map grows in size, while the performance of bots that
have more high-level reasoning capabilities (
Strategy Tactics and SCV) increases.
Nondeterminism Track
Seven bots were used for the nondeterminism track:
the five preexisting bots (RandomBiased, POWorkerRush, POLightRush, NaïveMCTS, PuppetSearch) and
two of the competition entries, StrategyTactics and
SCV. Each round robin tournament consisted of 7 ;
6 = 42 games (since we discarded self-play matches),
and we performed five full round-robin tournaments
in each of the eight maps, for a total of 5 ; 8 ; 42 =
1,680 games.
Figure 6 shows the win ratios achieved by each of
the bots in this track, organized by type of map. The
bot that achieved the highest win ratio over all maps
was StrategyTactics, with a win ratio of 0.655. The
second-best bot in this scenario was again
POLightRush, with a win ratio of 0.643. Figure 6 also
Figure 5. Win Ratios of the Bots in the Standard Track Plotted as a Function of Map Size.
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
64x6432x3224x2416x168x8
RandomBiased
POWorkerRush
POLightRush
NaiveMCTS
PuppetSearch
StrategyTactics
SCV
BS3NaiveMCTS
Figure 6. Win Ratios of the Bots in the Nondeterminism Track, by Map Type.
1
0.8
0.6
0.4
0.2
0
RandomBiased
POWorkerRush
POLightRush
NaiveMCTS
PuppetSearch
StrategyTactics
SCV
All Maps Hidden Maps Open Maps