What I Built
Using 8,619 NPB (Japanese professional baseball) game scores from 2016–2025, I calculated Park Factors (PF) for all 12 stadiums and visualized the year-by-year trends. The key focus: how do stadium renovations change the numbers?
→ GitHub: https://github.com/yasumorishima/npb-prediction
Key Terms (for first-time readers)
| Term | Meaning |
|---|---|
| Park Factor (PF) | A measure of how much a stadium inflates or suppresses scoring. 1.0 = neutral; >1.0 = hitter-friendly; <1.0 = pitcher-friendly |
| PF_5yr | 5-year rolling average park factor, smoothed with renovation breakpoints |
| Renovation breakpoint | The year a stadium renovation significantly changes the scoring environment. Data before/after must be separated |
| Home runs scored/allowed | Stats measured at a team's home stadium |
What Is a Park Factor?
Park Factor measures how much a stadium affects run scoring compared to a neutral park.
Formula (Baseball Reference standard):
PF = ((Home RS + Home RA) / Home G)
/ ((Away RS + Away RA) / Away G)
PF > 1.00 : Hitter-friendly (more runs scored)
PF = 1.00 : Neutral
PF < 1.00 : Pitcher-friendly (fewer runs scored)
Why Not Just Use Home Runs Scored?
# ❌ This mixes in the team's offensive strength
pf_bad = home_score / average_score
# ✅ Comparing the SAME team's home vs away performance isolates the park effect
pf = (home_RS + home_RA) / home_G / ((away_RS + away_RA) / away_G)
By comparing how the same team performs at home versus away, we isolate the stadium's contribution.
Why 5-Year Averages?
Single-season PF is noisy (especially for HR). FanGraphs and Baseball Reference typically use 3–5 year aggregates. I calculate both single-year PF and 5-year rolling average (PF_5yr).
The Renovation Problem
Here's a subtle but critical issue: 5-year averages become meaningless if they span a major renovation.
Example: Vantelin Dome Nagoya (Chunichi Dragons)
The 2025 PF_5yr is 0.844 (very pitcher-friendly). But a major HR wing installation is planned for 2026 (reducing left/right-center distance by 6 meters). If we keep including pre-renovation data after 2026, we'd show an artificially low PF for years.
My solution: renovation breakpoints. After a major change, only post-renovation data is used for multi-year averages.
# calc_park_factors.py
RENOVATION_BREAKS: dict[str, list[int]] = {
"ソフトバンク": [2015], # HR Terrace installed (L/R center -6m)
"ロッテ": [2019], # HR Lagoon installed (L/R center -4m)
"日本ハム": [2023], # New stadium (ES CON Field)
"楽天": [2016, 2026], # 2016: natural grass / 2026: fence moved in
"中日": [2026], # HR Wing installation planned
}
def calc_multiyear_pf(games, team, year, window=5):
"""Use only post-renovation data for multi-year PF calculation"""
reno = get_renovation_break(team, year)
data_start = reno if reno else games["year"].min()
years = list(range(max(year - window + 1, data_start), year + 1))
# ... aggregate and calculate
Rakuten has two renovation breakpoints (2016 and 2026), so both are registered.
The Visualization
Bars: 2025 PF (blue=hitter-friendly, red=pitcher-friendly) / Line: 5-year average PF / Orange vertical line: renovation completed / Purple dotted line: upcoming renovation
The visualization explicitly shows renovation timing for every stadium that had changes:
- Orange solid line: renovation already completed (e.g., ZOZO Marine 2019 HR Lagoon)
- Purple dotted line: upcoming renovation (e.g., Vantelin 2026, Rakuten 2026)
- Annotation: when renovation was before the data range (e.g., PayPay Dome 2015)
# plot_park_factors.py (excerpt)
for reno_year in breaks:
if reno_year < DATA_START:
# Before data range → text annotation on first bar
ax.text(DATA_START, YLIM[1] - 0.03, f"※{reno_year} renovated", ...)
elif reno_year > DATA_END:
# Future renovation → purple dotted line at right edge
ax.axvline(x=DATA_END + 0.45, color="#9333EA", linestyle=":", ...)
else:
# Within data range → orange solid line before the renovation year
ax.axvline(x=reno_year - 0.5, color="#F97316", linestyle="-", ...)
