Why Deadlock teams look uneven
Starting account Ranked points, not private MMR or hero-skill ratings. Historical win rates are group results, not match odds.
Rating gaps and results
All 111,037 eligible captured games, not just the examples.
Gap bands are not a linear scale. Bars count games; thin lines show reference ranges. * Fewer than 60 games.
Historical results, not exact match odds. Reference ranges are not adjusted for repeat players.
A high-score explanation—not the whole story
For match 137916267, a tested 65k ceiling reduces the scoring gap from 3,087.5 to 140.8. The reported points stay unchanged. This fits the newer EU/NA sample; it is not a universal clamp or confirmed Valve MMR.
Where gaps happen
Large gaps occur below the top ranks too. Each band is defined by its highest-ranked player.
All twelve Emissary or below
67,001 captured gamesHighest player Oracle / Phantom
29,095 captured gamesHighest player Ascendant
13,768 captured gamesAt least one Eternus
1,173 captured gamesLower-rank proof
This is not a low average hiding one very high player. Every starting badge is below Ascendant.
The 65k clamp cannot explain most of them
Capping at 65,000 changes none of those points, for any split. This rules out that particular clamp as their explanation—not every possible private-rating transformation.
By server region · every player ≤65k points
Uneven teams
>1,000 avg. points · 6 games · 2 Ascendant/Eternus
TEAM A
Silver↔Eternus IV77,390
Infernus↔Eternus III69,790
Holliday↔Eternus III66,205
RemAscendant VI64,700
VenatorAscendant IV61,715
HazeAscendant III61,225
TEAM B
CalicoEternus I65,705
ParadoxAscendant VI64,800
WraithAscendant VI63,975
PocketAscendant VI63,750
Yamato↔Ascendant IV62,220
IvyAscendant IV62,050
11 comparisons: A 8 wins, B 3. Too few for a percentage.
Comparison details
Observed wins: A 8/11 · B 3/11. Rough higher-point-side reference range: 43.4–90.3%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 401,025 · B 382,500. Sum gap 18,525; team-mean gap 3,087.5.
With a tested 65,000-point ceiling proxy the scoring gap would be 140.83. This does not change the reported points and is not confirmed private MMR.
TEAM A
Deadman DannyEternus III78,100
BillyEternus III73,370
Lash↔Eternus II66,595
Infernus↔Ascendant V62,960
Nurse HarrowAscendant V62,865
MinaAscendant IV62,070
TEAM B
WraithEternus III73,020
AbramsEternus II70,400
CelesteAscendant VI64,895
HollidayAscendant VI64,380
Lady GeistAscendant IV62,340
IvyAscendant IV62,040
Comparison details
Observed wins: A 43/69 · B 26/69. Rough higher-point-side reference range: 50.5–72.8%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 405,960 · B 397,075. Sum gap 8,885; team-mean gap 1,480.83.
With a tested 65,000-point ceiling proxy the scoring gap would be 126.67. This does not change the reported points and is not confirmed private MMR.
TEAM A
SolomonEmissary I39,650
Mirage↔Ritualist III34,500
McGinnisRitualist I33,200
DrifterMystic IV29,000
SevenMystic III28,000
ShivMystic II27,350
TEAM B
VictorRitualist II34,410
HazeRitualist I33,150
SilverMystic VI32,460
HollidayMystic III28,100
Nurse HarrowMystic II27,470
Lady GeistMystic II27,100
Comparison details
Observed wins: A 112/200 · B 88/200. Rough higher-point-side reference range: 49.1–62.7%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 191,700 · B 182,690. Sum gap 9,010; team-mean gap 1,501.67.
With a tested 65,000-point ceiling proxy the scoring gap would be 1,501.67. This does not change the reported points and is not confirmed private MMR.
TEAM A
YamatoOracle I45,570
Wraith↔Emissary V43,420
BebopRitualist VI37,500
KelvinRitualist V37,270
Nurse Harrow↔Ritualist I32,955
ParadoxMystic V30,350
TEAM B
VictorEmissary V43,700
Venator↔Emissary II40,350
IvyEmissary I39,650
Deadman DannyRitualist II33,620
Mo & KrillMystic VI32,210
LashMystic III28,560
Comparison details
Observed wins: A 131/200 · B 69/200. Rough higher-point-side reference range: 58.7–71.7%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 227,065 · B 218,090. Sum gap 8,975; team-mean gap 1,495.83.
With a tested 65,000-point ceiling proxy the scoring gap would be 1,495.83. This does not change the reported points and is not confirmed private MMR.
