Skip to the page
INTEGRATIONS/OPENAI AGENTS SDK

OpenAI Agents SDK

tradefloor.integrations.openai_agents wraps an OpenAI Agents SDK Agent. The adapter runs your agent inside a simulated market on your machine. At each decision tradefloor sends it a JSON observation, checks the decision it returns, places the orders in the simulation and scores the run. LLM adapters covers the observation payload and decisions and model calls for all the adapters.

Install

pip install "tradefloor[openai-agents]"

A first run

This runs with no API key. The adapter gets ScriptedModel, an offline model the SDK ships for testing.

ScriptedModel from agents.testing answers each call with a function of the input. payload_of reads back the observation the adapter sent.

import json

import tradefloor as tf
from agents import Agent
from agents.testing import ModelStep, ScriptedModel, assistant_message
from tradefloor.integrations.openai_agents import OpenAIAgentsAdapter, payload_of

def answer(call):
    # reads the same payload a real model is sent
    first = payload_of(call)["assets"][0]
    order = {"symbol": first["symbol"], "side": "BUY", "quantity": 200}
    return [assistant_message(json.dumps({"actions": [order],
                                          "rationale": "add"}))]

offline = ScriptedModel([ModelStep.respond(answer)] * 5)
pm = Agent(name="Portfolio Manager", instructions="You manage a small portfolio.")
adapter = OpenAIAgentsAdapter(pm, mode="live", model=offline)

market = tf.Universe.random(12, seed=4242)
card = tf.evaluate({"pm": adapter}, seed=4242, universe=market, days=5)["pm"]
print(card)
print(len(adapter.record), "decisions")
Scorecard('pm', pnl=1,621, return=+0.16%, trades=5, impact=+0.38bps, sharpe=n/a (short run), vol=0.7%, in_market=100%, exposure=0.03x)
5 decisions

OpenAIAgentsAdapter defaults to mode="replay". For a live run, pass mode="live" or use openai_agent(agent), which is this adapter with live as its default. Give the Agent a model your account can use, set OPENAI_API_KEY and leave out model=. The SDK has tracing on by default and sends traces to OpenAI, so the adapter turns tracing off for each run it starts unless you pass tracing=True.

Tested with

openai-agents 0.23.1 and tradefloor 0.11.0.