2.5. Demographic Asymmetry: Character Population, Matchup Familiarity, and Ecological Validity

A final, critical variable governing observational fighting game analytics is demographic distribution: the raw frequency with which specific characters exist within the competitive ecosystem. An analysis that assumes uniform distribution across all 19 combatants ignores the sociometric realities of the arcade cabinet. 

To establish the true demographic baseline of Japanese arcade culture, this study conducted an exhaustive census of the official Game Spot Versus Danisen registry, encompassing all 2,533 registered character cards from 1-Dan through Busei. The resulting population hierarchy exposes extreme asymmetry:

Cross-referencing the Danisen population against the match volume of the @necrocompendium archive establishes two foundational insights:

1. Absolute Ecological Validity (Arcade Population vs. Channel Volume)

To verify whether the @necrocompendium archive suffered from uploader selection bias, we mapped the relative rank of character popularity in the Danisen against the actual set volume recorded in the channel. The visual alignment demonstrates remarkable systemic fidelity:

Notice the grey lines indicating zero rank movement for apex fixtures: Chun-Li (#3), Dudley (#5), and Yun (#6) occupy identical structural demand in both populations. This confirms that the Necro Compendium did not curate favorable matchups; it absorbed the full, unvarnished brunt of Tokyo's authentic competitive meta.

• The Ken Parity Effect: With 427 registered pilots (nearly 17% of all players in Japan), Ken was ubiquitous. Japanese Necro mains possessed thousands of hours of muscle memory against Ken's options, neutralizing the theoretical 3–7 deficit predicted by Arcadia down to a functional 4.88 – 5.12 parity.
• The Low-Density Hazard (Oro): Conversely, Oro accounts for a meager 56 registered pilots (2.21%). Rare matchups deny specialists regular lab time. When confronted by Oro’s unconventional hurtboxes and double-jump angles, the lack of familiarity amplified character-specific disadvantages into a catastrophic 32.8% win rate.
• The Sean Purification: With a negligible 27 registered players in the entire nation, no "casual" Sean existed in the Danisen. The few competitors entering the arcade with Sean were mechanical savants operating at peak parry execution, explaining why Necro failed to achieve the theoretical 7–3 blowout predicted on paper.

4.6. Correction 6: Quarantining the Mirror Matches (The Civil War Exclusion)

The Problem: Mathematical Tautology

In our raw collection, 33 sets feature Necro vs. Necro. Whenever a mirror match concludes, the game engine inevitably declares: NECRO WINS. In naive character matchup tables, this generated an absurd statistical artifact: Necro holding an inflated win rate against himself caused by optical character recognition capturing Player 1 victories. By definition, in any closed mirror match, the true character win rate is strictly 50.0% (1 victory and 1 defeat per set).

The Methodological Shielding

All 33 mirror sets are completely expunged from the 18-character asymmetric matchup matrix. They are quarantined exclusively into a dedicated sub-chapter—The Civil War—analyzing human head-to-head records among Necro specialists.

4.7. The Triad of Rankings: Empirical, Smoothed, and Latent Skill

Rather than forcing the entire dataset into a single, arbitrary composite score, this study establishes a Three-Tier Ranking Framework. By presenting three distinct mathematical lenses side by side, we guarantee absolute transparency: fans of the arcade lore can examine the historical record exactly as it happened, while competitive analysts can inspect latent player ability stripped of schedule and character bias.

Tier 1: The Raw Observational Ranking (Empirical Record)

• Mathematical Definition: WR_raw = (W_j / N_j) × 100%
• Objective: Historical fidelity. It reports the unvarnished win/loss record of every competitor across the 3,049 cataloged sets without any algorithmic intervention.
• Interpretation: Answers the descriptive question: "What percentage of recorded sets did this player win in the arcade?"
• Known Limitations: Completely vulnerable to the Law of Small Numbers (e.g., inflating a 3–0 record to 100%) and blind to opponent strength.

Tier 2: The Regularized Baseline (Leave-One-Out Bayesian Shrinkage)

• Mathematical Definition: WR_B(−j) = (W_j + C · μ_(−j)) / (N_j + C) × 100%
• Where μ_(−j) is the Jackknife empirical prior computed across all sets excluding player j, and C = 20 represents the empirical stabilization constant (validated across sensitivity sweeps of C ∈ {10, 20, 40, 80}).
• Objective: Intuitive stability. It preserves the familiar 0–100% win-rate scale while eliminating micro-sample distortions without imposing ad-hoc ranking penalties.
• Interpretation: Answers the baseline evaluative question: "Assuming an unproven competitor regresses toward the collective Necro baseline, what is their expected long-run win rate?"
• Known Limitations: While it accounts for sample size variance, it remains agnostic to the specific human opponents faced and does not control for character matchup distribution.

Tier 3: The Latent Skill Ranking (L2-Regularized Logistic Regression)

Mathematical Definition:

An additive Generalized Linear Model based on the Bradley-Terry tournament framework, estimated via Ridge-regularized logistic regression (L2 penalty, C = 1.0) across all 3,049 discrete set events:

logit(P(Y_i = 1)) = ln( p_i / (1 − p_i) ) = β₀ + u_N(i) − v_O(i) + γ_C(i)

Where:

• u_N(i) = Latent skill parameter of the Necro player.
• v_O(i) = Latent skill parameter of the Human Opponent.
• γ_C(i) = Matchup resistance parameter of the Opponent Character.

