Chapter 23: Case Study 2 - Finding Gems: Jokic, Siakam, and Identifying Late-Round Value

Introduction

The 2014 NBA Draft's second round produced two future All-NBA players: Nikola Jokic (pick #41) and Pascal Siakam (pick #27, though evaluated after Jokic's selection as a 2016 draftee). These selections demonstrate that draft models can identify value overlooked by traditional scouting. This case study examines what made these players identifiable as undervalued prospects and how to build models that find similar late-round gems.

Part 1: Nikola Jokic - Pick #41, 2014 Draft

Pre-Draft Profile

Background: - Serbian center, 19 years old - Playing in Serbian KLS (second-tier league) and Adriatic League - Limited exposure to NBA scouts - Considered a draft-and-stash prospect

Statistics (2013-14, Mega Vizura - Adriatic League): - 11.4 PPG, 6.4 RPG, 2.5 APG - 53.4% FG in limited minutes - Age: 19

Physical Profile: - Height: 6'10" (listed, likely 6'11" without shoes) - Weight: 232 lbs - Wingspan: ~7'1" - Standing reach: Excellent for position - Athletic testing: Did not attend combine

Why Jokic Was Overlooked

1. Athletic Profile Concerns - Not a traditional NBA athlete - Lateral quickness questions - Vertical explosiveness below typical NBA center

2. Competition Level - Adriatic League not considered elite - Limited games against top EuroLeague competition

3. Playing Style - Unorthodox offensive game - Heavy reliance on skill over athleticism - Didn't fit traditional center molds

4. Information Asymmetry - Limited video available - Few scouts with expertise in Serbian basketball - Not a household name internationally

What Models Should Have Identified

Signal 1: Age-Adjusted Production

Age: 19 in Adriatic League
Per-40 minute projection:
- Points: 19.2
- Rebounds: 10.8
- Assists: 4.2

Age adjustment (19-year-old producing at this level): × 1.25
Adjusted production score: 85th percentile for second-round prospects

Signal 2: Passing for a Big Man Jokic's 2.5 APG (4.2 per 40) was extraordinary for a center: - 98th percentile among centers in comparable leagues - Historical correlation: High-passing bigs project to higher NBA value

Signal 3: Free Throw Percentage - 75.7% FT in 2013-14 - Improved to 80%+ the following year - Indicated shooting touch that would translate

Signal 4: International Production vs. Age Among international big men under 20 years old with similar production, the success rate was significantly higher than draft position suggested.

Model Projection (Retrospective)

Using the methodology from Chapter 23:

Base production translation:
- EuroLeague-equivalent factor: 0.75
- Age bonus: 1.20
- Position factor: 0.90 (center)

Projected Year 3 NBA production:
- 14.5 PPG, 9.2 RPG, 4.8 APG
- Win Shares: 6.0

Career projection: 55-75 Win Shares
90% CI: [15, 110]

This projection would have made Jokic a top-20 value in the draft, not a second-round afterthought.

Actual Outcomes

Career (through 2023-24): - 3× MVP (2021, 2022, 2024) - 6× All-Star - 5× All-NBA - NBA Champion (2023) - Career Win Shares: 100+ and climbing - Projected final career WS: 200+

Jokic has been the most valuable second-round pick in NBA history.

Part 2: Pascal Siakam - Pick #27, 2016 Draft

Pre-Draft Profile

Background: - Cameroonian forward - Played at New Mexico State (WAC Conference) - Late basketball starter (began at age 15) - Raw but athletic

College Statistics (2015-16): - 20.3 PPG, 11.6 RPG, 1.8 APG - 53.8% FG, 17.6% 3PT, 68.6% FT - Age at draft: 22

Physical Profile: - Height: 6'8.5" (no shoes) - Wingspan: 7'3" - Weight: 230 lbs - Athletic testing: Elite agility, good vertical

Traditional Evaluation Concerns

1. Age (22) - Old for a prospect - Limited development runway expected

2. Conference (WAC) - Low major conference - Conference factor: ~0.75

3. Shooting (17.6% 3PT, 68.6% FT) - Poor shooting indicators - Traditional stretch-4 role unlikely

4. Late Start in Basketball - Only 7 years of organized basketball - Seen as lacking fundamentals

What Models Should Have Identified

Signal 1: Elite Length

Wingspan ratio: 7'3" / 6'8.5" = 1.067
Classification: Elite length (99th percentile for position)
Defensive projection: High upside

Signal 2: Production Adjusted for Context

Conference factor: 0.75
Production: 20.3 PPG, 11.6 RPG

Adjusted for conference:
- Points: 20.3 × 0.72 × 0.75 = 11.0 pts/100 (NBA equivalent)
- Rebounds: 11.6 × 0.85 × 0.75 = 7.4 reb/100

Despite low conference, this production is still notable

Signal 3: Athletic Profile - Lane agility: Top 10% for forwards - Vertical: Above average - Sprint: Elite - Combined with length = switchable defensive profile

Signal 4: Late Bloomer Trajectory Players who started basketball late but showed rapid improvement: - Hakeem Olajuwon (started at 15) - Pascal's year-over-year improvement was extreme - Suggests untapped potential

Quantifying Development Trajectory

Siakam's college progression: - Year 1: 6.4 PPG, 5.4 RPG - Year 2: 20.3 PPG, 11.6 RPG

Improvement rate: +218% in points, +115% in rebounds

Historical late bloomers with similar improvement rates had significantly higher success rates than their draft position suggested.

Model Projection (Retrospective)

Standard projection (age penalty):
- Base: 35 career WS
- Age penalty (22): × 0.85
- Conference penalty: × 0.90
- Standard projection: 27 WS

Development-adjusted projection:
- Improvement rate bonus: × 1.30
- Length bonus: × 1.10
- Adjusted projection: 39 WS
90% CI: [10, 75]

The development-adjusted model would have flagged Siakam as potential first-round value at pick #27.

