1. Eric Lu (USA, 27, Fazioli): 22.43
2. Kevin Chen (Canada, 20, Steinway): 22.15
3. Zitong Wang (China, 26, Kawai): 21.45
4. Shiori Kuwahara (Japan, 29, Steinway): 21.30
5. Tianyao Lyu (China, 16, Fazioli): 21.30
6. Piotr Alexewicz (Poland, 25, Kawai): 21.03
7. Vincent Ong (Malaysia, 24, Kawai): 21.01
8. William Yang (USA, 24, Steinway): 20.91

Near finalists by cumulative: David Khrikuli, Tianyou Li, and Miyu Shindo finished just under the prize line.

How CMEAN Works in Plain Terms

1. Compute the mean of all juror marks for a performance.
2. Set the deviation limit k: k = 3 for Stage I, k = 2 for later stages.
3. For any juror score outside mean ± k, pull it back to that boundary.
4. Recompute the corrected mean (cmean) as the official score.

Why it matters: CMEAN dampens extreme outliers. Critics note that it can blur genuine minority preferences and can be gamed by jurors who understand the boundary effects.

Stage Momentum: Who Surged, Who Slipped

Biggest climbs by stage transition

• Tianyou Li: Stage I rank 37 → Stage II rank 4 (+33)
• William Yang: 31 → 3 (+28)
• Kai-Min Chang: 40 → 18 (+22)
• David Khrikuli: 25 → 5 (+20)
• Yang Gao: 26 → 8 (+18)
• Gabriele Strata: 34 → 16 (+18)
• Miki Yamagata: 38 → 20 (+18)
• Zitong Wang: 15 → 2 on Stage III entry (+13)
• Eric Lu: 7 → 1 by Stage II (+6)

Reading the momentum: Lu and Chen stabilized at the top by Stage III. Wang and Li posted late surges that set up strong prize finishes. Yang produced the biggest early jump into the top tier.

Agreement vs. Controversy: Where the Jury Split

Across stages, the standard deviation of juror scores flags consensus.

Most contentious examples

• Stage I: Xiaoyu Hu (σ = 3.71), Pedro López Salas (σ = 3.56), Piotr Pawlak and Vincent Ong near σ ≈ 3.5
• Stage II: David Khrikuli (σ = 3.22), Yang Gao (σ = 2.89)
• Stage III: David Khrikuli again (σ = 2.68), Piotr Pawlak (σ = 2.48)
• Final: Vincent Ong (σ = 2.58), Tianyao Lyu (σ = 2.42)

Most unanimous examples

• Stage I: Zihan Jin (σ = 1.46), Maiqi Wu (1.54)
• Stage II: Hyo Lee (1.29), Jacky Zhang (1.53)
• Stage III: Miyu Shindo (1.26), Xiaoxuan Li (1.27)
• Final: Miyu Shindo (1.50) among the lowest σ values

Takeaway: Khrikuli and Ong drew sharply split reactions, while performers like Jin and Shindo tended to produce steadier cross-jury alignment.

Piano Brands: Performance Snapshot

Average stage scores by instrument:

• Fazioli: rises over the rounds with 22.36 in Stage III and 21.33 in the Final
• Kawai: 20.68 in Stage III and 20.99 in the Final
• Steinway: consistent near 20.6 in Stage III and 20.44 in the Final
• Bechstein: limited sample in Stage I at 18.50
• Yamaha: Stage I 19.02, Stage II 18.72

Reading the trend: Among prizewinners, the Fazioli cohort includes the champion Eric Lu and the remarkable 16-year-old co-fourth prize Lyu. Kawai posts solid representation with Wang and Alexewicz. Steinway remains the volume leader with broad finalist presence.

Juror Profiles and Bias Signal

The bias metric compares a juror’s marks for compatriots to the average of other jurors for the same performances.

Positive net bias examples

• Piotr Paleczny (Poland) +1.66
• Momo Kodama (Japan) +1.13
• John Rink (USA) +1.00
• Robert McDonald (USA) +0.95

Negative net bias examples

• Dang Thai Son −1.21
• Ewa Pobłocka −1.36
• Krzysztof Jabłoński −2.14

Caution: Positive does not prove favoritism and negative does not prove counter-bias. It is a relative signal that should be viewed alongside variance and historical patterns.

Students of Jurors: Aggregate View

Jurors recuse for their own students. A few aggregates stand out:

• Dang Thai Son students: Stage II average 20.83, Stage III 23.05, Final 21.02
• Katarzyna Popowa-Zydroń students: Stage III 21.43, Final 21.25
• Robert McDonald students: solid progression into later stages with Final 20.48 for those who reached it

This view does not imply any scoring effect. It reflects studio pipelines and preparation quality seen through the independent marks of other jurors.

Alternate Leaderboard with JST

What JST does: normalizes each juror’s scale to a shared mean and spread while preserving the juror’s internal ranking. This gives equal voice to generous and severe scorers without clipping outliers.

Key differences vs. CMEAN

• Tiankun Ma advances to Stage II over Miki Yamagata at the cutoff.
• Vincent Ong would not advance to the Final.
• The margin between Eric Lu and Kevin Chen tightens to 0.07 in the final cumulative, with Lu still first.

JST prize set

1. Eric Lu: 22.28
2. Kevin Chen: 22.21
3. Zitong Wang: 21.28
4. Shiori Kuwahara: 21.24
5. Tianyao Lyu: 21.13
6. Piotr Alexewicz: 21.04

Remaining finalists by cumulative JST: William Yang, David Khrikuli, Tianyou Li, Miyu Shindo.

Interpretation: CMEAN and JST broadly agree on the podium and overall shape of the top ten, but JST slightly reshuffles the margins and one or two cutline outcomes.

Strategic Insights for Competitors

• Consistency beats volatility. Performers with low σ across jurors tended to progress smoothly. Aim for programs that reduce interpretive whiplash while still projecting personality.
• Late-round weight is real. Several medalists were not leading in Stage I but climbed with coherent narratives and strong concerto showings.
• Instrument fit matters. Fazioli artists featured at the top this year. Trial time is scarce. Arrive with a clear sound plan and a fallback.
• Know the math. Under CMEAN, extreme love or dislike gets pulled toward the pack. Under JST, scale differences are neutralized but rankings survive. Shape interpretations that convince across a spectrum of taste rather than gambling on polarizing outliers.

Notes for Organizers and Teachers

• Publish dual views. Providing both CMEAN and JST tables improves transparency and reduces suspicion around edge cases.
• Monitor variance hotspots. Repertoire or performance traits that repeatedly trigger very high σ can indicate ambiguous evaluation criteria.
• Pre-declare recusal and studio lists. The provided student aggregates are valuable for public understanding when paired with clear recusal protocols.

Data Access and Method

• Official scoring processed with CMEAN: k = 3 (Stage I) and k = 2 (Stages II to Final).
• JST normalization applied per juror set with preserved internal rank order.
• All tables and aggregates in this article are derived from the dataset described at the top.

If you want the Competopia community to explore the full JSON, we can host an interactive leaderboard with filters for stage, brand, jury variance, and a method toggle for CMEAN vs. JST.