• Objective: Causal isolation of competitive ability. By solving the entire 3,049-set network simultaneously in log-odds space, the model statistically disentangles three competing variables:
  • Player Competence (u_N): The intrinsic talent of the Necro player.
  • Opponent Resistance (v_O): How dangerous the human opponent was across the broader dataset.
  • Character Structural Advantage (γ_C): The systemic matchup difficulty of the opposing character (e.g., absorbing the structural disadvantage of Chun-Li or Yun).
• Interpretation: Answers the ultimate competitive question: "Holding character matchup difficulty and opponent caliber constant, which Necro player generated the highest probability of victory per set?"
• Key Advantage: A competitor whose raw win rate is suppressed by facing Meishou and Kensei titans playing Chun-Li is mathematically rewarded, whereas an inflated win rate built against low-dan opponents playing low-tiers is deflated to its true latent value.

Section 5: Macro-Level Empirical Findings and the Competitive "Reality Check"

5.1. The Definitive Macro Census: Archive Total vs. Asymmetric Combat Record

To maintain strict scientific consistency with the methodological quarantine established in Section 4.6, the macroscopic evaluation of our dataset must distinguish between two distinct populations: the total archival corpus and the asymmetric competitive combat record.

1. The Total Archival Volume (N = 3,049 Sets)

Across the 2,364 valid video sessions cataloged in the @necrocompendium repository, the absolute physical volume of competitive play encompasses 3,049 discrete sets. Of these, 33 sets represent closed mirror matches (Necro vs. Necro). As established, including mirror matches in character viability equations generates a mathematical tautology (each mirror set simultaneously guarantees a victory and a defeat for the character).

2. The Asymmetric Combat Record (Necro vs. The 18 External Characters)

Expunging the 33 mirror sets isolates the true, unadulterated baseline of Necro's performance against the external roster of Street Fighter III: 3rd Strike:

Remarkably, whether evaluated strictly through asymmetric combat (44.13% across 3,016 sets) or via the raw gross archive including closed mirror matches (44.15% across 3,049 sets), the empirical reality converges within a mere two-hundredths of a percentage point — a divergence so negligible it erases any methodological ambiguity regarding mirror inclusion. Against the world's sharpest arcade scene, Necro operated within less than six percentage points of dead-even competitive parity.

The Architecture of Play: Session Format Breakdown

Examining the distribution of match scores across all recorded video sessions reveals the structural nature of Japanese arcade play:

The 90% Reality: The Single-Credit Gauntlet

A critical observational finding is that 2,142 out of the ~2,364 valid video sessions (90.60%) were isolated single-set arcade encounters (1–0 or 0–1).

This completely refutes any assumption that our dataset is artificially dominated by long, casual exhibition sets. Over 90% of the entire archive represents the raw, unforgiving reality of Japanese coin-op culture: one 100-yen coin, two rounds to win, winner stays on, loser walks away.

5.2. The Competitive "Reality Check": Contextualizing 44.15% in Tokyo Arcades (2000–2026)

In modern esports discourse dominated by online matchmaking algorithms, an aggregate win rate of 44.13% might superficially appear to confirm that Necro is a weak, sub-optimal character. In competitive games, sub-45% win rates are frequently branded as unplayable "garbage tier."

However, evaluating this number within the historical, economic, and sociometric context of the Japanese arcade scene between 2000 and 2026 reveals the exact opposite: 44.13% across 3,049 sets is a staggering demonstration of competitive viability.

1. The 100-Yen Economic Filter

Unlike modern online ranked modes with infinite free rematches, every match at Game Spot Versus or Game Newton carried direct financial stakes. Players did not insert coins to lose.

Consequently, the opponent pool was intensely optimized: casual character experimentation was weeded out, and the machine queues were heavily dominated by players utilizing the strongest, most executionally safe tournament characters available (Chun-Li, Yun, Ken, Dudley). Necro players were not competing against a representative cross-section of the cast; they were running a permanent gauntlet against Tokyo's most dedicated specialists armed with top-tier tools.

2. The Structural Drag of the "Two-Tops"

As subsequent chapters will demonstrate, Necro suffers from structural, near-unplayable matchup deficits against the two apex characters of the game: Chun-Li (29.1% win rate) and Yun (36.5% win rate).

Because Chun-Li and Yun alone represent over 17% of all recorded sets in the database, their extreme mechanical dominance exerts a massive downward mathematical gravity on Necro's global record.

• To maintain an overall win rate of 44.15% while absorbing catastrophic, high-volume losses against Chun-Li and Yun, Necro pilots were mathematically required to hold near-even parity (48%–52%) against the rest of the high/mid-tier field (Ken, Urien, Ibuki, Elena) and establish dominant positive win rates against low-tiers.

3. The Verdict: The Ultimate Competitive Gatekeeper

Maintaining 44.15% across 3,049 documented arcade sets against the sharpest scene on earth proves that Necro was never an unviable joke character like Sean or Twelve.

Necro was the ultimate high-execution gatekeeper of 3rd Strike:

• He lacked the universal neutral safety of Chun-Li or the explosive damage robbery of Yun's Genei-jin.
• Every victory required flawless spacing, high-risk parry commitments, and encyclopedic matchup discipline.
• Yet, despite fighting against the structural design of the game on a weekly basis, Japan’s dedicated Necro corps stood only 5.87 percentage points away from true coin-flip parity (50/50).

The 44.15% figure does not document a failure; it documents the quantitative boundary of human execution pushing an asymmetrical, flawed character to the absolute edge of competitive viability.

Chapter 2: Japanese Theory vs. Arcade Reality — The Arcadia Matrix Demystified

For over fifteen years, the definitive blueprint for competitive matchups in Street Fighter III: 3rd Strike was the official 19-character matrix published in Monthly Arcadia. In that theoretical ledger, Row 13 belonged to Necro.

The magazine evaluated Necro with a total score of 80 points (−10 differential) across the 18 external matchups, predicting a bleak competitive life: Necro was assigned only five winning matchups (Elena, Remy, Hugo, Twelve, Sean), three dead-even matchups (Ibuki, Alex, Q), and ten losing matchups, including catastrophic 3–7 and 2–8 ratings against the upper half of the roster.

