market_params
{
"seeds": [
3,
44,
909
],
"drift": 0.0002,
"volatility": 0.007,
"num_bars": 150,
"initial_price": 500
}strategy_code
class Strategy:
def __init__(self):
pass
def decide(self, price_history, portfolio):
return 1.0
expected_metrics
{
"tolerance": {
"sharpe": 0.05,
"sortino": 0.05,
"max_drawdown": 0.02,
"cagr": 0.03,
"calmar": 0.3
},
"per_seed": [
{
"seed": 3,
"sharpe": -2.298606489703321,
"sortino": -2.9039487354072664,
"max_drawdown": -0.19061085335940964,
"cagr": -0.23556440306741255,
"calmar": -1.2358394021941654
},
{
"seed": 44,
"sharpe": 0.7754817600796895,
"sortino": 1.1409413255342131,
"max_drawdown": -0.07754004401760116,
"cagr": 0.07975482589694538,
"calmar": 1.0285630722474375
},
{
"seed": 909,
"sharpe": 2.5766708374812723,
"sortino": 4.238623178181203,
"max_drawdown": -0.04839396570104317,
"cagr": 0.27749577281118976,
"calmar": 5.7340986379466745
}
]
}task_description
A passive buy-and-hold baseline: allocate fully to the asset on the very first bar and never rebalance away from full exposure again, regardless of subsequent price action. This exists as a simple reference point against which more active strategies (moving-average crossovers, mean-reversion, momentum, etc.) can be compared -- with no trading frictions modeled, holding a constant 1.0 target weight also means the harness's own rebalance-to-target logic performs no further trades after the first bar, since the position's value already moves one-for-one with price at full exposure.