I couldn't find any docs about what the Benchmark is when running a backtest, and whether there's a way to adjust it.
Please checkout the dataset doc for benchmark used (typo-warning: both US equities data-set use SPY). You can use the 'set_benchmark' API function to select a different stock (only a single one) as the benchmark. For e.g.
from zipline.api import set_benchmark, symbol
def initialize(context):
set_benchmark(symbol('AAPL'))