Awards
$1,500, $1,000 and $500 in every track.
The first, second and third teams in each of the four tracks win cash awards: $12,000 in total.
NeurIPS 2026 Competition Track
Agenthon 2026
Verifiable AI for quantitative finance.
A four-track competition testing whether AI agents can produce finance outputs that survive automated, leakage-controlled, cheat-resistant evaluation.
Submission
Docker agent
Admissibility
g0-g3 gates
Ranking
Track metrics
We're delighted to welcome NVIDIA, Bloomberg and AllianceBernstein as sponsors of Agenthon 2026. Thank you for your support, for helping us organize the competition, and for collaborating with us on its science and task development. Meet our sponsors
Key dates
Key dates.
Registration runs from 17 August to 12 October 2026. Registration and Development close together at 23:59 Anywhere on Earth (AoE, UTC−12) on 12 October. The last Development runs start by 20:00 UTC on 12 October: an upload that has not started by then is not run. The evaluation fleet is in scheduled maintenance on 13 October from 08:00 to 12:00 UTC, when the Final + Verification phase opens.
-
17 Aug – 12 Oct
Registration
Sign up and form your team.
-
28 Aug – 12 Oct
Development
Public practice repositories, iteration, and the validation leaderboard.
-
30 Sep
Call for papers closed
Papers for the NeurIPS workshop were due at 23:59 AoE.
-
13 – 25 Oct
Final + Verification
One submission per entered track is evaluated on sealed held-out units; organizers rerun leading submissions and review reproducibility within the same phase.
-
Sat 12 Dec
Workshop at NeurIPS, Atlanta
The Agenthon workshop, with final presentations by the winners, in Atlanta, Georgia.
The competition closes with a workshop and final presentations from the winning teams at NeurIPS in Atlanta, Georgia, on Saturday 12 December. All dates are given in the official rules, which govern if anything here is out of step.
Teams
Enter as a team of one to three.
You can compete alone or with up to two teammates. You join one team and stay on it, and that one team can enter as many of the four tracks as it wants.
Registration is per team, and only registered teams can submit solutions or appear on the leaderboard. Your team uses one account on the competition platform, chosen by the team — the FAQ explains what that means for everyone else on the team.
Have a question about entering? Start with the FAQ.
Core question
Can AI agents produce finance answers that can be checked by machine?
Agenthon extends the Alphathon program into its first NeurIPS edition. The competition keeps the four-track structure and hard finance setting, while adding sealed held-out data, automated leakage controls, reproducible reruns, and public leaderboards.
Agents in all four tracks run in sandboxed Docker containers without general internet access. Supported Coding, Forecasting and Explainability submissions can call House Nemotron through a restricted connection. Simulation has no network access. Build required dependencies and permitted artifacts into your image before evaluation.
Four tracks
Same competition spine, four finance problems.
Quant-finance coding agents
Build a Docker agent that solves quantitative finance coding tasks under pytest and financial-invariant checks.
- Verb
- solve
- Metric
- pass@1
- Gate
- pytest + invariants
Reasoning-augmented time series
Forecast future panels using time-series data plus a time-stamped text corpus. The score evaluates the forecast distribution; information uplift is a research aim.
- Verb
- forecast
- Metric
- CRPS composite
- Gate
- as-of cutoff + calibration
Accelerated market simulation
Submit an ABIDES-compatible simulator that is faster while preserving matching-engine semantics and market stylized facts. Development results are practice feedback, not controlled Final timing.
- Verb
- simulate
- Metric
- events/sec
- Gate
- semantic regression
Evidence-grounded prediction
Predict labels, values, or rankings over tabular entities, supported by a frozen evidence corpus. Follow the track guide for prediction, evidence and reasoning evaluation.
- Verb
- analyze
- Metric
- Track 4 composite
- Gate
- faithfulness + embargo
Protocol
Submit, check, score.
Each unit is checked for admissibility and evaluated using its track's scoring rules. Participant failures remain in the evaluation denominator; organizer faults are handled separately. If two Final submissions finish a track with the same ranking score, the tie is broken in favour of the one uploaded earlier.
01
Submit
Upload the toolkit-generated ZIP identifying your agent image and proving team registration.
02
Check
g0-g3 verify integrity, schema, cutoff/resource rules, and domain semantics.
03
Score
Track-specific scoring produces Development feedback. Track 1 uses pass@1 without a confidence interval.
Public/private firewall
Transparent grading, sealed answers.
Each track has a public practice repo and a private sealed exam repo. Public repos include practice tasks, examples, local checks, available baselines and participant docs. Private repos hold evaluation tasks, reference answers, scoring material and audit records.