By cross-referencing this classical Japanese baseline with 3,016 non-mirror sets from the @necrocompendium archive, we can finally evaluate theory against empirical reality.

2.1. The Master Comparative Ledger

The table below contrasts the theoretical score assigned to Necro by the Arcadia editorial board against the observed empirical record across 3,016 non-mirror sets, sorted by total volume of sets played:

Not too shabby, Arcadia, not too shabby...

2.2. The Three Divergence Archetypes

Analyzing the deviations (Δ) between theoretical prediction and empirical reality reveals that the 18 matchups divide cleanly into three competitive archetypes:

2.3. Character-by-Character Forensic Breakdown

1. Urien — Volume: 353 Sets | 171–182 (48.4%) | Ratio: 4.8 – 5.2 | Arcadia: 4 – 6 (Δ = +0.8)

• The Context: Urien is the single most played matchup in the entire @necrocompendium archive, accounting for 353 sets.
• The Reality: Arcadia predicted a solid advantage for Urien (4–6), largely influenced by the terror of Aegis Reflector unblockables. The empirical data proves this was an exaggeration: across 352 high-level sets, Necro sits at 48.3%, transforming the matchup into an effective 5–5 coin-flip.
• Tactical Explanation: While Aegis Reflector remains devastating, Necro’s drills and the constant threat of eating 6x db.HP in the corner keep Urien from getting too confident. On top of that, Necro’s Snake Fang leaves Urien no breathing room when he whiffs to build meter, and SA3 punishes—even from range—stray fireballs, shoulder tackles, and the like, resulting in an even slugfest over long samples.

2. Ken — Volume: 299 Sets | 146–153 (48.8%) | Ratio: 4.9 – 5.1 | Arcadia: 3 – 7 (Δ = +1.9)

• The Context: The single largest theoretical error in the entire Arcadia matrix.
• The Reality: Monthly Arcadia assigned this as a devastating 3–7 disadvantage, treating Ken as an insurmountable barrier. Across 299 sets against the strongest Kens in Japan (including Matsuken), Necro won 146 sets (48.8%).
• Tactical Explanation: Ken relies on cr.MK > SA3 confirms. However, Necro's s.MP can safely challenge Ken outside of his cr.MK range and shut down dash-ins. Ken also can't afford to throw out thoughtless reversal shoryukens—especially in the corner, where it's a guaranteed stun. It's not an easy matchup, but it's a long way from the slaughter Arcadia predicted.

3. Chun-Li — Volume: 296 Sets | 86–210 (29.1%) | Ratio: 2.9 – 7.1 | Arcadia: 3 – 7 (Δ = −0.1)

• The Context: The highest volume matchup against an S-tier opponent.
• The Reality: Absolute predictive precision. Arcadia called it 3–7; 296 real-world arcade sets produced 2.9 – 7.1 (29.1%).
• Tactical Explanation: Chun-Li is Necro’s structural executioner. Her crouching fierce cleanly swats drill attempts, her back-fierce and standing fierce completely snuff Necro’s extended hurtboxes, and her SA2 punishes any extended limb on reaction. Necro must take monumental parry risks merely to establish offense, confirming that Japanese theory was entirely correct regarding this matchup.

4. Ryu — Volume: 257 Sets | 110–147 (42.8%) | Ratio: 4.3 – 5.7 | Arcadia: 4 – 6 (Δ = +0.3)

• The Context: High-volume legacy matchup against legendary Ryu masters (Vanao, Kuni).
• The Reality: Arcadia predicted a 4–6 disadvantage; the empirical record sits at 4.3 – 5.7.
• Tactical Explanation: The only Ryu that can genuinely scare Necro is Denjin Ryu—and that's most of the channel's matchups. Denjin Ryu is a problem for every character in the game, and Necro is no exception. But since Ryu doesn't pack as many extra tools as Ken, Necro has more room to establish his own offense.

5. Dudley — Volume: 249 Sets | 107–142 (43.0%) | Ratio: 4.3 – 5.7 | Arcadia: 3 – 7 (Δ = +1.3)

• The Context: Dudley represents the archetype of explosive 3-凶 burst damage.
• The Reality: Arcadia predicted a one-sided 3–7 beating. Empirical data reveals a far more resilient 4.3 – 5.7 performance.
• Tactical Explanation: Dudley destroys Necro if he achieves point-blank corner momentum. However, Dudley possesses no projectile and must advance on the ground. Necro’s standing fierce can create a formidable mid-screen wall that Dudley cannot effortlessly bypass without taking significant parry gambles.

6. Yun — Volume: 233 Sets | 85–148 (36.5%) | Ratio: 3.6 – 6.4 | Arcadia: 2 – 8 (Δ = +1.6)

• The Context: The second head of the S-tier duopoly.
• The Reality: Arcadia rated this as an unplayable 2–8 massacre. While Yun undeniably dominates, real-world Necros extracted an observed 36.5% win rate (3.6 – 6.4) across 233 sets.
• Tactical Explanation: Genei-jin (SA3) remains oppressive, frequently ending rounds off a single divekick confirm. However, Japanese Necros developed specialized anti-Yun protocols, pushing the matchup performance roughly 16 percentage points above the theoretical floor.

7. Makoto — Volume: 204 Sets | 79–125 (38.7%) | Ratio: 3.9 – 6.1 | Arcadia: 4 – 6 (Δ = −0.1)

• The Context: The terrifying touch-of-death command grab specialist.
• The Reality: Near-perfect predictive accuracy. Arcadia predicted 4–6; the data delivered 3.9 – 6.1 (38.7%).
• Tactical Explanation: Makoto’s Karakusa command grab into Abare Tosanami or Seichusen Godanzuki deletes 70% of Necro's lifebar. On top of that, Makoto’s hitboxes last an eternity while Necro’s last a single frame—two at most. But Necro also has some nasty combos against Makoto in the corner, and one slip-up can cost him a round.

