Find a strategy template, clone a public playbook, or bring your own Python. Paper-trade against live prediction-market data, then activate live trading with your own venue keys, hard limits, logs, fills, and P&L.
Most users should not start from a blank editor. Start from a strategy, inspect the assumptions, run it in paper, then decide whether it deserves live capital.
Browse all strategies ->A market-making style strategy on long-tail Polymarket markets. Places small bids and asks around the midpoint to capture spread. Educational only: real market-making needs inventory management and adverse-selection protection that a v0 strategy can't do safely. Use as a starting point.
Buys YES on Kalshi NFL game-line markets where the spread has moved against the underdog by ≥3 points in the final 6 hours before kickoff. Late line moves often overshoot; this fades the overshoot. Requires KALSHI_API_KEY_ID + KALSHI_PRIVATE_KEY in your venue credentials.
When a Polymarket market sees a sudden volume spike with directional move, follow it for a small position. Treats high-conviction informed flow as signal. Filters: min volume_24h, min absolute price move in the last hour, max position size capped tight.
Start from a Polymarket or Kalshi template, clone a public strategy, paste Python, or generate a first draft with your AI tool.
Run against live market data first. Inspect simulated fills, logs, positions, and P&L before any real order is allowed.
Add venue credentials, set caps, and promote the same strategy only when you decide the evidence is good enough.
Track fills, performance snapshots, logs, and stop controls across all running deployments from one surface.
Banger should make the risk boundary more obvious than the trade. Every activation path is built around caps, credentials, allowed markets, and a stop button.
Maximum dollars any strategy can place on one order.
New orders halt when a deployment breaches your loss limit.
Restrict a strategy to chosen venues, categories, or tickers.
The same strategy code moves from simulation to activation.
Stop the deployment from the dashboard or CLI.
Every simulated or live fill tied to venue, market, side, size, and price.
Strategy and runtime logs stream back to the deployment timeline.
Performance snapshots, open positions, realized and unrealized P&L.
The pitch is not that every strategy wins. The pitch is that every strategy can be tested, constrained, inspected, and stopped without rebuilding the same brittle bot infrastructure.
Claude, Cursor, and ChatGPT are useful for drafting strategy code. They are not the product. Banger is where the strategy gets tested, constrained, activated, and monitored.
from decimal import Decimal from banger import Strategy, signalsfrom banger.risk import RiskEnvelopefrom banger.venues.polymarket import Polymarket class SimpleFade(Strategy): name = "simple_fade" venues = [Polymarket()] universe = signals.polymarket_markets( time_to_resolution_lt="7d", min_volume_usd=10_000, ) risk = RiskEnvelope( max_position_usd=Decimal("100"), max_open_positions=10, ) fade_threshold: float = 0.92 _placed: set[str] = set() async def on_market_tick(self, market, tick, ctx): if market.market_id in self._placed: return if tick.yes_price is None: return if float(tick.yes_price) >= self.fade_threshold: self._placed.add(market.market_id) await ctx.place(Polymarket(), market, side="no", size=Decimal("25"))reported as extracted by arbitrage bots on Polymarket between April 2024 and April 2025. Of the top 20 wallets, 14 were reported as bots. In short-duration crypto markets, bots accounted for 55-62% of volume.
The conclusion is not that everyone should trust a black-box agent. The conclusion is that strategy testing, activation, and monitoring need to be built like a product.
Free paper trading keeps exploration cheap. Paid tiers unlock live strategy slots, attribution, MCP access, and higher limits.
Browse strategies, clone to paper, up to 3 paper deployments, 90-day history.
Unlimited paper, 1 live strategy included, basic MCP access, full attribution.
Pro plus 5 live strategies, higher MCP cap, custom risk envelopes, priority execution.