ANAC Competition

The Automated Negotiating Agents Competition (ANAC) is an international competition that has been running since 2010 to bring together researchers from the negotiation community. The competition challenges participants to design autonomous negotiation agents that can effectively negotiate with other agents in various scenarios.

NegMAS provides comprehensive support for ANAC competitions through the ANAC_INFO (also available as GENIUS_INFO) dictionary, which contains metadata about all ANAC competition years from 2010 to present. The ANAC community is committed to advancing he field through open-sourcing ALL strategies.

Note

The competition had one League and ran on the Java-based Genius platform. Starting in 2020, the main league began using the GeniusWeb platform, and from 2024, it was named the Automated Negotiation League (ANL) and runs natively on NegMAS.

Accessing Competition Data

You can access information about any competition year using the ANAC_INFO or ANAC_INFO dictionaries:

from negmas.genius.ginfo import ANAC_INFO, ANAC_INFO, get_anac_agents

# ANAC_INFO is an alias for ANAC_INFO
assert ANAC_INFO is ANAC_INFO

# Get information about a specific year
info_2024 = ANAC_INFO[2024]
print(info_2024["winners"])  # List of winners
print(info_2024["finalists"])  # List of finalists
print(info_2024["participants"])  # All participants

# Use helper function to get agents with specific criteria
winners_2024 = get_anac_agents(year=2024, winners_only=True)
bilateral_agents = get_anac_agents(bilateral=True)

Agent Availability

Agents from different competition years are available through different libraries:

Java/Genius Agents (2010-2019)

These agents were written in Java for the Genius platform. To use them, you need:

  1. GeniusBridge: A Java bridge that must be running to communicate with Java agents. See the Integrating with Genius tutorial for setup instructions.

  2. negmas-genius-agents (optional): A library that re-implements many of these agents in pure Python using AI-based conversion. This allows running them without the Java bridge. Install with: pip install negmas-genius-agents

GeniusWeb Java Agents (2020-2021)

These agents were written in Java for the GeniusWeb platform. To use them:

  • Install negmas-geniusweb-bridge: pip install negmas-geniusweb-bridge

  • The bridge provides AI-translated Python implementations of the Java agents.

  • All agents are available as GW-prefixed wrapped classes (e.g., GWAlphaBIU)

GeniusWeb Python Agents (2022-2023)

These agents were written in Python for the GeniusWeb platform. To use them:

  • Install negmas-geniusweb-bridge: pip install negmas-geniusweb-bridge

  • All agents are available as GW-prefixed wrapped classes (e.g., GWExploitAgent)

ANL Agents (2024+)

These agents are written in pure Python for NegMAS. To use them:

Platform Compatibility

The following table shows which platforms were used for each competition year and how to access the agents in NegMAS:

Years

Official Platform

NegMAS Compatible Implementation

Notes

2010-2019

Genius (Java)

GeniusBridge + negmas.genius

Requires Java runtime and GeniusBridge running. See Integrating with Genius.

2010-2019

Genius (Java)

negmas-genius-agents

AI-translated Python implementations. Install: pip install negmas-genius-agents

2020-2021

GeniusWeb (Java)

negmas-geniusweb-bridge

AI-translated from Java. 6 agents from 2021 are available. Install: pip install negmas-geniusweb-bridge

2022-2023

GeniusWeb (Python)

negmas-geniusweb-bridge

Native Python agents wrapped for NegMAS. Install: pip install negmas-geniusweb-bridge

2024+

NegMAS (Python)

anl-agents

Native NegMAS agents. Install: pip install anl-agents

Competition Years

Below is a detailed summary of each competition year, including the settings, winners, and agent availability.

2010

Settings: Bilateral, Linear utilities, Discounting

Winners:

  1. AgentK (agents.anac.y2010.AgentK.Agent_K)

  2. Yushu (agents.anac.y2010.Yushu.Yushu)

  3. Nozomi (agents.anac.y2010.Nozomi.Nozomi)

  4. IAMhaggler (agents.anac.y2010.Southampton.IAMhaggler)

Participants (7 agents): AgentFSEGA, AgentK, AgentSmith, Nozomi, IAMcrazyHaggler, IAMhaggler, Yushu

Agent Access: Java/Genius (requires GeniusBridge)

2011

Settings: Bilateral, Linear utilities, Discounting

Winners:

  1. HardHeaded (agents.anac.y2011.HardHeaded.KLH)

  2. Gahboninho (agents.anac.y2011.Gahboninho.Gahboninho)

  3. IAMhaggler2011 (agents.anac.y2011.IAMhaggler2011.IAMhaggler2011)

Finalists (8 agents): HardHeaded, Gahboninho, IAMhaggler2011, BramAgent, AgentK2, TheNegotiator, Nice Tit-for-Tat, ValueModelAgent

Agent Access: Java/Genius (requires GeniusBridge)

2012

Settings: Bilateral, Linear utilities, Reservation value, Discounting

Winners:

  1. CUHKAgent (agents.anac.y2012.CUHKAgent.CUHKAgent)

  2. AgentLG (agents.anac.y2012.AgentLG.AgentLG)

  3. OMACagent (agents.anac.y2012.OMACagent.OMACagent)

Finalists (8 agents): CUHKAgent, AgentLG, OMACagent, AgentMR, TheNegotiatorReloaded, BRAMAgent2, MetaAgent, IAMhaggler2012

Agent Access: Java/Genius (requires GeniusBridge)

2013

Settings: Bilateral, Linear utilities, Learning, Reservation value, Discounting

Winners:

  1. TheFawkes (agents.anac.y2013.TheFawkes.TheFawkes)

  2. MetaAgent2013 (agents.anac.y2013.MetaAgent.MetaAgent2013)

  3. TMFAgent (agents.anac.y2013.TMFAgent.TMFAgent)

Finalists (7 agents): TheFawkes, MetaAgent2013, TMFAgent, AgentKF, InoxAgent, Slavaagent, GAgent

Agent Access: Java/Genius (requires GeniusBridge)

2014

Settings: Bilateral, Non-linear utilities, Reservation value

Winners:

  1. AgentM (agents.anac.y2014.AgentM.AgentM)

  2. DoNA (agents.anac.y2014.DoNA.DoNA)

  3. Gangster (agents.anac.y2014.Gangster.Gangster)

Finalists (9 agents): AgentM, DoNA, Gangster, AgentYK, BraveCat, E2Agent, Atlas3, AgentTRP, WhaleAgent

Participants (18 agents): Full list in ANAC_INFO[2014]["participants"]

Agent Access: Java/Genius (requires GeniusBridge)

2015

Settings: Multilateral, Linear utilities, Reservation value

Winners:

  1. Atlas3 (agents.anac.y2015.Atlas3.Atlas3)

  2. ParsAgent (agents.anac.y2015.ParsAgent.ParsAgent)

  3. RandomDance (agents.anac.y2015.RandomDance.RandomDance)

Finalists (8 agents): Atlas3, ParsAgent, RandomDance, AgentX, Kawaii, Mercury, PhoenixParty, PokerFace

Participants (24 agents): Full list in ANAC_INFO[2015]["participants"]

Agent Access: Java/Genius (requires GeniusBridge)

2016

Settings: Multilateral, Linear utilities, Reservation value

Winners:

  1. Caduceus (agents.anac.y2016.caduceus.Caduceus)

  2. YXAgent (agents.anac.y2016.yxagent.YXAgent)

Finalists (10 agents): Caduceus, YXAgent, ParsCat, Farma, Atlas32016, MyAgent, Ngent, GrandmaAgent, AgentHP2, Terra

Participants (16 agents): Full list in ANAC_INFO[2016]["participants"]

Agent Access: Java/Genius (requires GeniusBridge)

2017

Settings: Multilateral, Linear utilities, Learning, Reservation value

Winners:

  1. PonPokoAgent (agents.anac.y2017.ponpokoagent.PonPokoAgent)

  2. CaduceusDC16 (agents.anac.y2017.caduceus.CaduceusDC16)

  3. Rubick (agents.anac.y2017.rubick.Rubick)

Finalists (10 agents): PonPokoAgent, CaduceusDC16, Rubick, AgentF, AgentKN, GeneKing, Mamenchis, ParsCat2, TucAgent, SimpleAgent2017

Participants (18 agents): Full list in ANAC_INFO[2017]["participants"]

Agent Access: Java/Genius (requires GeniusBridge)

2018

Settings: Multilateral, Linear utilities, Learning, Reservation value, Discounting

Winners:

  1. MengWan (agents.anac.y2018.meng_wan.Agent36)

  2. BetaOne (agents.anac.y2018.beta_one.Group2)

  3. AgentHerb (agents.anac.y2018.agentherb.AgentHerb)

Finalists (13 agents): MengWan, BetaOne, AgentHerb, ConDAgent, ExpRubick, FullAgent, IQSun2018, Libra, PonPokoRampage, Seto, Shiboy, SMAC_Agent, Yeela