8. Hugo — Volume: 164 Sets | 98–66 (59.8%) | Ratio: 6.0 – 4.0 | Arcadia: 6 – 4 (Δ = ±0.0)

• The Context: The premier giant-grappler matchup.
• The Reality: The single most accurate prediction in the entire matrix. Arcadia estimated 6–4 (+10.0% advantage); 164 sets yielded 59.8% (Ratio 6.0 – 4.0).
• Tactical Explanation: Hugo’s massive hurtbox is tailor-made for Necro. At least in this matchup, Necro’s elastic limbs are actually useful and cause Hugo plenty of trouble. On top of that, Hugo struggles to parry multi-hitting electric shocks, and Necro can confirm Magnetic Storm with immense reliability against Hugo’s colossal frame.

9. Yang — Volume: 149 Sets | 60–89 (40.3%) | Ratio: 4.0 – 6.0 | Arcadia: 3 – 7 (Δ = +1.0)

• The Context: Yun’s twin, operating with Seiei Enbu (SA3).
• The Reality: Arcadia assigned a harsh 3–7; the empirical data shows Necro holding a respectable 4.0 – 6.0 (40.3%).
• Tactical Explanation: While Yang’s divekicks and rekka confirms create immense pressure, he lacks Yun’s sheer unblockable robbery potential.

10. Alex — Volume: 136 Sets | 76–60 (55.9%) | Ratio: 5.6 – 4.4 | Arcadia: 5 – 5 (Δ = +0.6)

• The Context: A classic mid-tier brawler encounter.
• The Reality: Arcadia predicted an even 5–5; the data reveals a distinct Necro advantage (55.9% / Ratio 5.6 – 4.4).
• Tactical Explanation: With no good anti-air, Necro can abuse drills and put Alex in the corner, where Alex can't afford to get careless or he eats 6x db.HP too. On the other hand, if Alex manages to close the gap and get on top of Necro, Necro melts fast—and it suddenly looks like Alex has more stun damage than Necro.

11. Ibuki — Volume: 122 Sets | 64–58 (52.5%) | Ratio: 5.2 – 4.8 | Arcadia: 5 – 5 (Δ = +0.2)

• The Context: High-mobility vortex ninja matchup.
• The Reality: Arcadia predicted an even 5–5; 122 sets resulted in 52.5% (Ratio 5.2 – 4.8).
• Tactical Explanation: A pure test of execution. Ibuki’s Raida and target combos can loop into dizzy setups, but her low vitality makes her exceptionally fragile against Necro’s Magnetic Storm confirms and corner juggle damage.

12. Oro — Volume: 116 Sets | 38–78 (32.8%) | Ratio: 3.3 – 6.7 | Arcadia: 4 – 6 (Δ = −0.7)

• The Context: The single biggest hidden catastrophe in the database.
• The Reality: Arcadia treated Oro as a manageable 4–6 disadvantage. The empirical data reveals a catastrophic collapse to 32.8% (Ratio 3.3 – 6.7)—making Oro Necro’s second-worst matchup after Chun-Li.
• Tactical Explanation: Oro is a structural nightmare. His small hurtbox causes Necro's standard poke strings to whiff. His double jump invalidates Necro’s anti-air spacing, and Yagyou Dama generates inescapable unblockable chip damage setups that Necro’s wide hurtbox cannot escape. That's no exaggeration—the data proves this is one of the worst matchups for Necro, and it gets even worse because the number of Oro players is tiny, so when one shows up, he just bodies Necro.

13. Akuma (Gouki) — Volume: 115 Sets | 47–68 (40.9%) | Ratio: 4.1 – 5.9 | Arcadia: 4 – 6 (Δ = +0.1)

• The Context: The glass-cannon vortex matchup.
• The Reality: Direct validation. Arcadia estimated 4–6; 115 sets yielded 40.9% (Ratio 4.1 – 5.9).
• Tactical Explanation: Akuma’s air fireballs, teleport mixups, and pressure present lethal threats. However, Akuma possesses the lowest health pool in the game; a single confirmed SA3 or corner juggle wipes out half his vitality bar, keeping the contest around a stable 4–6 mark.

14. Elena — Volume: 99 Sets | 50–49 (50.5%) | Ratio: 5.1 – 4.9 | Arcadia: 6 – 4 (Δ = −0.9)

• The Context: A classic mid-tier battle.
• The Reality: Arcadia predicted a clear Necro favor (6–4). The empirical record shows a dead-even split: 50.5% across 99 sets (Ratio 5.1 – 4.9).
• Tactical Explanation: Arcadia underestimated Elena’s poke recovery and Healing (SA1) resource engine. While Necro out-spaces her, Elena’s wonky hurtboxes favor Necro’s pressure in the corner, enforcing a true 50/50 equilibrium.

15. Remy — Volume: 90 Sets | 43–47 (47.8%) | Ratio: 4.8 – 5.2 | Arcadia: 6 – 4 (Δ = −1.2)

• The Context: Charge zoning keep-away.
• The Reality: Arcadia predicted a 6–4 advantage for Necro. The data reveals a slight disadvantage (47.8% / Ratio 4.8 – 5.2).
• Tactical Explanation: Arcadia assumed Necro’s limbs could easily bypass Remy’s Light of Virtue fireballs. In practice, high-level Remys utilize high/low zoning alongside flash kicks, forcing Necro into high-risk forward parries to close the gap.

