Chapter 11: Key Takeaways

Quick Reference Guide

Expected Points (EP) Fundamentals

Definition: Expected Points represents the average number of points a team will score from a given game state, based on historical outcomes.

Core Components: 1. Field position (yard line 1-99) 2. Down (1st through 4th) 3. Distance (yards to first down/goal)

EP Approximations (1st & 10):

Down Adjustments (Approximate): - 1st down: baseline - 2nd down: -0.4 EP - 3rd down: -0.9 EP - 4th down: -1.5 EP

Expected Points Added (EPA)

The Formula:

EPA = EP_after - EP_before

Interpretation: - Positive EPA: Play helped the offense - Negative EPA: Play hurt the offense - Zero EPA: Play was neutral

Typical EPA Values:

Team EPA Benchmarks (Per Play):

Success Rate

Standard Thresholds:

Success Rate Benchmarks:

By Down (Typical Ranges): - 1st Down: 45-50% - 2nd Down: 40-45% - 3rd Down: 38-42%

Explosive Play Rate

Definitions: - Explosive Rush: 10+ yards - Explosive Pass: 15+ yards - Big Play: 20+ yards (any)

Benchmarks:

EPA vs Success Rate Matrix

Four Quadrants:

Quadrant Characteristics:

1. Elite (High EPA + High Success): - Consistent and explosive - Championship-caliber offense - Can win in any style

2. Explosive (High EPA + Low Success): - Big-play dependent - High variance - Can win shootouts, struggle in grind games

3. Grinding (Low EPA + High Success): - Ball control offense - Limited ceiling - Wins with defense and field position

4. Struggling (Low EPA + Low Success): - Needs improvement everywhere - Inconsistent and inefficient - Likely losing record

Key Formulas

Expected Points (Simplified):

def get_ep(yard_line, down, distance):
    # Base EP by field position
    if yard_line <= 50:
        base = -1.5 + (yard_line / 50) * 3.5
    else:
        base = 2.0 + ((yard_line - 50) / 49) * 4.0

    # Down adjustment
    down_adj = {1: 0, 2: -0.4, 3: -0.9, 4: -1.5}

    return base + down_adj.get(down, 0)

EPA:

def calculate_epa(play):
    ep_before = get_ep(play['yl_before'], play['down_before'], play['dist_before'])

    if play['touchdown']:
        ep_after = 7.0
    elif play['turnover']:
        ep_after = -get_ep(100 - play['yl_after'], 1, 10)
    else:
        ep_after = get_ep(play['yl_after'], play['down_after'], play['dist_after'])

    return ep_after - ep_before

Success Rate:

def is_successful(down, distance, yards_gained):
    thresholds = {1: 0.40, 2: 0.50, 3: 1.00, 4: 1.00}
    required = distance * thresholds[down]
    return yards_gained >= required or yards_gained >= distance

Common Applications

Player Evaluation: - QB: EPA per dropback - RB: EPA per carry + Success Rate - WR: EPA when targeted - Defense: EPA allowed

Team Analysis: - Overall offensive efficiency - Passing vs rushing efficiency - Situational performance (down, field position) - Red zone efficiency

Strategic Analysis: - Fourth down decisions - Play-calling tendencies - Situational aggressiveness

Minimum Sample Sizes

Limitations to Remember

1. Credit Assignment: Multiple players contribute to each play
2. Context: Doesn't capture score differential, time remaining
3. Sample Size: Small samples are noisy
4. Opponent Adjustment: Raw EPA doesn't account for opponent quality
5. Garbage Time: Include/exclude considerations needed

Quick Evaluation Guide

Is this offense efficient? 1. Check EPA per play (target: +0.10+) 2. Check success rate (target: 44%+) 3. Compare passing vs rushing EPA 4. Look at red zone performance 5. Examine third-down conversion

Is this player valuable? 1. Calculate player's total EPA 2. Compare EPA per opportunity to position average 3. Assess consistency (success rate) 4. Consider context (teammate quality, usage) 5. Check against sample size requirements

Why is an offense struggling? 1. Compare current EPA to baseline 2. Identify which phases declined (pass/rush) 3. Break down by down and distance 4. Examine field position performance 5. Look for root causes (protection, play-calling, personnel)

Key Vocabulary

Final Reminders

1. EPA measures value, Success Rate measures consistency
2. Both metrics together give complete picture
3. Always consider sample size before drawing conclusions
4. Context matters - adjust for situation when possible
5. Use as foundation, layer in additional analysis