Actual Outcomes

Career (through 2023-24): - NBA Champion (2019) - Most Improved Player (2019) - 3× All-Star - All-NBA Second Team (2020) - Career Win Shares: 50+ and climbing - Developed reliable three-point shot (37%+ in peak years)

Siakam exceeded typical second-round expectations by a massive margin.

Part 3: Common Patterns in Late-Round Gems

Identifying Factors

Based on Jokic, Siakam, and other successful late picks:

Building a Late-Round Value Model

Feature Priority: 1. Wingspan/length relative to position (weight: 0.20) 2. Age-adjusted production or improvement trajectory (0.20) 3. Unique skill indicators (0.15) 4. Athletic testing results (0.15) 5. Free throw percentage as shooting indicator (0.15) 6. Competition adjustment (0.15)

Key Insight: Late-round models should weight potential indicators MORE than production indicators, since production is already incorporated into consensus rankings.

Information Inefficiencies to Exploit

1. International Leagues - Less scouting coverage - Age-based production often overlooked - Different playing styles undervalued

2. Low-Major Conferences - Heavy discounting creates opportunities - Look for players with extreme physical tools

3. Late Bloomers - Traditional models penalize age - Improvement trajectory can indicate higher ceiling

4. Position Scarcity - Unique skill sets (e.g., passing big men) - Versatile defenders with length

Part 4: Quantitative Framework for Finding Value

Value Score Calculation

def calculate_late_round_value_score(prospect):
    """
    Calculate value score for late first/second round prospects.
    Higher scores indicate undervaluation potential.
    """

    base_score = 50  # Neutral

    # Physical tools bonus
    if prospect.wingspan_ratio > 1.06:
        base_score += 15
    elif prospect.wingspan_ratio > 1.04:
        base_score += 8

    # Age-adjusted production
    if prospect.age < 20 and prospect.production_percentile > 60:
        base_score += 20  # Young producer
    elif prospect.improvement_rate > 1.5:
        base_score += 15  # Rapid improver

    # Unique skills
    if prospect.assists_per_40 > position_avg * 1.5:
        base_score += 10  # Unusual passing
    if prospect.blocks_per_40 > position_avg * 1.5:
        base_score += 8   # Unusual rim protection

    # Athletic testing
    if prospect.athletic_composite > 75:
        base_score += 12

    # Shooting projection
    if prospect.ft_pct > 0.78:
        base_score += 10
    elif prospect.ft_pct < 0.68:
        base_score -= 8

    # Information inefficiency bonus
    if prospect.international or prospect.conference_rank > 20:
        base_score += 10  # Less scouted

    return base_score

Applying to Historical Drafts

Retroactively applying this framework to the 2014 draft:

The value score would have identified Jokic as significantly undervalued.

Part 5: Risk Factors and Limitations

Why Late-Round Gems Are Rare

Even with optimal modeling: - Most second-round picks fail (60%+ bust rate) - Physical tools don't guarantee skill development - Work ethic and character are hard to model - Injury risk affects all picks equally

False Positive Rate

For every Jokic, there are many players who: - Had similar statistical profiles - Were drafted late - Never developed into NBA players

The goal is improving hit rate from ~10% to ~20-25%, not guaranteeing success.

Model Limitations

1. Sample Size: Very few data points for late-round success
2. Changing Game: Style evolution affects which attributes matter
3. Development: Can't model work ethic or coaching quality
4. Injury: Career-altering injuries are unpredictable

Part 6: Practical Application

Scouting Protocol for Late-Round Picks

1. Screen for physical outliers - Extreme length - Elite athletic testing - Unique combinations

2. Identify production anomalies - Young age with production - Extreme improvement - Position-unusual skills

3. Investigate information gaps - Why is this player undervalued? - Is it correctable or fundamental?

4. Apply model scoring - Calculate value score - Flag prospects above threshold (e.g., 70+)

5. Deep dive on flagged prospects - Additional video study - Interview coaches/teammates - Medical evaluation

Draft Strategy Implications

For Teams with Multiple Second-Round Picks: - Use model-identified prospects as primary targets - Accept higher bust rate for upside - Draft-and-stash internationals when possible

For Teams Needing Immediate Contributors: - Weight floor over ceiling - Focus on athletes who can defend - Be cautious of "project" prospects

Exercises

Exercise 1

Using publicly available statistics, apply the late-round value score to a recent draft class. Identify the 3 second-round prospects with the highest value scores.

Exercise 2

Research a successful late-round pick not discussed in this case study. What factors made them identifiable as undervalued?

Exercise 3

Identify a prospect in the current/upcoming draft who fits the "late-round gem" profile based on the factors discussed. Justify your selection.

Exercise 4

Calculate the expected value of prioritizing model-identified late-round prospects versus random selection. Assume: - Base second-round success rate: 10% - Model-identified success rate: 25% - Expected value of success: 20 Win Shares - Expected value of failure: 0 Win Shares

Conclusion

The cases of Nikola Jokic and Pascal Siakam demonstrate that draft inefficiencies exist, particularly in later rounds. Key factors that identified these players as undervalued included:

1. Extreme physical tools (length, athleticism)
2. Age-adjusted or trajectory-based production
3. Unique skills unusual for their position
4. Information inefficiencies (international, low-major conferences)

While these methods cannot guarantee success, they can significantly improve the expected value of late-round selections. Teams that build robust models for identifying undervalued prospects gain meaningful competitive advantages in roster construction.

The lesson is not that models can find All-Stars consistently in the second round—Jokic remains a historic outlier. The lesson is that systematic evaluation can improve hit rates in a domain where consensus frequently leaves value on the table.