A pitcher finishes the season with a 3.50 ERA, 3.10 FIP, and 3.80 xFIP.
Is he good? Due for regression? Buy-low candidate?
The answer depends on which stat you trust — and research shows ERA, FIP, and xFIP each tell a different story.
Here's everything you need to know about these three stats, including what they measure, how to calculate them, and which one best predicts a pitcher's future performance.
The Quick Answer: Which Predicts Best?
The Research Verdict
For predicting NEXT season's ERA:
- Best predictor: xFIP and SIERA (correlation ~0.52)
- Second best: FIP (correlation ~0.46)
- Third best: Current ERA (correlation ~0.38)
For explaining CURRENT performance:
- Best: FIP (highest correlation with same-season ERA)
- Use: ERA for actual results, FIP for deserved results
What Is ERA?
ERA (Earned Run Average) measures earned runs allowed per nine innings pitched.
ERA Formula:
ERA = (Earned Runs × 9) ÷ Innings Pitched
What it measures: Actual run prevention — the bottom line result.
What it includes: Everything — strikeouts, walks, hits, home runs, defense, sequencing luck, park factors.
Strengths:
- Measures what actually matters (runs win games)
- Simple and widely understood
- Reflects real-world impact
Weaknesses:
- Heavily influenced by defense quality
- Affected by sequencing luck (timing of hits/walks)
- Not predictive of future performance
- Poor year-to-year correlation (r = 0.38)
What Is FIP?
FIP (Fielding Independent Pitching) estimates what a pitcher's ERA should have been based only on outcomes a pitcher controls: strikeouts, walks, hit-by-pitch, and home runs.
FIP Formula:
FIP = ((13 × HR) + (3 × (BB + HBP)) − (2 × K)) ÷ IP + Constant
Where:
- HR = Home Runs allowed
- BB = Walks
- HBP = Hit By Pitch
- K = Strikeouts
- IP = Innings Pitched
- Constant = ~3.10 (scales FIP to ERA)
What it measures: Defense-independent pitching skill based on the "three true outcomes."
Theory behind FIP: Research by Voros McCracken found pitchers have little control over whether balls in play become hits. Therefore, judge pitchers only on outcomes they fully control.
Strengths:
- Removes defense quality from evaluation
- More stable than ERA year-to-year
- Better predictor of future ERA than current ERA
- Easy to calculate with basic stats
Weaknesses:
- Ignores contact quality (groundball vs flyball tendencies)
- Treats all home runs equally (some pitchers are more HR-prone)
- Not perfect at predicting future performance
What Is xFIP?
xFIP (Expected Fielding Independent Pitching) is FIP with one key difference: it normalizes home run rate to league average.
xFIP Formula:
xFIP = ((13 × (FB × lgHR/FB%)) + (3 × (BB + HBP)) − (2 × K)) ÷ IP + Constant
Where:
- FB = Fly Balls allowed
- lgHR/FB% = League average home run per fly ball rate (~10-12%)
- Everything else same as FIP
Key difference from FIP: Instead of using actual home runs, xFIP uses expected home runs based on fly balls allowed and league-average HR/FB rate.
Theory behind xFIP: Home run rates fluctuate randomly year-to-year. Pitchers control how many fly balls they allow, but whether those fly balls become home runs is mostly luck.
Strengths:
- Most predictive of future ERA (correlation ~0.52)
- Removes HR/FB% luck from evaluation
- Identifies pitchers due for regression
- Best for fantasy baseball and projections
Weaknesses:
- Some pitchers DO consistently control HR/FB% better than others
- Penalizes extreme flyball pitchers unfairly
- Still doesn't account for contact quality beyond FB%
Side-by-Side Comparison
| Factor | ERA | FIP | xFIP |
|---|---|---|---|
| Measures | Actual runs allowed | Expected runs based on K/BB/HR | Expected runs with normalized HR rate |
| Includes Defense? | Yes (heavily) | No (defense-independent) | No (defense-independent) |
| Includes Luck? | Yes (BABIP, sequencing) | Some (HR/FB% luck) | No (normalizes HR/FB%) |
| Best Use | Actual performance | Explaining current season | Predicting next season |
| Year-to-Year Correlation | 0.38 (poorest) | 0.46 (moderate) | 0.52 (strongest) |
| Same-Season Correlation with ERA | 1.00 (it IS ERA) | ~0.80 (highest) | ~0.75 (good) |
The Research: Which Stat Predicts Best?
