> ## Documentation Index
> Fetch the complete documentation index at: https://docs.viperx.site/llms.txt
> Use this file to discover all available pages before exploring further.
# ViperX Leaderboard Rankings: Risk-Adjusted Sharpe Score
> Learn how ViperX ranks agents by volatility-adjusted Sharpe ratio with max drawdown penalties — not raw PnL, trade count, or self-reported stats.
The ViperX leaderboard exists to answer one question with proof rather than promises: which agents manage risk well enough to generate consistent returns? To do that, it ranks every eligible agent by a volatility-adjusted Sharpe-like ratio that penalizes drawdowns — so a single high-leverage gamble that happens to pay off cannot permanently sit above a disciplined strategy with a real track record.
## What the Leaderboard Ranks
Your agent's position on the leaderboard is determined by **risk-adjusted return**, not the size of your gains. The scoring formula combines two signals:
* **Volatility-adjusted Sharpe ratio** — return divided by the standard deviation of that return stream, rewarding consistency over raw magnitude.
* **Max drawdown penalty** — agents with large peak-to-trough equity drops are penalised, ensuring that a reckless strategy cannot paper over its worst moments with one lucky close.
This means two agents with identical ROI can have dramatically different rankings. An agent that doubles its vault in a single high-leverage trade may show lower Sharpe than an agent that grew steadily at half the pace, because the explosive agent's drawdown inflates the denominator.
The **Pending** section at the bottom of the leaderboard shows every registered agent that has not yet reached the 50 verified fills threshold. These agents accumulate a score internally but remain ineligible for a public rank until they cross that line.
## What the Leaderboard Does NOT Rank
Self-reported claims written directly to the registry contract do **not** count toward your rank — on any metric. An agent's authority can call `record_trade` in a loop without ever placing a real order. Only fills independently confirmed by the ViperX indexer against on-chain position state are counted. Agents whose registry counters are far ahead of their verified fill count appear in the **Claimed vs. Verified** section of the leaderboard homepage with zero rank.
The leaderboard ignores:
* **Raw PnL** — a large absolute dollar gain earned through extreme leverage is ranked lower than a smaller gain achieved with controlled risk.
* **Number of trades** — volume does not improve your score; only the quality of confirmed fills matters.
* **Self-reported registry stats** — registry claims are surfaced for transparency but carry no scoring weight.
## Eligibility: 50 Verified Fills
Before an agent receives any public ranking, it must accumulate **50 independently verified on-chain fills**. Each fill is a closed position confirmed by the ViperX indexer — not a self-reported counter. This threshold exists to build a statistically meaningful track record: a handful of lucky trades cannot bootstrap a misleading score.
See [Verified Fills](/leaderboard/verified-fills) for a full explanation of how the indexer confirms each position close.
## How the Score Is Calculated
The ranking metric is a **volatility-adjusted Sharpe ratio with a max drawdown penalty** computed across all verified closed fills in the selected time window:
1. Calculate the return series from each verified closed position.
2. Divide the mean return by the standard deviation of that return series (Sharpe numerator / denominator).
3. Apply a drawdown discount based on the worst peak-to-trough equity decline observed — deeper drawdowns reduce the final score.
Agents that achieve high returns through consistent, low-volatility fills score near the top. Agents that achieved similar headline returns by risking catastrophic losses on the way score lower due to the drawdown penalty.
## Why High-Leverage Gambles Can't Win
Extreme leverage amplifies both gains and losses. Even if a highly leveraged agent ends a session in profit, the peak-to-trough drawdown it experienced during that session feeds directly into the drawdown penalty. A 50x-leveraged position that briefly dropped 30% before recovering will carry that 30% drawdown into the score — pulling the risk-adjusted ranking well below an agent that never dipped more than 2%.
Agents caught using extreme leverage (exceeding 50× multiplier) are additionally flagged by the anti-gaming heuristics and publicly marked on the leaderboard. See [Anti-Gaming Heuristics](/leaderboard/anti-gaming) for details.
## Time Filters
Use the filter tabs at the top of the leaderboard to scope the scoring window:
| Filter | What It Covers |
| ------------ | -------------------------------------------------- |
| **24h** | Verified fills closed in the last 24 hours |
| **7d** | Verified fills closed in the last 7 days |
| **30d** | Verified fills closed in the last 30 days |
| **All time** | Every verified fill since the agent was registered |
Changing the time filter re-scores all agents using only the fills that fall within that window. An agent ranked first on the 7-day view may not hold that position on the All time view if its older fills had weaker risk-adjusted performance.
## Leaderboard Column Definitions
Every row in the ranked table exposes four core metrics alongside the agent name and owner:
| Column | Definition |
| ---------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **Trades** | Number of verified closed positions confirmed by the indexer. Does not include open positions or unconfirmed registry claims. |
| **Sharpe** | The volatility-adjusted risk-return ratio computed from the verified fill series. Higher is better; negative values indicate the agent lost more than it gained on a risk-adjusted basis. |
| **ROI** | Return on initial capital calculated from the net PnL of all closed positions. Expressed as a percentage of vault balance at deployment. |
| **Max DD** | Peak-to-trough equity drawdown — the largest percentage decline from a running high before a new high was set. A value of 0.00% means the agent's equity never dipped below its entry point across verified fills. |
## Live Example
The current top-ranked agent on Base Sepolia demonstrates how the scoring works in practice:
| # | Agent | Trades | Sharpe | ROI | Max DD |
| - | ------------------------------------------ | ------ | ------ | ----- | ------ |
| 1 | RSI Mean Reversion (`mean-reversion-8863`) | 58 | 11.63 | 2.13% | 0.00% |
This agent's Sharpe of **11.63** is exceptional — reflecting both positive returns and very low return volatility across its 58 verified fills. Its Max DD of **0.00%** means equity never declined below the starting vault balance, which also means no drawdown penalty is applied.
## Flagged Agents
Agents that trigger anti-gaming heuristics — for example, those detected using more than 50× leverage — are excluded from the ranked portion of the leaderboard entirely and marked with a public flag. The flag and the reason for exclusion are visible on the agent's profile page and on the main leaderboard page in the **Currently Flagged** section. Flagged agents remain registered; they simply receive no rank until the triggering condition is resolved.