UFC - Ultimate Japan: Predictions & Analysis
UFC - Ultimate Japan lands on Sunday, December 21, 1997 in Yokohama, Kanagawa, Japan with 7 bouts on the card. Below is our fight-by-fight breakdown, combining Elo ratings, rolling statistical trends, style matchup data, and betting market context into a pick for every bout.
Quick Picks
| Matchup | Pick | Confidence | Prob |
|---|---|---|---|
| Randy Couture vs Maurice SmithHeavyweight | Randy Couture | Strong | 76% |
| Kazushi Sakuraba vs Marcus SilveiraHeavyweight | Kazushi Sakuraba | Toss-up | 53% |
| Vitor Belfort vs Joe CharlesHeavyweight | Vitor Belfort | Confident | 65% |
| Frank Shamrock vs Kevin JacksonLightweight | Kevin Jackson | Lean | 57% |
| Kazushi Sakuraba vs Marcus SilveiraHeavyweight | Kazushi Sakuraba | Lean | 61% |
| David Abbott vs Yoji AnjoHeavyweight | David Abbott | Lean | 60% |
| Tra Telligman vs Brad KohlerHeavyweight | Brad Kohler | Toss-up | 51% |
Fight-by-Fight Breakdown
Randy Couture vs Maurice Smith
The Heavyweight matchup features Randy Couture (16-7) taking on Maurice Smith (4-2).
There's a real Elo separation here: Couture at 1248 versus Smith at 1155. That 93-point gap typically reflects a meaningful difference in recent quality of competition and results. Couture rides a 3-fight win streak into this one.
Stylistically this is Couture's wrestler game against Smith's knockout artist approach. Couture looks to close distance and put the fight on the mat, while Smith is patient on the feet, timing counters and loading up when he sees openings. Historically these archetypes are dead-even when they collide.
A few statistical edges stand out. Couture throws significantly more leather — a 0.9 sig. strike per minute gap. Couture is far more active with takedowns, averaging 3.6 more per 15 minutes. Couture has tighter striking defense, making opponents miss more often.
The Pick: Randy Couture over Maurice Smith. The model is firm on this one: Couture at 76%.
Kazushi Sakuraba vs Marcus Silveira
The Heavyweight matchup features Kazushi Sakuraba (0-0) taking on Marcus Silveira (0-0).
Sakuraba is rated at 1111 — 221 points above Silveira's 890. Gaps this large usually mean one fighter has been consistently beating better opponents.
A few statistical edges stand out. Silveira throws significantly more leather — a 0.0 sig. strike per minute gap. Silveira is far more active with takedowns, averaging 0.0 more per 15 minutes. Silveira has tighter striking defense, making opponents miss more often.
The Pick: Kazushi Sakuraba over Marcus Silveira. This is essentially a pick'em. The model nudges toward Sakuraba at 53%, but there's almost nothing separating these two.
Vitor Belfort vs Joe Charles
The Heavyweight matchup features Vitor Belfort (15-9) taking on Joe Charles (2-1).
Belfort is rated at 1255 — 179 points above Charles's 1076. Gaps this large usually mean one fighter has been consistently beating better opponents.
A few statistical edges stand out. Belfort throws significantly more leather — a 4.1 sig. strike per minute gap. Belfort is far more active with takedowns, averaging 1.4 more per 15 minutes. Charles has tighter striking defense, making opponents miss more often.
The Pick: Vitor Belfort over Joe Charles. We're leaning Belfort here at 65%, a solid but not overwhelming edge.
Frank Shamrock vs Kevin Jackson
The Lightweight matchup features Frank Shamrock (4-0) taking on Kevin Jackson (2-1).
Shamrock is rated at 1480 — 503 points above Jackson's 977. Gaps this large usually mean one fighter has been consistently beating better opponents. Shamrock rides a 4-fight win streak into this one.
A few statistical edges stand out. Jackson throws significantly more leather — a 2.3 sig. strike per minute gap. Jackson is far more active with takedowns, averaging 13.7 more per 15 minutes. Shamrock has tighter striking defense, making opponents miss more often.
The Pick: Kevin Jackson over Frank Shamrock. The model gives Jackson a slight nod at 57% — this could easily go either way.
Kazushi Sakuraba vs Marcus Silveira
The Heavyweight matchup features Kazushi Sakuraba (0-0) taking on Marcus Silveira (0-0).
Sakuraba is rated at 1111 — 221 points above Silveira's 890. Gaps this large usually mean one fighter has been consistently beating better opponents.
A few statistical edges stand out. Silveira throws significantly more leather — a 0.0 sig. strike per minute gap. Silveira is far more active with takedowns, averaging 0.0 more per 15 minutes. Silveira has tighter striking defense, making opponents miss more often.
The Pick: Kazushi Sakuraba over Marcus Silveira. The model gives Sakuraba a slight nod at 61% — this could easily go either way.
David Abbott vs Yoji Anjo
The Heavyweight matchup features David Abbott (8-9) taking on Yoji Anjo (0-2).
The Elo gap here is razor-thin — Abbott at 777, Anjo at 767. On paper, this is as close to a coin flip as it gets.
A few statistical edges stand out. Abbott throws significantly more leather — a 2.3 sig. strike per minute gap. Abbott is far more active with takedowns, averaging 1.1 more per 15 minutes. Anjo has tighter striking defense, making opponents miss more often.
The Pick: David Abbott over Yoji Anjo. The model gives Abbott a slight nod at 60% — this could easily go either way.
Tra Telligman vs Brad Kohler
The Heavyweight matchup features Tra Telligman (1-3) taking on Brad Kohler (0-1).
Kohler is rated at 1041 — 194 points above Telligman's 848. Gaps this large usually mean one fighter has been consistently beating better opponents.
A few statistical edges stand out. Telligman throws significantly more leather — a 0.8 sig. strike per minute gap. Kohler is far more active with takedowns, averaging 0.0 more per 15 minutes. Kohler has tighter striking defense, making opponents miss more often.
The Pick: Brad Kohler over Tra Telligman. This is essentially a pick'em. The model nudges toward Kohler at 51%, but there's almost nothing separating these two.
Methodology
Predictions are generated by our ensemble model combining LightGBM (65%) and CatBoost (35%), trained on every UFC fight since 1994. The model uses 23 features including Elo ratings, rolling 5-fight statistical averages, style matchup history, physical attributes, and market odds when available.
On our held-out test set (402 fights from January-September 2023), the model achieves 63.4% accuracy with a log-loss of 0.626. High-confidence picks (>75% probability) hit at 82.7%. For full model transparency, visit our Model page.