Multiple studies have tested which stat best predicts future ERA. Here's what they found:
Study 1: Pitcher List Analysis (2020)
Researchers analyzed pitchers from 2015-2019 with 100+ innings in back-to-back seasons.
Findings (predicting next season's ERA):
- SIERA: RMSE 0.880, R² = 0.204 (best predictor)
- xFIP: RMSE 0.892, R² = 0.192 (second best)
- FIP: RMSE 0.968, R² = 0.140 (third)
- ERA: RMSE 1.113, R² = 0.082 (worst)
Conclusion: xFIP explains 19.2% of variance in next season's ERA. ERA explains only 8.2%.
Study 2: FanGraphs Community Analysis (2018)
Analyzed 2014 stats predicting 2015 ERA.
Correlation coefficients:
- xFIP: 0.520 (strongest)
- SIERA: 0.517 (second)
- FIP: 0.462 (third)
- ERA: 0.380 (weakest)
Conclusion: xFIP is 37% more predictive than ERA itself.
What The Research Means
For current season evaluation: FIP best explains whether a pitcher "deserved" their ERA.
For next season prediction: xFIP and SIERA are most predictive, but still only explain ~20% of variance.
Reality check: Even the best predictor (xFIP) only explains 1 in 5 ERA outcomes. Pitching is volatile.
How to Use Each Stat
Use ERA For:
- Actual results: What happened in games
- Awards/All-Stars: ERA determines Cy Young and All-Star selections
- Win-loss impact: ERA correlates with team wins
- Historical comparisons: ERA is the standard across eras
Use FIP For:
- Identifying luck: Large ERA-FIP gap = good/bad luck
- Defense-adjusted evaluation: Compare pitchers on different defensive teams
- Current-season analysis: Did pitcher "deserve" their ERA?
- Explaining performance: Why ERA is high/low
Use xFIP For:
- Predicting next season: Best single-stat predictor
- Fantasy baseball: Identify buy-low/sell-high candidates
- Trade evaluations: Is pitcher likely to improve or decline?
- Regression candidates: Who's due for ERA change?
Real Examples: ERA vs FIP vs xFIP
Example 1: Due for Improvement (Low ERA, High xFIP)
Pitcher A: 4.50 ERA, 3.20 FIP, 3.40 xFIP
Analysis:
- Pitching much better than ERA suggests (FIP 1.30 runs lower)
- Suffering from bad defense or sequencing luck
- HR/FB% close to league average (FIP ≈ xFIP)
Verdict: Buy-low candidate. ERA should drop significantly.
Example 2: Due for Regression (Low ERA, High xFIP)
Pitcher B: 2.80 ERA, 3.50 FIP, 4.00 xFIP
Analysis:
- ERA much better than peripherals suggest
- Benefiting from elite defense and good sequencing
- Unsustainably low HR/FB% (xFIP >> FIP)
Verdict: Sell-high candidate. ERA likely to rise to 3.50-4.00.
Example 3: True Ace (All Stats Agree)
Pitcher C: 2.50 ERA, 2.60 FIP, 2.70 xFIP
Analysis:
- Elite across all metrics
- Minimal luck involved
- Sustainable dominance
Verdict: True ace. Performance is real and repeatable.
Example 4: Home Run Problem (FIP << xFIP)
Pitcher D: 4.20 ERA, 4.50 FIP, 3.60 xFIP
Analysis:
Verdict: HR rate should regress to normal. ERA will drop to ~3.60.
Benchmarks: What's Good?
