orbit-wars-v12 / README.md
subham007's picture
Upload README.md
65f9b91 verified
|
Raw
History Blame Contribute Delete
1.82 kB
# πŸ›Έ Orbit Wars v12 β€” Full-Spectrum Domination Agent
**For the [Orbit Wars Kaggle Competition](https://www.kaggle.com/competitions/orbit-wars)**
## Test Results
| Matchup | Result |
|---|---|
| V12 vs Random | **3/3 wins** (eliminated by step ~90-99) |
| V12 vs Starter | **3/3 wins** (eliminated by step ~73-159) |
| V12 vs V10 (1v1, alternating sides) | **5/6 wins** (83%) |
| V12 vs 3Γ— V10 (4-player) | **3/3 wins** (eliminated all 3 every game) |
## How to Submit
1. Download `submission.py`
2. Create a new Kaggle notebook
3. Copy the content into the notebook
4. Submit
## Architecture
Built from v10/v11 base + every technique from 1200+ Elo public agents:
1. **110-turn Forward Timeline Simulation** β€” binary-search exact keep_needed
2. **Surplus-based dispatching** β€” send EVERYTHING above keep_needed
3. **7-mission priority system**: reinforce > crash_exploit > capture > snipe > swarm
4. **Multi-source coordinated swarm attacks** (2/3 sources, Β±2 turn sync)
5. **Snipe detection** β€” arrive at neutrals right after enemy fleet battles
6. **Crash exploit** β€” arrive after inter-enemy fleet collisions (4P)
7. **Hostile reinforcement prediction** β€” estimate enemy can reinforce target
8. **Doomed planet evacuation** β€” save ships from unsaveable planets
9. **Exposed planet detection** β€” attack enemies who just launched fleets
10. **Dynamic game phases** with domination tracking
11. **Weakest enemy targeting** with elimination bonus (4P)
12. **Production snowball scoring**: score = prod Γ— remaining_turns / cost
13. **Proactive defense** with stacked multi-enemy threat detection
14. **Sun-aware fleet routing** with iterative intercept solver
15. **Indirect wealth valuation** (cluster bonus from nearby planets)
16. **Endgame consolidation** to highest-production planets