Key Findings
1. PayPay Dome (SoftBank) — 2015 HR Terrace converged to neutral
After the 2015 HR Terrace installation (L/R center -6m), the 2016 PF spiked to 1.171. Over 10 years, it gradually converged to PF_5yr=1.007 (2025) — essentially neutral.
| Year | PF | PF_5yr |
|---|---|---|
| 2016 | 1.171 | 1.171 |
| 2018 | 0.969 | 1.009 |
| 2025 | 0.976 | 1.007 |
2. ZOZO Marine (Lotte) — HR Lagoon transformed a pitcher's park
| Year | PF | Notes |
|---|---|---|
| 2018 | 0.874 | Pre-renovation (pitcher-friendly) |
| 2019 | 0.923 | HR Lagoon installed |
| 2020 | 1.101 | Immediate jump |
| 2021 | 1.235 | Strongly hitter-friendly |
| 2025 | 1.010 | PF_5yr=1.097 |
The renovation's effect appeared immediately in the following season.
3. ES CON Field (Nippon Ham) — New stadium peaked in Year 2
| Year | PF | PF_5yr | Notes |
|---|---|---|---|
| 2022 | 0.949 | 0.967 | Sapporo Dome (last year) |
| 2023 | 0.969 | 0.969 | ES CON Year 1 |
| 2024 | 1.212 | 1.089 | Year 2 surge |
| 2025 | 1.271 | 1.147 | Year 3, continued |
Year 1 was actually pitcher-friendly (right-center field is wide), but PF jumped to 1.212 in Year 2. Players may have needed time to adjust to the dimensions and wall caroms of the new park.
4. Vantelin Dome (Chunichi) — Pitcher's paradise for 10 straight years
PF_5yr ranged from 0.773–0.955 across the entire 2016–2025 period. Even the best year was 10%+ below average scoring.
| Year | PF | PF_5yr |
|---|---|---|
| 2016 | 0.773 | 0.773 |
| 2019 | 0.880 | 0.825 |
| 2022 | 0.867 | 0.839 |
| 2025 | 0.955 | 0.844 |
2026 HR Wing installation may change this dramatically — similar to what happened at ZOZO Marine in 2019.
5. Rakuten Mobile Park — Two renovations, 2026 is the big one
2016 natural grass conversion showed some effect, but the 5-year average has drifted back to pitcher-friendly territory (PF_5yr=0.908 in 2025). The 2026 fence adjustment could shift this significantly.
2026 Outlook
Two stadiums are due for major changes:
| Stadium | Renovation | Current PF_5yr | Expected Direction |
|---|---|---|---|
| Vantelin (Chunichi) | HR Wing (+6m closer) | 0.844 | → Hitter-friendly |
| Rakuten Mobile Park | Fence moved in | 0.908 | → More neutral |
The key implication: Chunichi and Rakuten pitchers may see their ERA rise in 2026, not because of their own performance decline, but because the park changed.
2025 Full Stadium Ranking (by PF_5yr)
| Rank | Team | Stadium | 2025 PF | PF_5yr | Character |
|---|---|---|---|---|---|
| 1 | Nippon Ham | ES CON Field | 1.271 | 1.147 | Most hitter-friendly |
| 2 | Swallows | Jingu | 1.096 | 1.129 | Hitter-friendly |
| 3 | DeNA | Yokohama Stadium | 1.184 | 1.102 | Hitter-friendly |
| 4 | Lotte | ZOZO Marine | 1.010 | 1.097 | Hitter-friendly |
| 5 | SoftBank | PayPay Dome | 0.976 | 1.007 | Neutral |
| 6 | Carp | Mazda Stadium | 1.065 | 0.996 | Neutral |
| 7 | Giants | Tokyo Dome | 0.878 | 0.981 | Neutral |
| 8 | Lions | Belluna Dome | 0.923 | 0.962 | Slightly pitcher-friendly |
| 9 | Orix | Kyocera Dome | 0.931 | 0.943 | Slightly pitcher-friendly |
| 10 | Hanshin | Koshien | 0.830 | 0.942 | Slightly pitcher-friendly |
| 11 | Eagles | Rakuten Mobile Park | 0.931 | 0.908 | Pitcher-friendly |
| 12 | Dragons | Vantelin Dome | 0.955 | 0.844 | Most pitcher-friendly |
Single-Year PF Ranking by Year (2016–2025)
Ranked by actual single-year PF each year — cleaner than PF_5yr for year-to-year comparison since multi-year averages mix in pre-renovation data.