Public practice
Private exam
Public practice units
Private-test held-out units
Examples, local checks and available baselines
Oracle solutions and final scorer
Manifest and canary safety checks
Canary registry and audit logs
Leaderboard
Competition scores, one track at a time.
Development leaderboards on agenthon.net are public and visible to everyone. Each track uses its own scale. Higher is better for the displayed leaderboard scores; the raw Forecasting loss is converted for this display. Use CodaBench for uploads and processing status. Follow the logged-in website for submission opening status. From 5 October 2026, 00:00 AoE (12:00 UTC), the Track 1 board ranks only runs uploaded after 25 September 2026 at 05:00 UTC, for every team (Track 1 README, rules 8 and 9).
The leaderboard is updated periodically from the competition platform.
T1 Coding
T2 Forecasting
T3 Simulation
T4 Explainability
Search teams
T1 Coding
T1 Coding: teams 1–25 of 94
Rank
Team
Score
1
Kaile
0.7674
2
Saifuddin
0.7209
3
Pluvia
0.5349
4
Proxima
0.5000
5
Arsen Ibragimov
0.4535
6
Chirayu
0.3837
7
Yan Su
0.3605
8
RandomGuyHavingFun
0.3488
9
MindAlchemist
0.2907
10
RizYL
0.2791
11
abc123
0.2674
11
Nana Champions
0.2674
13
dbuddha
0.2558
14
DeepSucc
0.2442
15
Cornfield Chase
0.2326
15
Formaggio
0.2326
17
Autonomous Alpha
0.2209
17
fiftyfifty
0.2209
19
Liqourena
0.2093
19
Tengen Toppa
0.2093
19
Pluto
0.2093
22
Anonymous
0.1977
22
in3lab
0.1977
22
APTAPT
0.1977
22
Quiet Signal
0.1977
T1 Coding: teams 26–50 of 94
Rank
Team
Score
22
Husky River Trading
0.1977
27
eagle
0.1860
27
Submartingale
0.1860
27
CMCCGDYDDICT
0.1860
27
eoo0m
0.1860
31
Just-in-Time
0.1744
31
VerityLovityFalsityCruelty
0.1744
31
BoGo
0.1744
31
Paragon
0.1744
31
Make It Make Cents 🙏
0.1744
36
Undefined
0.1628
36
TappuKiMKC
0.1628
36
Proof of Alpha
0.1628
36
AbangFrabi
0.1628
36
verifiquant
0.1628
36
FedorAzarov
0.1628
36
Powerhouse
0.1628
36
Eyyyyyy
0.1628
44
Sean
0.1512
44
Four Sigma
0.1512
44
Money Miner
0.1512
44
1991
0.1512
44
Give me a job
0.1512
49
gheware
0.1395
49
Warkop PuteraPrakoso
0.1395
T1 Coding: teams 51–75 of 94
Rank
Team
Score
49
Masato
0.1395
52
iluss
0.1279
52
Duo_tech
0.1279
52
Jade
0.1279
55
MicroCat
0.1163
55
hnhparitosh
0.1163
55
Agent_007
0.1163
55
always win
0.1163
55
Nietzsche
0.1163
60
TEst
0.1047
61
UNIVERSEPLAYER
0.0930
61
Sir Thaddeus
0.0930
61
HTHY
0.0930
61
Banana Kingdom
0.0930
65
Agent Alpha
0.0814
66
The Unscheduled Penguins
0.0698
67
Apex
0.0581
68
GAOKUO
0.0465
69
attention is qkv
0.0349
70
volatility_labs_stony_brook
0.0233
71
Urithiru
0.0116
—
Aamir Abdul Azeez
— not scored
—
Jin & Pei
— not scored
—
AgentReady
— not scored
—
3226
— not scored
T1 Coding: teams 76–94 of 94
Rank
Team
Score
—
AlbertQuant
— not scored
—
Verifiable Capital
— not scored
—
a-team
— not scored
—
FAWDA
— not scored
—
9:00pm
— not scored
—
C_cake
— not scored
—
huhudawang
— not scored
—
Quant Science Flow
— not scored
—
phtree
— not scored
—
DaoyuShu4094
— not scored
—
Future Ability
— not scored
—
Echoing
— not scored
—
Luca
— not scored
—
Artifical General Invesment
— not scored
—
Flowfront
— not scored
—
nogugu
— not scored
—
Outsiders
— not scored
—
Wealth-Building Lab
— not scored
—
IMAnonymous
— not scored
No team matches that search.
