; B > -1 and < +1; C > -3 and at or below -1; D at or below -3. A dash means not enough data. The displayed win rate is the observed rate; grades use the adjusted rate. Hero-summary grades use all ranks, clean matches for this patch.

Win rate when bought is the observed match win rate among buyers. Win rate impact compares eligible same-tier alternatives in the same purchase phase after giving them a common mix of purchase times, team wealth leads and ranks. Compare only rows with the same alternatives and mix; numbered groups distinguish different comparisons when a table contains more than one. The reference includes the item itself; zero is the reference average. Data covered is the share of the intended comparison mix supported by the data, counted in groups of purchase time, team wealth lead and rank.

The likely range is a 95% interval accounting for purchases from the same match, conditional on the chosen alternatives and mix. Likely helps means the entire range is above zero; likely hurts means below zero; unclear includes zero. Not enough data means a comparison failed its sample, support or display requirements. These associations do not establish cause, account for unmeasured differences or repeated players across matches, predict future patches, or add up to a build win rate. Filtering items does not change the comparison groups, weights or estimates.

Expected win chance at purchase is the average win chance a separate model of the game gave the buyers just before they bought. Win rate vs expected is the buyers' actual win rate minus that expected chance, measured on the same data used to fit the model. Edge over same-tier items is that difference minus the average difference for same-tier items on this hero in the same phase; it is observational, not the win rate impact comparison. Purchase shares use all hero appearances. Rows are limited to 10 per comparison tier.

Win chance effect comes from a statistical model over clean matches of this patch and match tier. It is shown in percentage points above a 50% baseline from this hero's side: positive means this hero's side wins more often. Tables are sorted by the underlying model coefficient.