Participants (20 agents): Full list in ANAC_INFO[2018]["participants"]

Agent Access: Java/Genius (requires GeniusBridge)

2019

Settings: Bilateral, Linear utilities, Reservation value, Uncertainty

Platform: Genius (Java)

Winners:

  1. AgentGG (agents.anac.y2019.agentgg.AgentGG)

  2. KakeSoba (agents.anac.y2019.kakesoba.KakeSoba)

  3. SAGA (agents.anac.y2019.saga.SAGA)

Finalists (6 agents): AgentGG, KakeSoba, SAGA, DandikAgent, GaravelAgent, MINF

Participants (18 agents): Full list in ANAC_INFO[2019]["participants"]

Agent Access: Java/Genius (requires GeniusBridge)

2020

Settings: Linear utilities, Reservation value, Uncertainty, Elicitation

Platform: GeniusWeb (Java)

Finalists (5 agents, alphabetically - no explicit ranking published):

  • AgentKT (Brown University, USA)

  • AhBuNeAgent (Ozyegin University, Turkey)

  • Angel (University of Tulsa, USA)

  • HammingAgent (TUAT, Japan)

  • ShineAgent (Bar-Ilan University, Israel)

Participants (13 agents): AgentKT, AgentP1DAMO, AgentXX, AhBuNeAgent, Anaconda, Angel, AzarAgent, BlingBling, DUOAgent, ForArisa, HammingAgent, NiceAgent, ShineAgent

Agent Access: negmas-geniusweb-bridge

2021

Settings: Linear utilities, Learning, Reservation value, Uncertainty, Elicitation

Platform: GeniusWeb (Java)

Winners:

  1. AlphaBIU (Bar-Ilan University, Israel)

  2. MatrixAlienAgent (University of Tulsa, USA)

  3. TripleAgent (Utrecht University, Netherlands)

Agent Access: negmas-geniusweb-bridge

Participants (6 agents): AgentFO2021, AlphaBIU, GamblerAgent, MatrixAlienAgent, TheDiceHaggler2021, TripleAgent

Agent Access: negmas-geniusweb-bridge

Note

Six ANAC 2021 agents have been AI-translated from Java and are available in the bridge package. Additional agents may be added in future releases.

2022

Settings: Linear utilities, Learning, Reservation value, Elicitation

Platform: GeniusWeb (Python)

Winners (Individual Utility):

  1. DreamTeam109Agent (College of Management Academic Studies, Israel)

  2. ChargingBoul (University of Tulsa, USA)

Winners (Social Welfare):

  1. DreamTeam109Agent (College of Management Academic Studies, Israel)

  2. Agent007 (Bar-Ilan University, Israel)

Finalists (3 agents): Agent007, ChargingBoul, DreamTeam109Agent

Participants (19 agents): Agent007, Agent4410, AgentFish, AgentFO2, BIU_agent, ChargingBoul, CompromisingAgent, DreamTeam109Agent, GEAAgent, LearningAgent, LuckyAgent2022, MiCROAgent, Pinar_Agent, ProcrastinAgent, RGAgent, SmartAgent, SuperAgent, ThirdAgent, Tjaronchery10Agent

Agent Access: negmas-geniusweb-bridge

2023

Settings: Linear utilities, Learning, Reservation value, Elicitation

Platform: GeniusWeb (Python)

Winners (Individual Utility):

  1. ExploitAgent - Bram Renting (Leiden University, Delft University of Technology)

  2. MiCRO2023 - Dave de Jonge (IIIA-CSIC)

Winners (Social Welfare):

  1. AntHeartAgent - Kaiyou Lei et al. (Ant Group)

  2. SmartAgent - Jianing Zhao et al. (Southwest University)

Finalists (4 agents): ExploitAgent, MiCRO2023, AntHeartAgent, SmartAgent

Participants (15 agents): AgentFO3, AmbitiousAgent, AntAllianceAgent, AntHeartAgent, ColmanAnacondotAgent2, ExploitAgent, GotAgent, HybridAgent2023, KBTimeDiffAgent, MiCRO2023, MSCAgent, PopularAgent, SmartAgent, SpaghettiAgent, TripleEAgent

Agent Access: negmas-geniusweb-bridge

2024 (ANL)

Settings: Bilateral, Linear utilities, Reservation value, Known opponent utility function

Platform: NegMAS (Native Python)

Winners (Individual Advantage):