TEAM A
HollidayAscendant II60,405
IvyPhantom VI57,660
WraithPhantom V56,710
AbramsPhantom IV55,380
InfernusPhantom IV55,375
HazePhantom I52,770
TEAM B
Lady GeistAscendant IV61,920
BebopAscendant I58,695
VindictaPhantom III54,820
Pocket↔Phantom II53,280
GravesPhantom II53,120
VenatorOracle III47,500
Comparison details
Observed wins: A 116/176 · B 60/176. Rough higher-point-side reference range: 58.6–72.5%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 338,300 · B 329,335. Sum gap 8,965; team-mean gap 1,494.17.
With a tested 65,000-point ceiling proxy the scoring gap would be 1,494.17. This does not change the reported points and is not confirmed private MMR.
TEAM A
RemSeeker VI11,500
BillySeeker V10,500
SolomonSeeker IV10,150
Rat KingSeeker II8,360
MinaSeeker I7,150
YamatoSeeker I6,630
TEAM B
Deadman DannySeeker V10,620
ShivSeeker IV9,850
GravesSeeker II8,150
IvyInitiate VI6,100
AbramsInitiate VI5,375
HazeInitiate VI5,245
Comparison details
Observed wins: A 49/85 · B 36/85. Rough higher-point-side reference range: 47.0–67.6%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 54,290 · B 45,340. Sum gap 8,950; team-mean gap 1,491.67.
With a tested 65,000-point ceiling proxy the scoring gap would be 1,491.67. This does not change the reported points and is not confirmed private MMR.
Middle gap
>250–1,000 avg. points · 6 games · 2 Ascendant/Eternus
TEAM A
CelesteEternus I67,240
VyperEternus I67,150
PocketAscendant IV62,280
Mina↔Ascendant IV61,640
ParadoxAscendant IV61,600
Grey TalonAscendant II60,470
TEAM B
WraithEternus I65,160
ViscousAscendant VI63,625
LashAscendant V62,995
Silver↔Ascendant V62,540
Rat King↔Ascendant IV61,760
Lady GeistAscendant III60,705
Comparison details
Observed wins: A 95/160 · B 65/160. Rough higher-point-side reference range: 51.6–66.7%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 380,380 · B 376,785. Sum gap 3,595; team-mean gap 599.17.
With a tested 65,000-point ceiling proxy the scoring gap would be 105.83. This does not change the reported points and is not confirmed private MMR.
TEAM A
BillyEternus II67,335
MinaEternus II65,685
ApolloAscendant VI64,255
Ivy↔Ascendant V62,985
Abrams↔Ascendant IV62,470
PaigeAscendant IV61,980
TEAM B
Lady GeistEternus II66,715
RemAscendant VI63,700
HazeAscendant V62,930
BebopAscendant V62,815
YamatoAscendant V62,640
BabaAscendant IV62,305
Comparison details
Observed wins: A 112/200 · B 88/200. Rough higher-point-side reference range: 49.1–62.7%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 384,710 · B 381,105. Sum gap 3,605; team-mean gap 600.83.
With a tested 65,000-point ceiling proxy the scoring gap would be 383.33. This does not change the reported points and is not confirmed private MMR.
TEAM A
Lady GeistPhantom II53,740
IvyPhantom II53,350
SevenPhantom II53,075
Sinclair↔Phantom I52,000
Venator↔Oracle IV49,410
YamatoOracle II46,750
TEAM B
BebopPhantom III54,035
MinaPhantom I52,070
HazeOracle V50,130
DrifterOracle V49,920
Paige↔Oracle V49,720
ShivOracle IV48,850
Comparison details
Observed wins: A 108/200 · B 92/200. Rough higher-point-side reference range: 47.1–60.8%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 308,325 · B 304,725. Sum gap 3,600; team-mean gap 600.
With a tested 65,000-point ceiling proxy the scoring gap would be 600. This does not change the reported points and is not confirmed private MMR.
TEAM A
Deadman DannyEmissary II40,700
WardenEmissary I39,650
Paige↔Emissary I39,350
CalicoRitualist VI37,850
DynamoRitualist VI37,850
MinaRitualist V37,150
TEAM B
SinclairEmissary II40,350
Lady GeistEmissary II40,000
IvyEmissary I39,000
HazeRitualist VI37,850
CelesteRitualist VI37,600
ShivRitualist II34,150
Comparison details
Observed wins: A 105/200 · B 95/200. Rough higher-point-side reference range: 45.6–59.3%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 232,550 · B 228,950. Sum gap 3,600; team-mean gap 600.
With a tested 65,000-point ceiling proxy the scoring gap would be 600. This does not change the reported points and is not confirmed private MMR.
TEAM A
SilverAscendant VI64,415
BebopAscendant V62,785
Paige↔Ascendant II60,385
Deadman DannyAscendant I59,090
Drifter↔Ascendant I59,010
Lady GeistPhantom VI58,255
TEAM B
VindictaAscendant V62,995
WardenAscendant IV62,200
BabaAscendant IV62,170
ParadoxAscendant II59,915
HollidayPhantom VI57,500
Lash↔Phantom IV55,560
Comparison details
Observed wins: A 41/90 · B 49/90. Rough higher-point-side reference range: 35.7–55.8%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 363,940 · B 360,340. Sum gap 3,600; team-mean gap 600.