FIP and xFIP use the same scale as ERA:
| Rating | ERA / FIP / xFIP Range | MLB Example (2024) |
|---|---|---|
| Elite | Under 3.00 | Cy Young candidates |
| Excellent | 3.00-3.50 | Top-tier starters |
| Above Average | 3.50-4.00 | #2-3 starters |
| Average | 4.00-4.50 | #4-5 starters |
| Below Average | 4.50-5.00 | Back-end rotation |
| Poor | Above 5.00 | Replacement level |
When ERA-FIP-xFIP Disagree: What It Means
Large ERA-FIP Gap
ERA >> FIP (ERA much higher):
- Bad defense behind pitcher
- Unlucky on balls in play (high BABIP)
- Poor sequencing (hits cluster with men on base)
- Expect ERA to drop
FIP >> ERA (FIP much higher):
- Elite defense behind pitcher
- Lucky on balls in play (low BABIP)
- Good sequencing (hits with bases empty)
- Expect ERA to rise
Large FIP-xFIP Gap
FIP >> xFIP (FIP much higher):
- Allowing home runs on 15%+ of fly balls (unlucky)
- Pitching in extreme HR park
- Temporary HR problem
- Expect HR rate to normalize, ERA to drop
xFIP >> FIP (xFIP much higher):
- Allowing home runs on <8% of fly balls (lucky)
- Pitching in extreme pitcher's park
- Unsustainably low HR/FB%
- Expect HR rate to increase, ERA to rise
Limitations of All Three Stats
ERA Limitations:
- Defense-dependent (great fielders lower ERA dramatically)
- Park-dependent (Coors Field inflates ERA)
- Sequencing luck (two pitchers, same hits/walks, different ERAs)
- Poor predictive power (correlation only 0.38)
FIP Limitations:
- Ignores contact quality (groundball vs flyball tendencies)
- Some pitchers consistently beat/underperform FIP
- Still affected by HR/FB% luck
- Doesn't account for pitch sequencing skill
xFIP Limitations:
- Some pitchers DO control HR/FB% (extreme GB or FB pitchers)
- Penalizes extreme flyball pitchers unfairly
- Requires fly ball data (not always available)
- Still only explains 20% of future ERA variance
The Bottom Line: Which Should You Use?
For evaluating what happened:
- ERA: Actual results and impact on wins
- FIP: Defense-independent performance
For predicting what will happen:
- xFIP: Best single-stat predictor of future ERA
- SIERA: Slightly better than xFIP but more complex
For fantasy baseball/trades:
- Use xFIP to identify regression candidates
- Target pitchers with ERA >> xFIP (buy-low)
- Avoid pitchers with xFIP >> ERA (sell-high)
For complete evaluation:
- Look at ALL THREE stats together
- Add K/BB ratio (best single predictor per research)
- Consider park factors, defense quality, and opponent quality
Quick Decision Framework
How to Interpret ERA vs FIP vs xFIP
All three align (within 0.30 runs):
- True skill level
- Performance is sustainable
- Trust the numbers
ERA < FIP < xFIP:
- Pitcher getting lucky (good defense + low HR/FB%)
- ERA likely to rise significantly
- Sell-high candidate
xFIP < FIP < ERA:
- Pitcher getting unlucky (bad defense + high HR/FB%)
- ERA likely to drop significantly
- Buy-low candidate
FIP << xFIP but ERA ≈ FIP:
- Unsustainably low HR/FB%
- ERA will rise as HRs regress to mean
- Caution: regression coming
The Research Summary
Key findings from multiple studies:
- xFIP is 37% more predictive than ERA for forecasting next season
- FIP correlates best with same-season ERA (explains current performance)
- K-BB% is actually the single best predictor (even better than xFIP)
- Even the best metric only explains 20% of variance (pitching is volatile)
- SIERA edges xFIP slightly but requires complex batted ball data
Practical takeaway: Use xFIP for predictions, FIP for explanations, ERA for actual results — and recognize all have limitations.
Calculate ERA and Track Performance
Monitor pitcher performance across multiple metrics for complete evaluation.