16. Q — Volume: 75 Sets | 37–38 (49.3%) | Ratio: 4.9 – 5.1 | Arcadia: 5 – 5 (Δ = −0.1)

• The Context: Heavyweight armor and taunt defense.
• The Reality: Flawless predictive alignment. Arcadia predicted 5–5; the data produced 49.3% (Ratio 4.9 – 5.1).
• Tactical Explanation: Q's defensive buffs from taunts give him immense survivability, while his command grab deals severe damage. Necro easily keeps Q at bay with pokes, but finishing off an armored Q requires extended neutral interactions, locking the matchup at an even split.

17. Twelve — Volume: 35 Sets | 23–12 (65.7%) | Ratio: 6.6 – 3.4 | Arcadia: 6 – 4 (Δ = +0.6)

• The Context: The bottom-tier air-dash matchup.
• The Reality: Strong confirmation of dominance. Arcadia estimated 6–4; observed play delivered 65.7% (Ratio 6.6 – 3.4).
• Tactical Explanation: Twelve’s low damage output, erratic air dashes, and weak reversal options allow Necro to dominate neutral. Necro’s anti-air pokes cleanly swat Twelve out of the sky, turning this into one of Necro’s most reliable hunting grounds.

18. Sean — Volume: 24 Sets | 11–13 (45.8%) | Ratio: 4.6 – 5.4 | Arcadia: 7 – 3 (Δ = −2.4)

• The Context: The undisputed worst character in 3rd Strike.
• The Reality: The most bizarre empirical anomaly in the entire database. Arcadia rated this as Necro’s easiest matchup (7–3). In practice across 24 sets, Necro holds a losing record (45.8% / Ratio 4.6 – 5.4).
• Tactical Explanation: This divergence is primarily a product of the "Craftsman Selection Bias" (described in Section 2.4). In high-level Tokyo arcades, nobody plays Sean casually; the only competitors who pick Sean are supreme mechanical savants operating at the outer limits of parry execution. Across a modest sample of 24 sets, these elite Sean craftsmen achieved parity against Necro, subverting baseline character math.

Chapter 4: The Pantheon of Necro — The Triad of Rankings and the Judgment of the Players

In fighting game lore, player reputations are routinely forged on word-of-mouth, localized tournament runs, and highlight clips. A player who wins a local weekly tournament with a mid-tier character is quickly mythologized as a savant, while a dedicated grinder maintaining an even record against premier competition is overlooked as unremarkable.

To deliver an honest, scientifically grounded evaluation of the world’s premier Necro pilots, we deploy our Three-Tier Evaluation Framework. By moving sequentially from the Raw Observational Record to Leave-One-Out Bayesian Smoothing, and culminating in an L2-Regularized Bradley-Terry Latent Skill Model, we systematically strip away the distortions of sample size, schedule difficulty, and character matchup penalties.

The threshold for inclusion in this primary evaluation is set at a minimum of 15 discrete sets, ensuring every ranked competitor possesses a meaningful sample across the archive.

4.1. Ranking 1: The Raw Observational Record & Wins Above Expected (WAE)

The first tier presents the empirical reality of the arcade floor: exactly how many sets each competitor won and lost, their raw win rate (W/N), and their Wins Above Expected (WAE).

The Intuition: "Net Profit Above the Baseline"

Raw win rate can be deeply misleading. A 50% record against a cast of low-tiers is standard, but maintaining a 36% win rate against a gauntlet of Chun-Lis is an act of competitive heroism.

To measure this, we introduce Wins Above Expected (WAE). WAE measures net competitive surplus:

If an average Necro played this pilot’s exact distribution of character matchups, how many sets would they be expected to win? And how many surplus victories did this specific pilot deliver beyond that baseline?

Any positive WAE represents pure, unadulterated value created for the character where community baseline expectation predicted defeat.

*Note: Kuroda is quarantined as a multi-character guest outlier.

The Revelation: The Monumental Value of Sugiyama (+19.39 WAE)

The raw win rate table immediately exposes why evaluating volume solely through percentages fails.

At 49.21%, Sugiyama sits in 7th place in raw win rate. Yet, look at his WAE: +19.39 surplus victories.

• Across his 441 cataloged sets, expected performance for an average pilot facing his exact character gauntlet predicted roughly 197.61 wins. Sugiyama delivered 217.
• He generated 1.65× the total surplus value of any other dedicated Necro main in history (Ikeyama sits at +11.75; PinoAB7 at +8.05; Konitan at +7.86).
• Sugiyama is not a 50/50 coin-flip player; he is the single greatest creator of competitive surplus in the character's history. He extracted water from stone against hundreds of top-tier opponents where the community baseline predicted failure.

4.2. Ranking 2: The Leave-One-Out Bayesian Stress Test (Stability Across C = 10…80)

To eliminate the "Aaron Fallacy" and small-sample luck without suffering from the circularity of pulling a player toward their own data, we implement Leave-One-Out (Jackknife) Empirical Bayes.

The Intuition: "The Tank vs. The Plastic Toy"

In sports data science, parameter sensitivity tests act as a durability test:

• If a competitor's performance is real, running their record through increasing degrees of shrinkage (C = 10, 20, 40, 80) will barely move their rating. They are a battleship: their sample size anchors them firmly in place.
• If a competitor's record is an artifact of short-term variance, tightening the statistical grip (C = 80) will cause their rating to collapse toward the community baseline. They are a plastic toy tossed around by the math.

The Takeaway: The Unshakeable Foundations of the Veterans

• The High-Volume Monoliths: Sugiyama, Kusakabe, Tamu, Konitan, and PinoAB7 display virtually zero variance. Kusakabe moves by an infinitesimal 0.19% across all four regularization strengths (46.0% → 45.81%). Sugiyama moves by 0.72%. Their win rates are hardened steel.
• The Regression of the Fragile: Conversely, Victor (20 sets) plummets by 11.90 percentage points (60.0% → 48.10%). Aaron (53 sets) sheds 8.14 percentage points (58.5% → 50.36%).
• The Bayesian stress test proves that without massive sample sizes, high win rates are merely transient spikes.

