Hi CAMEL-AI / OASIS team!
Sharing a real-world application of OASIS-style crowd simulation for commodity futures trading.
The use case: Before my IBKR futures bot executes any trade on Micro Gold (MGC), Crude (MCL), or other CME micros, it runs a pre-signal crowd simulation using OASIS principles.
- 50 commodity trader agents with different personas (gold bugs, momentum traders, contrarians, macro analysts...)
- 5 rounds of social interaction reacting to overnight news
- Output:
dissent_rate - the key signal for whether the market has conviction
dissent_rate < 0.25 -> strong consensus -> proceed
dissent_rate > 0.40 -> market split -> raise threshold, sit tight
This is the signal I wasn't getting from traditional sentiment scoring. OASIS made it possible.
Repo: https://github.com/vinayr1973-sudo/trading-ril (MIT)
Your research on herd effects and group polarization in LLM agents is directly applicable to real market dynamics. Thank you for open sourcing OASIS.
Will close if showcase issues aren't appropriate here.
Hi CAMEL-AI / OASIS team!
Sharing a real-world application of OASIS-style crowd simulation for commodity futures trading.
The use case: Before my IBKR futures bot executes any trade on Micro Gold (MGC), Crude (MCL), or other CME micros, it runs a pre-signal crowd simulation using OASIS principles.
dissent_rate- the key signal for whether the market has convictiondissent_rate< 0.25 -> strong consensus -> proceeddissent_rate> 0.40 -> market split -> raise threshold, sit tightThis is the signal I wasn't getting from traditional sentiment scoring. OASIS made it possible.
Repo: https://github.com/vinayr1973-sudo/trading-ril (MIT)
Your research on herd effects and group polarization in LLM agents is directly applicable to real market dynamics. Thank you for open sourcing OASIS.
Will close if showcase issues aren't appropriate here.