J League · Score Research

08-22 18:15 CST · J League · Separate HT/FT models

Hiroshima Sanfrecce

Home side / left market direction

Full-time first pick 2:1 Probability 15.9% · High-total score cluster High-total score cluster: 2:1 15.9%; 3:1 10.3%; 4:2 4.5%; 3:2 4.0%; 3:0 3.6%; 4:1 3.3%; 5:1 2.4%; 4:0 2.1% Half-time first pick 0:0 (51.3%)

Kawasaki Frontale

Away side / right market direction

Five-Factor Read

1. Market strength Hiroshima Sanfrecce side rates higher

The handicap leans toward Hiroshima Sanfrecce; the market prices the home side as stronger.

2. Form and pace 0:0

HT first pick is 0:0 (51.3%); HT draw probability is 57.4%, so the model still respects early caution.

3. AI Sidecar Read FT total 3.25

This match uses a DeepSeek AI sidecar factor. The AI signal only applies a small candidate-score reweighting and does not override the market/history main model by itself. AI note: 主队广岛主场强势，但高温与双方防守稳健或压低节奏；川崎防守不弱且有偷分能力，倾向主胜但总进球谨慎。

4. Venue factor non-neutral / heat

Venue: 广岛和平之翼. Non-neutral venue. Weather/temperature: 天晴, 33℃～34℃. The schedule marks this as non-neutral, so home/away identity remains a weak background factor. Higher temperature raises fatigue risk and late-game tempo drop risk. Together with the totals line, the venue layer can support a more open upper tail.

5. Sample and soft factors Usable sample

144 comparable matches; useful for probability ranking

Full-Time Score Distribution

2:1 15.9%

1:1 12.9%

3:1 10.3%

1:2 7.4%

2:0 5.0%

1:0 4.8%

2:2 4.8%

4:2 4.5%

FT market: side line 1.00 · total 3.25 · Bucket ah_ou:1.0/3.0

Score cluster: High-total score cluster: 2:1 15.9%; 3:1 10.3%; 4:2 4.5%; 3:2 4.0%; 3:0 3.6%; 4:1 3.3%; 5:1 2.4%; 4:0 2.1%

Half-Time Score Distribution

0:0 51.3%

1:0 23.4%

0:1 9.1%

1:1 5.8%

2:0 5.4%

2:1 1.4%

3:0 1.2%

0:2 0.9%

HT market: side line 0.50 · total 1.25

Hiroshima Sanfrecce win61.3%

Draw21.7%

Kawasaki Frontale win17.0%

Hiroshima Sanfrecce expected goals2.29

Kawasaki Frontale expected goals1.28

Under 3.554.3%

FT temperature0.95

Empirical weight55.0%

Sample size144

Europe 1X2home 62.9% / draw 20.4% / away 16.7%

Key Variables

First-goal path Hiroshima Sanfrecce benefits most from opening the scoring; if HT stays 0:0, the FT distribution compresses toward lower scores.

Deciding factor Draw probability is 21.7%, so the exact-score set should keep draw and one-goal-margin scripts alive.

Risk point Under 3.5 is 54.3%; if the live total moves up materially, rerun the model.

Pre-Match Update · Market Movement

Published 2026-08-25 00:37:59 CST (started/finished about 54h 23m ago) | Snapshot window: pending -> 2026-08-25 00:37:59

First snapshotFT handicap 待补 · total 待补

Latest marketFT handicap 1 · total 3.25

Movement: FT handicap data pending; FT total data pending; HT handicap data pending; HT total data pending.

Pre-match note: Market snapshots are thin; refresh Feijing data again before publishing the final pre-match version.

Final Read: 2:1

Predicted side: Hiroshima Sanfrecce win · Actual side: draw · side missed

Using the full-time handicap, totals, comparable market buckets and half-time market, the model's full-time first pick for Hiroshima Sanfrecce vs Kawasaki Frontale is 2:1 (15.9%). Hiroshima Sanfrecce side rates higher. This match uses a DeepSeek AI sidecar factor. The AI signal only applies a small candidate-score reweighting and does not override the market/history main model by itself. AI note: 主队广岛主场强势，但高温与双方防守稳健或压低节奏；川崎防守不弱且有偷分能力，倾向主胜但总进球谨慎。 HT first pick is 0:0 (51.3%); HT draw probability is 57.4%, so the model still respects early caution. Exact-score outputs are probability rankings for pre-match research and review.

Risk note: exact scores are low-hit-rate events. The model output is a probability ranking, not a certainty; late market moves, lineups, red cards and early goals can reshape the distribution.

Sources

Feijing · fixtures, results, Asian handicap, totals, half-time markets, Europe 1X2 · Local history database · historical market lines, full-time scores, half-time scores · DeepSeek · AI sidecar context, bounded candidate reweighting only