The Agent in the rail, backtests a keystroke away, deploy in the same window. The room already knows your brokers.
class Momentum(Strategy): universe = sp500() # point-in-time lookback, skip = 252, 21 def prices_to_signals(self, prices): r = prices.pct_change(self.lookback - self.skip) return rank(r, pct=True) > 0.9
Illustrative preview. No broker connection or actual run results.
Plain English in; drafted code, a backtest, and honest coverage out. It never leaves your machine.
The Agent on the homepage ↗Editor, backtests, tearsheets, blotter, notebooks. One room.
class PairsEnergy(Strategy): universe = constituents("XLE") z_entry, z_exit = 2.0, 0.5 def prices_to_signals(self, prices): z = zscore(spread(prices), 60) return (z < -self.z_entry) | (z > self.z_entry)
In [3]: artifacts = ["strategy.py", "report.html", "run.json"] artifacts strategy.py report.html run.json
Ship to paper with one action. Real capital stays behind a separate gate.
| Strategy | Status | Broker | Mode | Report | Review |
|---|---|---|---|---|---|
| momentum | paper | simulator | Demo | sample | required |
| meanrev-pairs | paper | simulator | Demo | sample | required |
| vol-carry | paper | simulator | Demo | sample | required |