How Arena Score Works
Arena Score is a single 0-100 rating that answers one question: how good is this trader, really? The current version — Arena Score v4 — rewards genuine profit while staying honest about risk and statistical confidence.
The Formula
Arena Score = 100 × Quality × Confidence
Quality blends five performance dimensions. Confidence scales the score down when there is not enough data to trust it. The result is mapped to a percentile — roughly the percentage of traders you beat.
Quality: Five Dimensions
Quality is split evenly between earnings (50%) and skill (50%):
- PnL — weight 0.30: absolute profit in USD, on a log-magnitude scale
- ROI — weight 0.20: return on investment
- Drawdown — weight 0.20: worst peak-to-trough loss (smaller is better)
- Sharpe — weight 0.20: risk-adjusted return
- Consistency — weight 0.10: win rate and profit factor
Earnings (PnL 0.30 + ROI 0.20 = 0.50) reward traders who actually make money. Skill (Drawdown 0.20 + Sharpe 0.20 + Consistency 0.10 = 0.50) rewards doing it with controlled risk. Only the dimensions a trader actually has data for are included, then re-normalized.
PnL Uses Log-Magnitude
Absolute profit spans a huge range, so PnL is scored logarithmically: roughly $1K maps to 0 and $10M maps to 1. A $200 account scores near zero no matter how high its percentage return — real capital is what counts.
Other Dimensions Are Percentile Ranks
ROI, drawdown, Sharpe, and consistency are scored as within-cohort percentile ranks, not fixed thresholds. Each trader is compared against everyone else in the same period, so the scale self-calibrates — a "good" Sharpe is defined by the field, not a hard-coded constant.
Confidence: Statistical Honesty
Confidence = between 0.35 and 1.0
Confidence scales with two things:
- Sample size: more closed trades means more trust —
trades / (trades + 50), so 50 trades gives about 0.5 - Data completeness: how many risk metrics (Sharpe, drawdown, win rate, profit factor) are actually available
A trader with only 3 trades, or with missing risk data, is structurally capped and cannot reach the top no matter how lucky. Confidence never drops below 0.35, so real traders are never zeroed out.
The Displayed Score
The composite (Quality × Confidence) is turned into a percentile: your score is approximately the percentage of traders you beat. It is blended 70/30 with the relative composite so the very best traders do not all pile up against 100.
Why It Resists Gaming
- ROI is compressed — a 300% return and a 10,000% return land close together, so you cannot farm the top with a tiny account
- PnL rewards real money — small accounts score near zero
- Confidence caps thin track records — a handful of trades cannot buy a top rank
This keeps the leaderboard honest: consistent, well-capitalized, risk-aware traders rise to the top.