With a tested 65,000-point ceiling proxy the scoring gap would be 600. This does not change the reported points and is not confirmed private MMR.
TEAM A
ApolloSentinel VI25,360
Rem↔Sentinel III21,970
WraithSentinel III21,500
Bebop↔Sentinel I19,500
Mo & Krill↔Acolyte V17,000
MinaAcolyte IV16,470
TEAM B
VenatorSentinel V23,500
CelesteSentinel II21,200
VindictaSentinel II20,500
BillySentinel II20,500
InfernusSentinel I19,500
SinclairAcolyte I13,000
Comparison details
Observed wins: A 99/200 · B 101/200. Rough higher-point-side reference range: 42.6–56.4%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 121,800 · B 118,200. Sum gap 3,600; team-mean gap 600.
With a tested 65,000-point ceiling proxy the scoring gap would be 600. This does not change the reported points and is not confirmed private MMR.
Close teams
≤250 avg. points · 6 games · 2 Ascendant/Eternus
TEAM A
Lady GeistEternus I65,290
HazeAscendant V63,120
Nurse HarrowAscendant IV61,760
LashAscendant III61,455
IvyAscendant III61,080
Baba↔Ascendant II60,315
TEAM B
HollidayEternus I65,265
VindictaAscendant IV62,305
WraithAscendant IV61,750
McGinnisAscendant III61,125
InfernusAscendant III60,995
ParadoxAscendant III60,930
Comparison details
Observed wins: A 91/200 · B 109/200. Rough higher-point-side reference range: 38.7–52.4%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 373,020 · B 372,370. Sum gap 650; team-mean gap 108.33.
With a tested 65,000-point ceiling proxy the scoring gap would be 104.17. This does not change the reported points and is not confirmed private MMR.
TEAM A
HollidayEternus II66,230
VictorAscendant VI64,695
Drifter↔Ascendant VI64,405
McGinnisAscendant VI64,210
WraithAscendant VI64,145
BebopAscendant IV61,870
TEAM B
BillyEternus I67,805
MinaAscendant VI64,815
ShivAscendant VI64,250
CalicoAscendant VI63,965
HazeAscendant V62,725
Nurse HarrowAscendant III61,470
Comparison details
Observed wins: A 94/200 · B 106/200. Rough higher-point-side reference range: 40.2–53.9%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 385,555 · B 385,030. Sum gap 525; team-mean gap 87.5.
With a tested 65,000-point ceiling proxy the scoring gap would be 350. This does not change the reported points and is not confirmed private MMR.
TEAM A
Mo & KrillEmissary III41,855
MinaEmissary II40,000
PaigeEmissary I39,650
VindictaRitualist VI38,340
CalicoRitualist V37,150
Warden↔Ritualist I33,200
TEAM B
RemEmissary III41,070
Lady GeistEmissary II40,650
LashEmissary I39,000
PocketRitualist V36,525
BebopRitualist V36,500
BillyRitualist IV35,850
Comparison details
Observed wins: A 93/200 · B 107/200. Rough higher-point-side reference range: 39.7–53.4%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 230,195 · B 229,595. Sum gap 600; team-mean gap 100.
With a tested 65,000-point ceiling proxy the scoring gap would be 100. This does not change the reported points and is not confirmed private MMR.
TEAM A
HazeMystic VI31,220
InfernusMystic IV29,700
KelvinMystic IV29,350
CelesteMystic IV29,070
LashMystic IV29,000
YamatoMystic III28,420
TEAM B
BebopMystic V30,700
MirageMystic V30,350
CalicoMystic IV29,350
VenatorMystic IV29,000
RemMystic III28,410
Mina↔Mystic III28,350
Comparison details
Observed wins: A 109/200 · B 91/200. Rough higher-point-side reference range: 47.6–61.3%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 176,760 · B 176,160. Sum gap 600; team-mean gap 100.
With a tested 65,000-point ceiling proxy the scoring gap would be 100. This does not change the reported points and is not confirmed private MMR.
TEAM A
LashMystic III28,450
DynamoMystic I26,350
InfernusMystic I26,350
GravesSentinel VI25,245
CalicoSentinel VI25,200
Paige↔Sentinel VI25,000
TEAM B
CelesteMystic III28,000
KelvinMystic III28,000
AbramsMystic II27,700
HazeSentinel VI25,200
Deadman DannySentinel VI24,600
McGinnisSentinel III22,495
Comparison details
Observed wins: A 99/200 · B 101/200. Rough higher-point-side reference range: 42.6–56.4%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 156,595 · B 155,995. Sum gap 600; team-mean gap 100.
