About the Metrics

FantasyMeter helps you understand how each hero contributes to results across tournaments. Our metrics blend hero performance (scoring, participation, consistency) with player choices (deck building, entry selection, timing). The sections below tell you what a metric represents, why it matters, and how to read it. Click Formula (for the curious) if you’d like the math.

Initial Value (Vi)

Fragment Rate

Deck Score & Hero Score

Hero Participation

Deck Rewards → Hero Share

ETH-equivalent Reward

Per-Tournament ROIₜ

Overall ROI

σROI (Volatility)

Momentum-Level (from gliding_24h / gliding_7d)

Momentum (24h vs 7d)

Snapshot of current “heat” versus the weekly baseline.

• > 0 → trending up; < 0 → cooling versus the 7-day glide.
• Great for spotting fresh moves or post-spike cool-downs.

Momentum (24h vs 3d)

• > 0 → short-term outpacing the 3-day glide; < 0 → losing pace.

Momentum (24h vs 21d)

• Helps separate noise from real regime shifts.

Momentum (3d vs 21d)

• Confirms sustained improvement or deterioration.

Alignment score

Count of positive momentum reads (0–4).

• 3–4 → broad strength; 0–1 → broad weakness; 2 → mixed.

Velocity (rank)

• > 0 → rank improving (smaller rank numbers); < 0 → slipping.

Acceleration (rank)

• > 0 → improvement is speeding up; < 0 → fading.

Velocity (score)

• Pairs with rank velocity to confirm/deny the move.

Acceleration (score)

• Positive and rising → building power; negative → stalling.

Volatility (24h σ)

How choppy the last day has been.

Lower = steadier; higher = whipsaw.

Volatility (7d σ)

Baseline choppiness over the week.

Use as a context anchor for 24h moves.

Δσ (24h − 7d)

• < 0 → calming down (often constructive); > 0 → heating up.

CV (7d)

Scale-free stability: volatility relative to the average.

Lower CV = more dependable prints.

Z (24h vs 21d)

Where the latest 24h score sits versus the 21-day average.

• ~0 = typical; > +1 strong; < −1 weak.

% from 21d high

• Closer to 0% → pressing highs; larger negatives → pulled back.

% from 21d low

• Closer to 0% → retesting lows; higher → above the floor.

RSI (14)

Overbought/oversold style momentum read on score changes.

• > ~55 constructive; < ~45 weak; extremes near 70/30 can mean exhaustion.

Percentile (21d)

How today’s print ranks within the last 21 days.

• 80% → top quintile; 20% → bottom quintile.

Star Efficiency (live)

“Points per current star” snapshot. If stars go up, this can dip even as score improves.

ΔSE (24h − 7d)

• > 0 → converting better than the weekly norm; < 0 → softer conversion.

SE z(21d)

• Positive → unusually efficient; negative → below usual conversion.

Divergence (score vs rank)

Score and rank disagree.

• Score↑ + Rank↓ → stealth strength; Score↓ + Rank↑ → hollow momentum.

Composite momentum

Single view blending momentum, rank velocity, context, and stability.

• > +0.20 often bullish; < −0.20 often bearish; in-between = mixed/indecisive.

Tournament-Level (from tournament_histories + usage)

These compare tournament N vs N−1 and incorporate deck participation. “LA” means league-aware: in Reverse leagues, lower hero score is better and all deltas that measure “better vs worse” flip sign accordingly.

ΔHero (norm, LA)

ΔHero (raw, LA)

TPD (Tournament Performance Delta, LA)

HP (Hero Participation)

DPI (Deck Push Index)

Note: A Reverse “great” score (often near 0) can still yield a negative DPI if this event’s hp_raw is below your own historical average. A per-league baseline may be added in the future.

RAL (Reward Attribution Lift, ETH-eq)

Consistency (σ, CV)

Practical reading patterns

• Reverse quick read: lower score is better; big positive TPD/ΔHero (LA) after a zero means improvement.
• DPI vs Reverse scores: zero score can still show DPI < 0 if your hp_raw this event is below your own historical average; this is expected.
• Which ΔHero to trust? Use normalized for rarity-neutral performance; use raw for reward relevance.
• Baseline scope: DPI baseline currently spans all leagues; per-league baseline may be introduced later.
• Star Efficiency (live): uses current star count for both 24h/7d; per-tournament star history is not available.