4.3. Ranking 3: Latent Skill Modeling (L2-Regularized Bradley-Terry GLM)

While Bayesian smoothing solves sample size distortion, it remains blind to who was standing on the other side of the arcade cabinet.

To isolate latent pilot skill, we implement an L2-Regularized Logistic Regression Model (a generalized Bradley-Terry tournament network). Each of the 3,049 sets is treated as an independent trial, solving for player talent (β) while simultaneously controlling for opposing human talent and opponent character matchup resistance in an additive log-odds space.

The Intuition: "The Competitive Lie Detector"

Think of the L2 Model as an x-ray of competitive conditions:

• It discounts wins achieved against low-dan arcade casuals.
• It rewards competitors who took sets off Kensei and Meishou grandmasters.
• It absorbs the structural penalty of facing Chun-Li or Yun, measuring purely: holding character and opponent difficulty completely equal, who gave you the highest mathematical probability of winning a set?

4.4. Deep Analytical Findings from the Triad

1. The Resurrections: Yakkun and Uni

The L2 model delivers justice to two players whose reputations were suppressed by raw percentages:

• Yakkun (+0.408 Beta): In raw win rate, Yakkun sat in a mediocre 13th place with a losing record of 43.86%. But the L2 model cross-referenced his opponents against the Versus Danisen registry and discovered an astonishing reality: Yakkun faced a Strength of Schedule (SoS) of 3.15x—the absolute highest in the history of the channel. Yakkun was fed almost exclusively into the meatgrinder of Kensei and Meishou royalty (Tominaga, Matsuken, Nuki, Vanao). The model recognizes this, elevating him into the Top 8 world-class tier.
• Uni (+0.572 Beta): Holding a 47.53% raw rate, Uni surges into 3rd place among specialists (trailing only Aaron and Sugiyama) because his 2.84x SoS was second only to Yakkun. Uni proved that his ~48% was forged against the sharpest arcade killers in Tokyo.

2. The Contextualization of Aaron (+0.604 Beta | SoS 1.59x)

Aaron sits at the top of the specialist Beta ratings (+0.604) with a strong raw win rate of 58.49%. However, the Danisen SoS metric provides the necessary critical nuance:

• While Sugiyama (2.21x SoS), Uni (2.84x SoS), and Ikeyama (2.40x SoS) fought through heavy concentrations of Danisen generals, Aaron operated against an SoS of 1.59x.
• Aaron was remarkably lethal and efficient, but his schedule was significantly softer than the Tokyo meatgrinder endured by the Game Spot Versus vanguard.

3. Ikeyama: The Surgical Master (+0.532 Beta | 52.00% Raw)

Across 200 sets, Ikeyama is the only high-volume Necro player in history to maintain a definitively positive raw win rate (52.00%) while simultaneously conquering a 2.40x SoS. He is the epitome of clinical execution—combining volume, positive conversion, and a top-tier opponent gauntlet.

4. Sugiyama: The Apex Titan (+0.579 Beta | 441 Sets | +19.39 WAE)

The synthesis of the three rankings cements Sugiyama as the undisputed sovereign of competitive Necro play:

• The Volume: 441 sets (accounting for 14.5% of all sets in the archive).
• The Opposition: A brutal 2.21x SoS facing Japan’s finest every week.
• The Model: A dominant +0.579 Beta rating alongside the world-record +19.39 WAE.
• Sugiyama did not merely play Necro; he sustained the character’s competitive legitimacy in Japan almost single-handedly.

5. Buba: The Tragic Workhorse (238 Sets | -0.800 Beta)

At the opposite end of the spectrum sits Buba. Holding 73 wins and 165 defeats (30.67%), Buba generated a severe deficit of -34.51 WAE.

Yet, in the lore of 3rd Strike, Buba commands profound respect. Despite suffering relentless defeats against an above-average schedule (1.71x SoS), Buba never switched characters, inserting coin after coin across 238 sets. He is the tragic patron saint of character loyalty—the warrior who absorbed defeats so that the Necro Compendium could exist.

Chapter 5: Taming the Apex — Matchup Specialization via WAE and the DNA of the Eight Greats

While aggregate rankings provide an invaluable macro evaluation of pilot competence, they mask the defining virtue of the fighting game specialist: matchup-specific mastery.

In high-level 3rd Strike, no competitor plays every character uniformly. One pilot may decipher the neutral spacing required to contest Chun-Li, while another develops specialized parry protocols against Yun’s Genei-jin.

By applying our Wins Above Expected (WAE) framework to character subsets, we eliminate the distortive "Aaron Fallacy" (where micro-samples masquerade as elite performance) and determine which Necro pilots generated true competitive surplus against the game's four most oppressive threats: Chun-Li, Yun, Ken, and Makoto. Following this, we dissect the individual Competitive DNA of the eight foundational pillars of the @necrocompendium archive.

5.1. Taming the Top Tiers: The WAE Specialization Breakdown

To qualify for evaluation against a specific top tier, competitors were filtered by a minimum threshold of 8 discrete sets against that specific character.