| Rank | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 |
|---|---|---|---|---|---|---|---|---|---|---|
| 1st | DeNA 1.205 | Swallows 1.422 | Swallows 1.319 | Giants 1.193 | Swallows 1.230 | Lotte 1.235 | Eagles 1.121 | Swallows 1.374 | Ham 1.212 | Ham 1.271 |
| 2nd | SB 1.171 | Eagles 1.148 | Lions 1.230 | Lions 1.115 | Lotte 1.164 | Giants 1.201 | Swallows 1.087 | Lotte 1.120 | Swallows 1.166 | DeNA 1.184 |
| 3rd | Swallows 1.127 | DeNA 1.063 | DeNA 1.107 | Swallows 1.085 | DeNA 1.138 | DeNA 1.117 | DeNA 1.073 | Orix 1.089 | DeNA 1.146 | Swallows 1.096 |
| 4th | Lions 1.077 | Orix 1.056 | Giants 1.044 | DeNA 1.062 | Eagles 1.079 | Lions 1.067 | SB 1.044 | Carp 1.056 | Lotte 1.095 | Carp 1.065 |
| 5th | Eagles 1.072 | Giants 1.018 | Eagles 1.013 | SB 1.038 | Lions 1.044 | Ham 1.042 | Giants 1.034 | SB 1.056 | Carp 1.079 | Lotte 1.010 |
| 6th | Carp 1.058 | Lions 1.011 | Ham 0.980 | Ham 1.028 | Giants 1.023 | Hanshin 1.015 | Lotte 1.016 | Hanshin 1.023 | Lions 1.003 | SB 0.976 |
| 7th | Ham 0.996 | Ham 0.932 | SB 0.969 | Orix 0.969 | Orix 0.963 | Swallows 0.978 | Hanshin 0.967 | DeNA 0.993 | SB 1.002 | Dragons 0.955 |
| 8th | Hanshin 0.962 | Lotte 0.921 | Orix 0.963 | Carp 0.958 | Carp 0.941 | SB 0.964 | Carp 0.953 | Ham 0.969 | Giants 0.932 | Orix 0.931 |
| 9th | Giants 0.919 | SB 0.919 | Carp 0.955 | Lotte 0.923 | SB 0.912 | Carp 0.879 | Ham 0.949 | Lions 0.872 | Orix 0.932 | Eagles 0.931 |
| 10th | Orix 0.896 | Carp 0.901 | Lotte 0.874 | Hanshin 0.888 | Dragons 0.862 | Eagles 0.868 | Lions 0.938 | Giants 0.871 | Hanshin 0.868 | Lions 0.923 |
| 11th | Lotte 0.797 | Dragons 0.866 | Hanshin 0.804 | Dragons 0.880 | Hanshin 0.860 | Orix 0.864 | Orix 0.916 | Eagles 0.841 | Eagles 0.814 | Giants 0.878 |
| 12th | Dragons 0.773 | Hanshin 0.863 | Dragons 0.789 | Eagles 0.839 | Ham 0.846 | Dragons 0.806 | Dragons 0.867 | Dragons 0.797 | Dragons 0.803 | Hanshin 0.830 |
Key takeaways:
- Dragons (Vantelin) ranked last in 8 of 10 years (only exceptions: 2020, 2021)
- Swallows (Jingu) ranked 1st in 4 years (2017, 2018, 2020, 2023), always near the top
- Lotte (ZOZO Marine) went from 11th in 2016 to 1st in 2021 — direct result of 2019 HR Lagoon renovation
- Nippon Ham went from 12th in 2020 to 1st in 2025 — ES CON Field effect after 2023 move
Code & Data: https://github.com/yasumorishima/npb-prediction
Data sources: baseball-data.com / npb.jp (raw data not redistributed)
Top comments (0)