Extended Momentum & Stability Metrics

Z(7d vs 3w)

Z-score comparing the 7-day average to the 3-tournament (≈21-day) mean.
> +1 = strong short-term uptrend; < −1 = short-term weakness vs baseline.

Warmth (3d)

Alias of Z(3d vs 3w). Measures how “hot” the last 3 days were compared to the 3-tournament baseline.
High positive → building momentum; negative → cooling down.

Z(max 3d vs 3w)

Z-score of the peak 24 h in the last 3 days versus the 3-week mean.
Highlights spikes: > +2 = viral moment or breakout; values near 0 = typical range.

Spike Breadth (3d)

Fraction of 3-day samples ≥ 80 % of the max.
0–0.2 = one-off burst; 0.4+ = sustained heat.

Z(24h vs 28d)

Where the latest 24 h score sits relative to the 4-tournament (~28 d) mean.
> +1 = breaking higher vs long baseline; < −1 = lagging long term.

σ (28d)

Standard deviation of last 4 tournament scores.
High σ → volatile performer; low σ → stable and predictable.

CV (28d)

Coefficient of variation over 4 tournaments (σ / mean).
< 0.10 = excellent steadiness; > 0.30 = erratic results.

σ Ratio (7d / 28d)

Compares short-term to long-term volatility.
> 1 = short-term turbulence; < 1 = calming trend.

Signal Quality (7d)

Reliability of recent momentum, blending z7_vs_3w and σ7d.
> 0.6 = clean, reliable trend; < 0.3 = noisy/uncertain.

Confirmations (short + mid)

Count of confirming conditions among {z7_vs_3w &gt; 0, ema7 &gt; ema14, mom21_28 &gt; 0}.
3 = strong multi-timeframe alignment; 0 = no agreement.

Composite Momentum Score

Weighted summary of short + mid momentum vs anchors, used for labeling.
> +0.20 = bullish; < −0.20 = bearish; near 0 = neutral.

EMA 7 / 14 & Cross-Age

Short and medium exponential averages of the 24 h series. Cross-Age = time since last bullish 7 > 14 crossover.
Long recent Cross-Age → trend persistent; very short → fresh reversal.

Max Drawdown (14d)

Largest peak-to-trough loss over 14 days.
< −10 % = healthy pullback; > −25 % = deep correction.

Ulcer Index (14d / 28d)

Root-mean-square drawdown — penalizes sustained stress more than brief drops.
< 5 % = smooth; > 15 % = rough ride.

Hurst Exponent (7–14d)

Trend persistence index.
> 0.55 = sustained trend; < 0.45 = mean-reverting chop.

Anchor Sharpe (8 w)

Reward-to-risk measure using 8-week tournament anchors.
Higher = more efficient long-term performance.

Recovery Rate (14d)

Speed of rebound from the last trough relative to its depth.
> 0.5 = strong comeback; < 0.2 = slow recovery.

Ulcer Performance Ratio

Anchor Sharpe / Ulcer Index — combines return and pain into one score.
> 2 = excellent efficiency; < 1 = inefficient or volatile.

Consensus Momentum

Robust composite blending all timeframe momentum readings.
> +0.5 = broad strength; < −0.5 = broad weakness.

Average Score (last K tournaments)

Rolling mean of last 4 / 8 / 12 tournament scores.
Use as steady anchors to judge regime shifts and normalize newer prints.

DDIᴺ, DDIᴿ & Rarity Premium

DDIᴺ — rarity-neutral

Uses normalized scores (no rarity multipliers).

DDIᴿ — rarity-weighted

Uses raw scores (with rarity multipliers) — aligns with reward splits.

Interpretation bands (rules of thumb):

• < 0.15 — Low dependency (filler role).
• 0.15–0.30 — Normal share.
• > 0.30 — High dependency (carry).

Rarity Premium

Hero Win (per tournament)

Rounding & Display Rules

• ETH values typically show 5 decimals.
• Fragments are rounded in the UI; calculations use exact fractional values.
• Percentages display to 2 decimals.
• When deck rewards are zero and Vi > 0, ROIₜ displays 0.00% (or — where intentionally muted).

Break-even (by league)

League breakdown table

Notes: Participation and rewards are Reverse-aware; fragment rate affects ETH-equivalent and thus ROI.

Zero Rewards & Missing Data

• If DeckScore = 0, we set hp = 0 for that tournament.
• If rewards are missing, we treat them as zero for ROI and label the row accordingly.
• If Vi = 0, ROI metrics show — to avoid division by zero.

Worked Examples

FAQ