T2 Forecasting
T2 Forecasting: teams 1–25 of 116
Rank
Team
Score
1
Yan Su
-1.3403
2
DuML
-1.4388
3
Ywin
-1.5483
4
Flowfront
-1.5548
5
huhudawang
-1.5682
6
alpha-extractor
-1.5725
7
Made in Heaven
-1.6351
8
Samoyed
-1.6775
9
Chirayu
-1.7656
10
Not Today
-1.7697
11
Quiet Signal
-1.7917
12
HBIT
-1.8358
13
F1_1
-1.8367
14
Jin & Pei
-1.8424
15
Proof of Alpha
-1.8515
16
labubu666
-1.8693
17
CMCCGDYDDICT
-1.8758
18
Pawsibble
-1.9221
19
Apex
-1.9618
20
HackStreet Boys
-1.9722
21
Nexus of f(x)
-2.0134
22
Warkop PuteraPrakoso
-2.0303
23
mikelou1
-2.0579
24
Just-in-Time
-2.0898
25
VerityLovityFalsityCruelty
-2.0903
T2 Forecasting: teams 26–50 of 116
Rank
Team
Score
26
DKYnumber1
-2.1028
27
Autonomous Alpha
-2.1097
28
Jigglyduck
-2.1111
29
Paragon
-2.1128
30
Super Debugging
-2.1651
31
Saifuddin
-2.1769
32
spidy
-2.1999
33
Sapien
-2.2129
34
urayaha
-2.2155
35
lifeisluck
-2.2172
36
TradeFlare
-2.2310
37
LastDigitsOfPi
-2.2338
38
Masato
-2.2448
39
HG Market System
-2.2556
40
Cornfield Chase
-2.2726
41
quack
-2.2893
42
Sitadel Insecurities
-2.2962
43
JJ LAB
-2.3146
44
Formaggio
-2.3310
45
iluss
-2.3557
46
Money Miner
-2.3569
47
abc123
-2.3662
48
5 o'clock
-2.3690
49
time
-2.3694
50
fiftyfifty
-2.3772
T2 Forecasting: teams 51–75 of 116
Rank
Team
Score
51
pranshu rastogi
-2.3774
52
1991
-2.3901
53
AsOf Quant
-2.3990
54
Tengen Toppa
-2.4107
55
Bayes Street
-2.4108
56
Auror
-2.4301
57
QIE DOG
-2.4329
58
Javier Amo
-2.4429
59
Dog parents
-2.4467
60
lingsio
-2.4480
61
hnhparitosh
-2.4484
62
C_cake
-2.4513
63
in3lab
-2.4570
64
randomwalk
-2.4666
65
LiLaiLai
-2.4670
66
DoublePai
-2.4681
67
Aamir Abdul Azeez
-2.4722
68
ZZKK
-2.4795
69
In Sync
-2.4848
70
Natpaphon
-2.4891
71
Verifiable Capital
-2.4918
72
GAOKUO
-2.4932
73
SMTM
-2.4938
74
krnb
-2.4961
75
LIZARD
-2.5086
T2 Forecasting: teams 76–100 of 116
Rank
Team
Score
76
1111
-2.5090
77
PSS_
-2.5096
78
Garros Solo Lab
-2.5139
79
GXR123
-2.5169
80
Nietzsche
-2.5272
81
Yongchang
-2.5273
82
Try hard
-2.5347
83
AbangFrabi
-2.5381
84
Banana Kingdom
-2.5388
85
Jade
-2.5458
86
JiaxinLi
-2.5482
87
pluuuuuus
-2.5521
88
Husky River Trading
-2.5567
89
Probably Right
-2.5643
90
Liqourena
-2.5769
91
Launchlane Forecast
-2.5799
92
iKun-XinglinForrest
-2.5994
93
TheAxiom
-2.5999
94
YIC Quant
-2.6067
95
Luca
-2.6137
96
lwz888
-2.6141
97
Undefined
-2.6151
98
ARKANE
-2.6357
99
OOBs
-2.6373
—
Reference forecaster M0 reference, not ranked
-2.6412
100
SoloTraveler
-2.6765
T2 Forecasting: teams 101–116 of 116
Rank
Team
Score
101
ah_a1g
-2.6829
101
Give me a job
-2.6829
103
OffTheTape
-2.7369
104
BigWorld
-2.7476
105
katharsis
-2.7795
106
K-uant
-2.7985
107
PA-Agent
-2.8354
—
9:00pm
— not scored
—
ZOLDIK
— not scored
—
Urithiru
— not scored
—
Quant Science Flow
— not scored
—
phtree
— not scored
—
Hatenabase
— not scored
—
Edge-X
— not scored
—
Artifical General Invesment
— not scored
—
Outsiders
— not scored
No team matches that search.
