How the rating is calculated

The exchange shows you ROI, win rate and drawdown for each master trader. None of those numbers answers the question that actually matters: what happens to your money if you copy this account? So we recompute the picture from the daily equity series Bybit exposes publicly, and turn it into one number from 0 to 120.

Everything below is implemented in one function, so the leaderboard and the "Rating Breakdown" block on a trader's page can never disagree.

The base: 100 points minus penalties

The account is scored out of 100, then a return bonus is added on top (see below).

Component	Penalty
Liquidation days (a day when the account fell 90% or more)	−18 / −25 / −30 for 1 / 2 / 3+ days
Max drawdown of the compounded equity curve	0 at 15% → −20 at 100%
Worst single day	0 at −15% → −12 at −90%
Daily ROI volatility	0 at 5% → −15 at 40%
Daily win rate	0 at 55% → −10 at 10%
Sharpe (our own, daily series)	0 at 1.0 → −9 at −0.5
Exchange Stability Index	0 at 5 → −10 at 2
Exchange win rate (per trade, median of windows)	0 at 85% → −20 at 100%
90-day average holding time	0 at 12 h → −20 at 480 h
All-time cumulative ROI	0 at −10% → −12 at −70%
Current drawdown (below its own peak right now)	−3 at 30%, another −2 at 50%
High Leverage / Low Leverage tag	−5 / +3

All scales are linear: a 22% drawdown costs 2.8 points, not a step to 6. Steps are used only where the metric is discrete (liquidation days, the ROI bonus) or categorical (leverage tags).

Bonuses

Component	Bonus
Tenure (trading days)	+1 per full 90 days, only if all-time ROI is not negative
Follower PnL (all time)	up to +10 for what followers earned, down to −15 for large follower losses
Low Leverage tag	+3

The return bonus: up to +20

+1 point for every full 20% of 180-day ROI, capped at +20 (400% ROI and above). Grids and martingale accounts (180-day exchange win rate ≥ 90%) are capped at +10: near-perfect trading with extreme ROI is a grid, and a grid should not buy the top of the leaderboard with a single bonus.

The final score is 0–120: min(120, base + bonus).

Why these components and not others

We validated the scales on 7,173 traders. The top 1% by rating has 0% liquidation days, a median drawdown of 12 points, a median Sharpe of +1.93 and a median 180-day ROI of +24%. The bottom 10% has 99% liquidation days.

Two components carry most of the predictive power for survival:

• Daily volatility — the strongest single predictor of a liquidation (AUC 0.772).
• Max drawdown — second (AUC 0.715).

Return itself has almost none (AUC 0.471). That is the whole point of the rating: a leaderboard sorted by money alone measures who was lucky, not who is safe to copy.

What we deliberately do not score

• Follower count and AUM. They describe popularity, not skill, and a large base makes execution worse for the people copying.
• The 7-day window. Too short to mean anything; every leaderboard that ranks by a week is a roulette wheel.
• Symbols traded. A grid on one token and a swing trader on forty are not comparable by any single number, so symbol mix is shown as information, not a score.
• Profit sharing. It reduces the follower's return, so it belongs in the review and in your own calculation - not in the trader's score.

Limits of the calculation

• We see the last 180 days of the daily series. If a trader's account has been running for two years, earlier history is not part of the rating.
• Where the daily series shows −99% but the exchange reports −20% for the same 90 days, we trust the exchange: that pattern is a data artefact, not a wipeout.
• A liquidation day is defined as a daily ROI of −90% or worse. This counts account wipeouts visible in the series, not leverage the trader used on individual trades.

Past performance does not determine future results, and no score makes copying a trader safe. The rating is a way to compare accounts on the same basis, nothing more.