Most "Competitive" Maps

Out of pure interest, I've compiled a list of the maps with the least random outcomes. This means I applied Whole History Rating to each of these maps independently and scored the accuracy of it. The better the rating system can predict the outcomes of games on a specific map, the higher the score. Maps with purely random outcomes are most unpredictable and thus earn the smallest score.

Sadly this requires a lot of games on each map to achieve usable data and even using the entire zk dataset (filtered for fair 1v1-3v3) proved insufficient for all but the most popular maps. Anyway I found it interesting to see that maps one might consider "random" are actually very predictable, probably due to a strong correlation between skill and players' experience in map specific strategies.

1. Red Comet (4551 games played): 0.067
2. Comet Catcher Redux (1208 games played): 0.061
3. SpeedMetal (1239 games played): 0.053
4. DesertCliffsV1 (1383 games played): 0.049
5. TitanDuel (2931 games played): 0.048
6. Archers_Valley_v5 (1710 games played): 0.047
7. DarkSide Remake (2160 games played): 0.045
8. Icy Run v2 (2269 games played): 0.043
9. Into Battle_v3 (1097 games played): 0.042
10. Ravaged_v2 (2086 games played): 0.042
11. Avalanche-v2 (1591 games played): 0.042
12. AlienDesert (1373 games played): 0.042
13. Barren (1257 games played): 0.041
14. Geyser_Plains_TNM04-V3 (1243 games played): 0.039
15. Small_Divide-Remake-v04 (4128 games played): 0.038
16. Aquatic Divide Revised v02 (1055 games played): 0.036
17. Trojan_Hills_v02 (1037 games played): 0.034
18. wanderlust_V01 (1320 games played): 0.033
19. SuperSpeedMetal (1420 games played): 0.033
20. Badlands (1291 games played): 0.032
21. Fields_Of_Isis (1404 games played): 0.031
22. Cooper_Hill_TNM02-V1 (1338 games played): 0.030
23. FrozenPlanetV3 (1017 games played): 0.029
24. Altair_Crossing_v3 (1410 games played): 0.023
25. Intersection_v3 (1280 games played): 0.023
26. Battle for PlanetXVII-v01 (1205 games played): 0.014
27. Vittra_v1 (1096 games played): 0.013
28. IsisDelta_v02 (1005 games played): 0.010
29. Small Supreme Battlefield V2 (1714 games played): 0.006

Speed metal is predictable, cuz better players have more experience in zk, thus they know they need NASA computer to play speed metal, while the nubs dont know this and lagg away.

Alternatively, pros just spam fleas and wait for nubs to leave.

I was astonished to see SpeedMetal faring much better than SuperSpeedMetal. Other maps with different versions, like Intersection_v3 and Intersection_v2 (not shown due to only having <1000 battles) have nearly the same score. The supermex and superfus apparently make it a lot more random.

What is the number? How do I interpret this?

Geyser isn't number one? Must be a bad algorithm

Btw, DeinFreund, the link to your java parser objects is dead

You do realize that IsisDelta is the worst 1v1 map on this list? Even much worse than CooperHill.

Thats because IsisDelta might look good but it plays like utter shit. In fact I would say it was only played because of the fact that Skasi made it be a map for the sandbox mission.

Isn't this a list with the least random outcomes?

So, I have to guess wich side is the most favoured and try to set it as the map of my homeworld....