algoglide 0.2.0

Algoglide agent runtime SDK

Algoglide agent runtime SDK for Python

Build, backtest, and deploy prediction-market trading agents.

Docs  ·  algoglide.com  ·  MCP Plugin

---

Install

pip install algoglide

Quick start

from algoglide import BaseStrategy, TickContext

class MomentumAgent(BaseStrategy):
    markets = ["will-bitcoin-hit-100k-by-2026"]

    def on_market_update(self, ctx: TickContext) -> None:
        yes = ctx.market.outcome("Yes")
        if not yes:
            return

        # Buy YES when price is below 40 cents
        if yes.price < 0.40 and ctx.market.elapsed_pct > 0.3:
            ctx.buy("Yes", size_usd=10.0, limit_price=yes.price + 0.01)
            ctx.log(f"Buying YES at {yes.price:.3f}")

        # Take profit above 65 cents
        if yes.price > 0.65:
            ctx.sell("Yes", size_usd=10.0)
            ctx.log(f"Taking profit at {yes.price:.3f}")

Concepts

Strategies are event-driven. You define a class that extends BaseStrategy, list the markets you want to trade, and implement on_market_update. The runtime calls your method on every price tick.

No polling loops. Never use while True or time.sleep. The runtime handles scheduling.

Context object. Every hook receives a TickContext (or FillContext) with market state, position info, and order methods.

Market state

def on_market_update(self, ctx: TickContext) -> None:
    # Current market snapshot
    yes = ctx.market.outcome("Yes")   # OutcomeSnapshot or None
    no  = ctx.market.outcome("No")

    # Price, spread, volume
    price = yes.price        # 0.0 - 1.0
    spread = ctx.market.spread
    volume = ctx.market.volume_24h

    # How far through the market's lifetime (0.0 = start, 1.0 = end)
    elapsed = ctx.market.elapsed_pct

Orders

# Market buy
ctx.buy("Yes", size_usd=10.0)

# Limit buy
ctx.buy("Yes", size_usd=10.0, limit_price=0.55)

# Sell
ctx.sell("Yes", size_usd=10.0)

# Cancel
ctx.cancel(order_id)

Fill handling

from algoglide import BaseStrategy, FillContext

class MyAgent(BaseStrategy):
    markets = ["some-market-slug"]

    def on_market_update(self, ctx):
        pass

    def on_fill(self, ctx: FillContext) -> None:
        ctx.log(f"Filled: {ctx.outcome} @ {ctx.price:.3f}, size={ctx.size:.2f}")
        ctx.log(f"Position: {ctx.position.size} @ avg {ctx.position.avg_price:.3f}")

Event decorators

from algoglide import BaseStrategy, on_news, on_signal_change

class NewsAgent(BaseStrategy):
    markets = ["fed-rate-decision-june"]

    def on_market_update(self, ctx):
        pass

    @on_news("federal reserve rate decision")
    def handle_fed_news(self, item):
        self.log(f"News: {item.headline}")

    @on_signal_change("fed-rate-decision-june", threshold=0.1)
    def handle_signal_shift(self, before, after):
        self.log(f"Signal shifted: {before.confidence:.2f} -> {after.confidence:.2f}")

Deploy

# Validate locally
algoglide validate agent.py

# Deploy to algoglide cloud
algoglide deploy agent.py --name "my-momentum-bot"

Or deploy via the MCP plugin in Claude Code / Cursor / Cline.

Links

• Documentation
• MCP Plugin
• TypeScript SDK
• Rust SDK