Analytical Highlights:

• The Chun-Li Ceiling (Sugiyama +4.28 WAE): WAE permanently corrects the historical record. Aaron’s superficial 50.0% win rate collapses to 4th place (+1.68) due to a compound statistical problem: a micro-sample of just 8 sets against the softest opposition in the pool (SoS 1.32x, Light/Casual). Sugiyama stands as the undisputed sovereign against Chun-Li: enduring 61 brutal sets against Tokyo’s premier Chun-Li mains (Raoh, Nuki, Ochibi, MOV), he generated +4.28 surplus victories, followed closely by Uni (+2.35 at 2.54x SoS) and Konitan (+2.22 at 2.88x SoS) — both of whom earned their surplus against High-Density veterans.
• The Yun Demolition (Konitan +7.69 & PinoAB7 +5.79 WAE): Across the entire character meta, the single most shocking empirical revelation is the complete subversion of Yun by Konitan and PinoAB7. While the global Necro baseline suffers a miserable 36.48% against Yun, Konitan posted a staggering 61.29% (19–12, +7.69 WAE at 1.60x SoS) and PinoAB7 delivered 57.14% (16–12, +5.79 WAE at 1.74x SoS). Both ran on mid-level Danisen competition rather than the elite Kensei gauntlet, but the sheer surplus remains historic: between them, they converted an ostensibly unplayable matchup into personal positive hunting grounds.
• The Ken Neutralizer (Kusakabe +3.84 & Sugiyama +3.59 WAE): Proving that the Ken matchup is true parity, Kusakabe dismantled Ken specialists at a 62.07% clip (18–11) — though his 1.45x SoS reveals a notably softer opposition profile (Light/Casual). By contrast, Sugiyama’s +3.59 was forged against a 2.91x High-Density gauntlet while absorbing a massive 50-set workload (28–22, 56.00%) — the heavier lift of the two on a per-set basis. Tamu also deserves a nod: 62.50% (10–6) against a 2.46x schedule is the most efficient Ken-tamer ratio in the archive.
• The Makoto Defense (PinoAB7 +3.74 WAE): Facing Makoto’s lethal touch-of-death command-grab setups, PinoAB7 achieved an astonishing 72.73% win rate (8–3) — and did so against a 2.43x High-Density opposition profile, neutralizing Makoto’s Karakusa offense through disciplined spacing against veterans rather than casuals. Kuroda’s 57.89% (11–8, +3.64) against a 4.50x Extreme Elite schedule is, on paper, the most statistically absurd Makoto result in the entire archive — but it remains quarantined as a guest outlier.

5.2. The Anatomy of the Eight Greats: Individual Competitive DNA

By isolating each pilot’s non-mirror dataset, we map the Competitive DNA of the eight foundational pillars of Japanese Necro history: identifying each player’s Favorite Prey (most sets won), their Kryptonite (most sets lost), and their Defining Human Rivalry.

1. Sugiyama (杉山) — The Unshakable Foundation

• Record: 214W – 222L across 436 sets (49.08%)
• Favorite Prey: Urien (36 Victories)
• Kryptonite: Chun-Li (39 Defeats)
• Defining Rival: RX [Urien] (24 Sets Disputed)

Sugiyama represents the spine of the entire archive. His career is defined by his epic, long-running war against RX, Japan’s premier Urien master. Sugiyama turned Urien into his primary source of competitive nutrition, extracting 36 total victories. However, as Game Spot Versus’ premier vanguard, he bore the brunt of Tokyo’s Chun-Li armada, absorbing 39 defeats against the game’s apex predator while keeping his global win rate within inches of true 50/50 parity.

2. Kusakabe (日下部) — The Classical Purist

• Record: 162W – 192L across 354 sets (45.76%)
• Favorite Prey: Urien (21 Victories)
• Kryptonite: Urien (26 Defeats)
• Defining Rival: Kuni [Ryu] (27 Sets Disputed)

Kusakabe’s profile presents a fascinating mechanical paradox: Urien was simultaneously his greatest hunting ground and his primary executioner. Across 47 sets against Urien, Kusakabe fought an unceasing, volatile war of attrition (21–26). His human narrative, however, was anchored by a legendary 27-set rivalry against Kuni, the classical Ryu general (Chishou), representing the quintessential Shoto vs. Zoner arcade duel.

3. Tamu (たむ) — The Methodical Sentry

• Record: 132W – 158L across 290 sets (45.52%)
• Favorite Prey: Urien (18 Victories)
• Kryptonite: Chun-Li (23 Defeats)
• Defining Rival: Furo [Yang] (18 Sets Disputed)

Tamu operated with cold, spacing-oriented discipline. Like his peers, Tamu feasted on Urien (18 wins), but his true legacy lies in his intense 18-set conflict against Furo, Tokyo’s most terrifying Yang strategist (Chishou). Tamu demonstrated elite resistance against Yang’s relentless Seiei Enbu divekick mixups, holding the line where other mid-tiers crumbled.

4. Konitan (こにたん) — The Danisen General & Yun Hunter

• Record: 132W – 151L across 283 sets (46.64%)
• Favorite Prey: Urien (20 Victories)
• Kryptonite: Chun-Li (29 Defeats)
• Defining Rival: Raoh [Chun-Li] (19 Sets Disputed)

Holding the official rank of Meishou (名将) alongside Sugiyama, Konitan’s record reflects the absolute pinnacle of arcade pressure. His defining rivalry was a grueling 19-set war against Raoh, one of Japan’s most celebrated Chun-Li pioneers. While Chun-Li claimed 29 sets from him, Konitan compensated by becoming the single greatest Yun hunter in history, shattering the Genei-jin matchup at an unthinkable 61.29% win rate.

5. Ikeyama (いけやま) — The Surgical Dudley Hunter

• Record: 103W – 95L across 198 sets (52.02%)
• Favorite Prey: Dudley (18 Victories)
• Kryptonite: Ken (16 Defeats)
• Defining Rival: Matsuken [Ken] (22 Sets Disputed)

Ikeyama stands alone as the only high-volume Necro pilot with a net-positive record (52.02%). His DNA diverges sharply from the rest of the pantheon: while others specialized against Urien, Ikeyama built his empire by dismantling Dudley (18 wins), completely shutting down the boxer’s forward momentum. His nemesis was Ken (16 losses), personified by an unforgettable 22-set gauntlet against Matsuken, the undisputed Ken god (Kensei) of Japan.