T3 Simulation
T3 Simulation: teams 1–25 of 41
Rank
Team
Score
1
fuzz
231024.2407
2
abc123
221088.2954
3
IMAnonymous
215963.1445
4
Proof of Alpha
211712.3699
5
Cornfield Chase
210201.3605
6
Yan Su
210016.3629
7
Warkop PuteraPrakoso
201094.4945
8
VerityLovityFalsityCruelty
198767.4670
9
DKYnumber1
196323.2757
10
iMak AI Lab
190224.1037
11
S2WISH
189138.3673
12
Autonomous Alpha
185936.6020
13
Lumia
185889.4276
14
Paragon
184046.5937
15
Made in Heaven
181871.6467
16
Quantumonster
179955.5150
17
Money Miner
179846.1354
18
Verifiable Capital
178103.6278
19
PA-Agent
177343.4124
20
Jin & Pei
176009.6963
21
Yongchang
161575.8191
22
Probably Right
157361.2474
23
HackStreet Boys
156661.8792
24
fiftyfifty
152138.2119
25
huhudawang
134182.0441
T3 Simulation: teams 26–41 of 41
Rank
Team
Score
26
chilli
110245.5255
27
Saifuddin
84836.4432
28
PumpkinChicken
20625.2674
29
1991
17167.8369
30
Apex
16467.6082
31
CMCCGDYDDICT
15559.8991
32
Banana Kingdom
14180.6623
33
Just-in-Time
13813.8474
34
Quant Science Flow
13797.7076
35
Auror
13708.7408
36
Outsiders
12116.2977
—
TEst
— not scored
—
thalachira
— not scored
—
Hertz
— not scored
—
Artifical General Invesment
— not scored
—
Sequoia
— not scored
No team matches that search.
T4 Explainability
Rank
Team
Score
1
Genshin, launch!
0.8401
2
Just-in-Time
0.5787
3
Saifuddin
0.5234
4
abc123
0.5058
5
MALIU
0.5025
6
Paragon
0.4920
7
stone stone
0.4902
8
Sitadel Insecurities
0.4797
9
Quiet Signal
0.4683
10
huhudawang
0.4512
11
OmniSync
0.4365
12
ccczy
0.4029
13
FAWDA
0.4001
14
SignalCraft
0.3993
15
Apex
0.3954
16
Flowfront
0.3400
17
LOOOONG
0.1475
—
yohoho
— not scored
—
Chaewon-Research
— not scored
—
Northline
— not scored
No team matches that search.
2026 phases
Development, then Final + Verification.
Development runs through 12 October 2026. The joint Final + Verification phase runs from 13 to 25 October 2026. Registration and Development close on 12 October at 23:59 Anywhere on Earth (AoE, UTC−12). Final + Verification closes on 25 October at 23:59 AoE.
Upcoming
Development
Build and test with the public starter packages. Check the logged-in website for submission opening status.
0%
complete
-
Upcoming Aug 28 - Oct 12 Development
Teams practice and iterate with the public starter packages. Submission opening is announced separately.
-
Upcoming Oct 13 - Oct 25 Final + Verification
One submission per entered track is evaluated on sealed private-test units. Organizers rerun leading submissions and review reproducibility in the same phase. There is no separate verification submission. Run outputs, logs, and score files stay hidden.
Awards
Cash awards in all four tracks.
The top three teams in each track win a cash award, decided by the final standings after verification. That is $3,000 per track and $12,000 in total.
-
1st place
$1,500
In each of the four tracks.
-
2nd place
$1,000
In each of the four tracks.
-
3rd place
$500
In each of the four tracks.
Amounts are in US dollars. A team's award is split equally among its registered members. Eligibility, verification, tax and payment requirements are set by the official rules, and a confirmed winner who accepts an award publishes the code needed to reproduce the method.
Scientific outputs
Agenthon is designed to explain failures, not just rank winners.
Cross-track failure map
Gate failures are labeled and aggregated to show where finance agents break down.
Information uplift
T2 investigates whether text and reasoning improve forecasts; this is not a separate scored component.
Speed-realism frontier
T3 measures throughput only after semantic fidelity and stylized facts survive checks.
Faithfulness under embargo
T4 requires evidence-backed predictions that do not cite future or unsupported facts.
Call for papers
The call for papers is closed.
Accepted papers will be presented as posters at the Agenthon workshop at NeurIPS in Atlanta on Saturday 12 December. The program will follow.
