🚧 Cutoff Mechanism
Lunor uses a cutoff system to ensure only high-quality, competitive, and forward-ready submissions are surfaced on leaderboards and considered for rewards.
This system is fundamental to maintaining the integrity, signal quality, and real-world relevance of every Lunor Quest.
🎯 Purpose of the Cutoff
The cutoff mechanism serves as a quality filter to:
- Prevent low-effort or random submissions from appearing on leaderboards
- Ensure only balanced and robust strategies are evaluated further
- Save review and compute bandwidth for meaningful solutions
- Promote real-world deployable systems over leaderboard gaming
📊 How Cutoffs Work
Each Lunor Quest defines a cutoff score — the minimum score a submission must achieve to:
- Appear on the public leaderboard (based on backtesting)
- Qualify for the forward simulation test (private leaderboard)
- Be considered for final rankings and rewards
🔁 Cutoff Applies in All Evaluation Phases
| Phase | Cutoff Enforced | Result of Failing |
|---|---|---|
| Backtest Phase (Public Leaderboard) | ✅ Yes | Submission is hidden from the public leaderboard |
| Forward Test Phase (Private Leaderboard) | ✅ Yes | Submission is excluded from final evaluation and prize pool |
🔍 Submissions must meet the cutoff in both phases to be eligible for final consideration.
🧩 Round-Based Quests
In quests with multiple rounds, each round has its own cutoff target, which may increase in difficulty over time.
Participants must meet the cutoff for each round independently to:
- Appear on that round's leaderboard
- Qualify for forward testing in that round
- Remain eligible for final rewards
⚠️ Failure to Meet Cutoff
If a submission fails to meet the cutoff score, it will:
- ❌ Not appear on any leaderboard (public or private)
- ❌ Not be evaluated further
- ❌ Not be eligible for rewards or recognition
If no submission meets the cutoff in a given round, the round may conclude with no winners selected.
🧠 Why This Matters
The cutoff mechanism ensures Lunor leaderboards showcase:
- Strategies that are both performant and stable
- AI or trading logic that holds up in unseen environments
- Contributions that reflect real-world deployability
It’s not about gaming a score — it’s about building something that works, holds, and scales.