With a tested 65,000-point ceiling proxy the scoring gap would be 100. This does not change the reported points and is not confirmed private MMR.
TEAM A
Lady GeistMystic II27,650
Graves↔Mystic I26,350
Mo & KrillMystic I26,350
InfernusSentinel VI24,500
ApolloSentinel V23,500
MinaSentinel IV23,150
TEAM B
ParadoxMystic III28,000
ShivMystic I26,200
MirageMystic I26,060
LashSentinel VI25,620
PocketSentinel VI25,520
CelesteSentinel I19,500
Comparison details
Observed wins: A 103/200 · B 97/200. Rough higher-point-side reference range: 44.6–58.3%.
Same region, point-gap band and all-twelve-high status; nearby rank and player-point spread. Heroes/roles/parties not matched. All 18 examples excluded.
Raw sums: A 151,500 · B 150,900. Sum gap 600; team-mean gap 100.
With a tested 65,000-point ceiling proxy the scoring gap would be 100. This does not change the reported points and is not confirmed private MMR.
Duos
Keeping a duo together removes possible splits. That can raise the minimum public-point gap, but does not establish duos as the cause of these lower-rank gaps.
Europe: large-gap rate
North America: large-gap rate
More detected pairs: small observed edge
Median public-point difference on that side: EU -6.7; NA -23.3 points per player.
Can the same twelve be split more evenly with pair locks?
Asia
Match 136546918 · max player 39,6501,501.7actual team-mean point gapUnder recorded-badge restrictions: at least five hypothetical duos. Actual parties are unknown.
No qualifying pair detected. This is not confirmation of solos.
Oceania
Match 135857886 · max player 45,5701,495.8actual team-mean point gapUnder recorded-badge restrictions: at least five hypothetical duos. Actual parties are unknown.
No qualifying pair detected. This is not confirmation of solos.
Europe
Match 137688471 · max player 61,9201,494.2actual team-mean point gapUnder recorded-badge restrictions: at least five hypothetical duos. Actual parties are unknown.
2 prior-co-play pairs detected. Their locks permit a 19.2-point gap.
North America
Match 135224756 · max player 11,5001,491.7actual team-mean point gapNo solo/duo pattern makes this split point-optimal. Even the most restrictive pattern still allows 851.7 rather than 1,491.7.
No qualifying pair detected. This is not confirmation of solos.
Which rules are actually published?
Below Ascendant, with three prior captured games per player. Groups are not experimentally matched. The pair-lock test uses public points only; hypothetical hard type quotas or other private constraints could change it.
Published queue limits come from the cached official Matchmaking Update (30 July 2026), not a live enforcement test. No exact duo-MMR penalty or mandatory equal-party-count rule was identified.
Hero roles
44 heroes grouped by the saved client types. Development flags are not a live Ranked eligibility list.
assassin
9 heroes
Apollo
Calico
Drifter
Haze
Lash
Mirage
Pocket
Vyper
Yamato
brawler
10 heroes
Abrams
Bebop
Billy
Mo & Krill
Rat King
Shiv
Silver
Victor
Viscous
Warden
marksman
6 heroes
Deadman Danny
Holliday
Paradox
Venator
Vindicta
Wraith
mystic
13 heroes
BabaDev
Celeste
Graves
Grey Talon
Infernus
Lady Geist
McGinnis
Mina
Seven
Sinclair
Solomon
The Doorman
VioletDev
healer
6 heroes
Dynamo
Ivy
Kelvin
Nurse Harrow
Paige
Rem
Saved build 6861. Healer tags come from explicit client data where the API field is missing.
Team building
Queue and party eligibility
Published restrictions limit who can queue together. The actual ready population, full hero choices, latency constraints and queue ages are not in finished-match data.
The selected twelve
We can see who played, their initial public points/badges, and recorded pregame/played heroes. We cannot recover the whole candidate pool or actual arrival order.
Split into two teams
Points, composition and possible pair locks help explain splits. High-end score compression is supported in the relevant high-rank sample, not a universal answer for lower ranks.
Search, changes and final lock
Whether the backend drafts, optimizes, repairs or uses different private ratings is not uniquely identified. Exact coefficients and finalization/revalidation rules remain unknown.
Data & sources
111,037 complete, calibrated Normal/Ranked games from public Deadlock API snapshot 1552, public rank interval 2, captured through 15 November 2026. Large gap: over 1,000 team-average public points. These are not private MMR or exact match odds. All valid extremes remain; metric ties have no higher side.
Example matches were chosen from inputs, not wins. Co-play pairs are not confirmed parties. Roles use client build 6861; the 65k ceiling is a tested high-rank proxy, not recovered code. Private ratings, queue pools and complete constraints remain unknown. Reference ranges are not adjusted for repeat players.
Download lineups and results · Provenance
Valve hero and rank artwork via public Deadlock API assets.