  1. Shochan - Takayama, TUAT, Japan (anl_agents.anl2024.takafam.Shochan)

  2. UOAgent - Hirotada Matsumoto, TUAT, Japan (anl_agents.anl2024.team_moto.UOAgent)

  3. AgentRenting2024 - Mick Elshout, Utrecht University, Netherlands (anl_agents.anl2024.team_renting.AgentRenting2024)

Winner (Nash Optimality):

  1. Shochan - Takayama, TUAT, Japan

Finalists (10 agents): Shochan, UOAgent, AgentRenting2024, AntiAgent, HardChaosNegotiator, KosAgent, Nayesian2, CARCAgent, BidBot, AgentNyan

Participants (19 agents): Full list in ANAC_INFO[2024]["participants"]

Agent Access: anl-agents (pip install anl-agents)

More Information: ANL 2024 Results

2025 (ANL)

Settings: Bilateral, Linear utilities, Reservation value, Multi-deal

Platform: NegMAS (Native Python)

Winners:

1. RUFL - Garrett Seo, Rutgers University, USA (anl_agents.anl2025.team_271.RUFL) - 250€ 1. SAC Agent - Hossein Savari, University of Tehran, Iran (anl_agents.anl2025.university_of_tehran.SacAgent) - 250€ (tied for 1st) 3. UFunAtAgent - Fukutoku Yuma, TUAT, Japan (anl_agents.anl2025.team_305.UfunATAgent) - 100€

Finalists (12 agents, by qualification score):

  1. SacAgent (University of Tehran)

  2. ProbaBot (CWI, Netherlands)

  3. RUFL (Rutgers University)

  4. KDY (TUAT)

  5. JeemNegotiator (Bar-Ilan University)

  6. Astrat3m (Chongqing Jiaotong University)

  7. A4E (TUAT)

  8. OzUAgent (Ozyegin University)

  9. SmartNegotiator (Bar-Ilan University)

  10. CARC2025 (HIT Shenzhen)

  11. UfunATAgent (TUAT)

  12. Wagent (Chongqing Jiaotong University)

Participants (17 agents): Full list in ANAC_INFO[2025]["participants"]

Agent Access: anl-agents (pip install anl-agents)

More Information: ANL 2025 Results

Competition Settings Reference

The following table summarizes the settings for each competition year:

Year

Type

Linear

Learning

Reservation

Discount

Uncertainty

Elicitation

Multi-deal

Known Opp. Ufun

Known Opp. RV

Platform

2010

Bilateral

Yes

No

No

Yes

No

No

No

No

No

Genius

2011

Bilateral

Yes

No

No

Yes

No

No

No

No

No

Genius

2012

Bilateral

Yes

No

Yes

Yes

No

No

No

No

No

Genius

2013

Bilateral

Yes

Yes

Yes

Yes

No

No

No

No

No

Genius

2014

Bilateral

No

No

Yes

No

No

No

No

No

No

Genius

2015

Multilateral

Yes

No

Yes

No

No

No

No

No

No

Genius

2016

Multilateral

Yes

No

Yes

No

No

No

No

No

No

Genius

2017

Multilateral

Yes

Yes

Yes

No

No

No

No

No

No

Genius

2018

Multilateral

Yes

Yes

Yes

Yes

No

No

No

No

No

Genius

2019

Bilateral

Yes

No

Yes

No

Yes

No

No

No

No

Genius

2020

Mixed

Yes

No

Yes

No

Yes

Yes

No

No

No

GeniusWeb (Java)

2021

Mixed

Yes

Yes

Yes

No

Yes

Yes

No

No

No

GeniusWeb (Java)

2022

Mixed

Yes

Yes

Yes

No

No

Yes

No

No

No

GeniusWeb (Python)

2023

Mixed

Yes

Yes

Yes

No

No

Yes

No

No

No

GeniusWeb (Python)

2024

Bilateral

Yes

No

Yes

No

No

No

No

Yes

No

NegMAS

2025

Bilateral

Yes

No

Yes

No

No

No

Yes

No

No

NegMAS

Legend:

  • Multi-deal: Whether agents negotiate multiple deals simultaneously (introduced in ANL 2025)

  • Known Opp. Ufun: Whether agents have access to their opponent’s utility function (ANL 2024 only)

  • Known Opp. RV: Whether agents know their opponent’s reserved value (not used in any year so far)

Acknowledgements

Compiling the complete list of agents for ANAC since 2010 would have been impossible without the support of many people including:

External Resources