6. Uni (ウニ) — The Iron Fortress

• Record: 75W – 84L across 159 sets (47.17%)
• Favorite Prey: Hugo (7 Victories)
• Kryptonite: Ken (14 Defeats)
• Defining Rival: Boss [Makoto] (10 Sets Disputed)

Uni’s profile is defined by unmatched defensive fortitude. Holding a terrifying 2.84x Strength of Schedule, Uni spent his career battling the sharpest blades in Tokyo. He effortlessly collected free wins against Hugo, but was continually tested by Ken (14 losses) and his intense, high-risk 10-set clash with Boss, the fearsome Makoto master (Chishou).

7. PinoAB7 (ピノAB7) — The Tactical Anomaly

• Record: 102W – 114L across 216 sets (47.22%)
• Favorite Prey: Yun (16 Victories)
• Kryptonite: Oro (20 Defeats)
• Defining Rival: Key [Oro] (19 Sets Disputed)

PinoAB7 possesses the most radical, divergent DNA in the entire archive. Completely breaking character convention, Yun was his favorite prey (16 victories at 57.14%). PinoAB7 solved the most broken character in the game, repeatedly swatting divekicks and shutting down offense.

Conversely, his career was haunted by Oro (20 losses), driven almost entirely by an agonizing 19-set grudge match against Key, Tokyo's premier Oro specialist. PinoAB7 could dismantle god-tier Yuns with ease, but was repeatedly snared by Key’s Tengu Stone unblockables.

8. Buba (ブバ) — The Undying Workhorse

• Record: 73W – 165L across 238 sets (30.67%)
• Favorite Prey: Urien (10 Victories)
• Kryptonite: Chun-Li (26 Defeats)
• Defining Rival: Kuni [Ryu] (24 Sets Disputed)

Buba is the emotional heartbeat of the Necro Compendium. Operating at a severe mathematical disadvantage (30.67%), Buba was battered mercilessly by Chun-Li (26 losses).

Yet, Buba never wavered, fighting an unyielding 24-set rivalry against Kuni, one of Tokyo’s 智将 — Chishou generals. And here is the brutal truth of that rivalry: Buba won zero of those 24 sets. Not one. A full 0–24 shutout against one of the sharpest minds in the arcade — and he kept walking back to the cabinet anyway. Kuni saga is the real monument: the raw, unglamorous grit of the arcade grinder who refuses to abandon his character, regardless of how steep the mountain — or how one-sided the beatdown.

Chapter 6: The Human Ecosystem — Nemeses, Victims, and the Civil War

Fighting games are ultimately contested between human beings, not abstract character models. While aggregate matchup percentages describe mechanical parameters, the historical reality of Game Spot Versus was forged through long-term rivalries, psychological warfare, and player reputation.

In this chapter, we examine the human architecture of the archive across 3,016 non-mirror sets—identifying the definitive executioners (Nemeses) and the most loyal arcade rivals (Victims)—before isolating the 33 historical Mirror Match sets to resolve the internal hierarchy of the Necro Civil War.

6.1. The Hall of Nemeses: The Apex Executioners of Tokyo

To identify the greatest historical threats to the Necro corps, opponents were filtered by a minimum volume of 15 discrete sets against Necro, sorted by total sets won:

Analytical Breakdown:

• The Twin Towers of Tokyo (Kuni & Matsuken): Between them, Kuni (49 wins) and Matsuken (45 wins) extracted 94 set victories over the Necro archive.
  • Kuni now leads the Nemesis board outright: holding a terrifying 71.01% win rate across 69 sets, his fundamental spacing and denjin setups shut down Necro's neutral pokes, making him the single most suffocating Shoto matchup in Japanese history.
  • Matsuken operated as the ultimate high-volume crucible, fighting 68 total sets across two decades and asserting the dominance of his Kensei-ranked Ken (66.18%).
• The High-Efficiency Guillotines (Tominaga & Mester): While Kuni and Matsuken accumulated wins through pure volume, Tominaga and Mester operated with terrifying conversion rates:
  • Tominaga (Makoto — 77.14%): Winning 27 of 35 sets, Tominaga was an unavoidable lethal threat. If a Necro pilot committed a single spacing error, Tominaga’s Karakusa confirms ended the set.
  • Mester (Yun — 78.57%): Winning 22 of 28 sets, Mester demonstrated the absolute ceiling of Genei-jin pressure, converting nearly 8 out of every 10 encounters into victory.

6.2. The Faithful Rivals: The Historic "Victims"

In arcade fighting game culture, a "victim" is not an insult; it is a testament to dedication. These competitors were high-volume regulars who sat at the cabinet repeatedly, exchanging dozens of sets against Necro specialists:

Analytical Highlights:

• The Subjugation of Higa (Ibuki): Higa, one of Tokyo’s premier Ibuki strategists (Chishou), suffered 27 set defeats against Necro across 47 encounters (57.45% loss rate). Necro's disjointed limbs and defensive electric buffers consistently disrupted Ibuki’s low-vitality vortex.
• The RX Equilibrium (17–17): Across 34 documented sets, the legendary Urien pioneer RX and the Necro corps fought to an exact 50.00% dead-heat (17 wins, 17 losses), symbolizing the ultimate competitive stalemate between two of 3rd Strike's most iconic mechanical innovators.

6.3. The Civil War: Isolating the 33 Mirror Matches

As established in Chapter 4, mirror matches represent a mathematical tautology that corrupts character matchup charts. Quarantined into their own discrete laboratory, the 33 recorded sets of Necro vs. Necro resolve the ultimate internal arcade debate: Who reigned supreme when character asymmetry was removed?

Individual Mirror Match Performance (The Kings of the Mirror)

When aggregating each pilot's individual record strictly within the mirror laboratory, a striking internal hierarchy emerges: