Team Management · Season Leaderboard
Titan Index
A composite performance metric that balances raw speed and power-to-weight efficiency into a single score, designed to capture the fundamental tradeoff in erg-based athlete evaluation.
Why a Composite Metric?
Coaches face a recurring tradeoff when ranking athletes on the erg:
- Raw speed (split time per 500 m) favors heavier, more powerful athletes who move the flywheel fastest in absolute terms.
- Efficiency (watts per pound) favors lighter athletes who produce more power relative to their body weight — critical for on-water performance where the boat carries every kilogram.
Neither dimension alone tells the full story. A pure speed ranking under-values technical lightweights; a pure efficiency ranking under-values the raw horsepower needed for competitive boat speed. The Titan Index combines both into a single leaderboard score so coaches can quickly identify athletes who excel across both dimensions.
How It’s Calculated
The Titan Index uses z-score standardization — the same technique used in sports analytics metrics like baseball’s OPS+ and basketball’s PER — to put speed and efficiency on a common scale before combining them.
Step 1 — Standardize Speed
For each athlete, compute a speed z-score from the assignment’s average split:
speedZ = -(split - meanSplit) / stdSplit
Negated because lower split = faster = better. An athlete 1σ faster than average gets a speedZ of +1.
Step 2 — Standardize Efficiency
Compute an efficiency z-score from watts per pound (wplb):
effZ = (wplb - meanWplb) / stdWplb
Higher wplb = better, so no negation needed. An athlete 1σ more efficient than average gets an effZ of +1.
Step 3 — Combine & Normalize
Average the two z-scores with equal weight, then rescale to the output range:
raw = (speedZ + effZ) / 2 titanIndex = ((raw - min) / (max - min)) × 100
The best athlete in the visible group scores 100, the lowest scores near 0. Everyone else is distributed proportionally between them based on their combined z-score.
Per-Workout vs. Season Titan Index
The Titan Index is computed at two levels:
- Per-workout Titan Index
- Calculated for each scored assignment relative to the athletes who completed that specific workout. Stored as part of the workout result. This is a snapshot of where an athlete stood in a single session.
- Season Titan Index (rolling window)
- The value shown on the Season Leaderboard. Computed as a recency-weighted average of per-workout Titan Indexes over a rolling window of recent assignments. This smooths out single-day variance and reflects an athlete’s trend rather than a single performance.
The rolling window approach means an athlete who has a bad day won’t see their season score crater, and an athlete who peaked once won’t ride that score all season. More recent workouts carry more weight than older ones.
Interpreting the Score
| Range | Interpretation | Coaching Signal |
|---|---|---|
| 80–100 | Elite within the group | Strong on both dimensions. Top candidates for racing lineups. |
| 50–80 | Above average | Solid performers. Drill into Speed vs. Efficiency columns to see which dimension is driving the score. |
| 20–50 | Below average | May be developing in one dimension. Compare the individual Speed and Efficiency rank columns to identify focus areas. |
| 0–20 | Lowest in the group | Not necessarily “bad” — could be a novice in a group of varsity rowers. The index is always relative. |
The season-level rolling average naturally compresses scores toward the middle — an athlete who scores 0 in one workout but 40 in others will have a season Titan that is much more reasonable than the single-workout floor.
Key Properties
- Relative, not absolute
- The Titan Index is calculated within the currently visible athlete group. Changing filters (team, squad, tier) recalculates the index against only the visible athletes. A score of 75 means “75th percentile of this group” — not a fixed benchmark across all teams.
- Equal weighting (50/50)
- Speed and efficiency contribute equally. This is a deliberate design choice: in rowing, both absolute boat-moving power and power-to-weight ratio matter for on-water performance. Neither is inherently more important for general team assessment.
- Requires weight data
- The efficiency component (watts per pound) requires athlete body weight. Athletes missing weight data will not have a Titan Index calculated. The leaderboard shows “—” for these athletes.
- Minimum 2 athletes required
- Z-scores require variance to be meaningful. When fewer than 2 athletes have both speed and efficiency data, the system falls back to composite rank as an approximation.
Statistical Foundation
The approach — standardize heterogeneous metrics via z-scores, combine with a weighted average, then normalize to a readable scale — is a standard technique in sports analytics and composite index construction:
- Z-score normalization removes unit dependence (seconds vs. watts/lb) and naturally accounts for the spread of each metric within the cohort.
- Equal-weight averaging is the simplest defensible combination when neither input is demonstrably more predictive. This mirrors approaches used in metrics like the NBA’s Player Efficiency Rating and baseball’s OPS+.
- Min-max rescaling to 0–100 produces an intuitive “how good within this group” score without requiring external reference distributions.
Relationship to Other Leaderboard Columns
| Column | What It Measures | Time Window |
|---|---|---|
| Titan Index | Z-score composite of speed + efficiency, 0–100 scale | Rolling window (recent workouts) |
| Composite | Average of ordinal speed rank + ordinal efficiency rank | Season average |
| Speed | Average ordinal rank by split time (lower = faster) | Season average |
| Efficiency | Average ordinal rank by watts per pound (lower rank = more efficient) | Season average |
Use the Titan Index for a recency-weighted view of overall performance. Use Composite for a full-season ordinal view. Compare individual Speed and Efficiency columns to understand what’s driving an athlete’s position.
Scale Design: Why 0–100 (and Alternatives)
The current implementation uses min-max normalization to a 0–100 scale. This is the most intuitive for coaches but has a known drawback: someone is always 0 and someone is always 100 within a per-workout snapshot.
The season-level rolling average naturally softens this — averaging across multiple workouts means nobody stays pinned at 0 unless they are consistently the lowest on every dimension in every workout.
Alternative Scales Considered
| Approach | Formula | Range | Trade-off |
|---|---|---|---|
| Min-max 0–100 (current) | (raw−min)/(max−min) × 100 | 0–100 | Simple, intuitive. But bottom athlete is always 0. |
| Compressed floor (20–100) | 20 + (raw−min)/(max−min) × 80 | 20–100 | Nobody sees “0.” Still a relative scale but less demoralizing. |
| T-score | 50 + 10 × rawZ | ~20–80 | Classic psychometrics. Average is always 50, most athletes fall 30–70. Nobody near 0. |
| Percentile rank | ordinal position / N × 100 | 0–100 | Even distribution (every athlete has a unique percentile). Loses magnitude information — barely-different athletes can have very different percentiles. |
Limitations
- Not cross-team comparable. Because the index is normalized within the visible group, a Titan Index of 80 on one team is not equivalent to 80 on another. It measures relative standing, not absolute performance.
- Equal weighting may not fit every selection context. When selecting for a specific boat class (e.g., lightweight 4×), efficiency may matter more. When selecting for an open 8+, raw speed may dominate. Coaches should use the individual Speed and Efficiency columns alongside the Titan Index.
- Small cohorts amplify extremes. With only 3–4 athletes, top and bottom scores are more volatile. The metric becomes more informative with larger groups (8+ athletes).
- Requires weight data for efficiency component. Athletes without body weight recorded cannot receive a Titan Index.