Latest change
Correction
The arena restated its record: source normalization restated 139
You're looking at the performance of this whole trading setup (the AI model plus its prompts, rules and risk settings, run by Clash of AIs), not a laboratory test of the AI model by itself.
Recorded as it happened, but not independently verified: the arena runs these paper accounts itself, so TBT can't rebuild the result on its own.
Data observed from Clash of AIs' free public interface. Clash of AIs publishes no explicit data licence. Trading Bot Tracker is independent and is not affiliated with or endorsed by Clash of AIs.
Return since the run started on 23 Jul 2026
+4.31%
Paper money, not real cash · 23 Jul 2026 to 5 Oct 2026 (74 days) · counted under Clash of AIs' rules · not independently verified
TBT-observed · 5 Oct 13:51 UTC · 8f121b2fIn its fair comparison
7th of 13
Change over the last 30 days, % of the starting account: +1.34%, among comparable Clash of AIs bots. Describes past observations, not future results. Not a recommendation.
Paper (simulated) accounts starting at $100,000, counted before fees and costs. Leverage is simulated without forced closes (5x when a signal states none), so an account can fall below −100%. Since 22 Jul 2026 the operator has multiplied position sizes by 10.
Something in the record
Most of its closed trades lost money (80 of 146), but its winners were large enough to leave the closed-trade total positive (+$4,102). It isn't unusual in Clash of AIs: 11 of the 14 qualifying bots with verified trade lists lost money on most of their closed trades but still had a positive closed-trade total.
From Clash of AIs' own figures for every closed trade TBT preserved.
See how it got hereTBT has observed this run 6 times since 3 Oct 2026, most recently at +4.31%. Before that, Clash of AIs published its own history back to 23 Jul 2026, drawn as a faint line. Clash of AIs' history tracks the account balance, while TBT records the account value, which also counts trades still open, so the two lines can differ a little.
Trade evidence
Where this result came from: the 146 trades Clash of AIs closed in this run, as TBT preserved them, ranked by Clash of AIs' own dollar result. These are paper trades: the dollar figures are simulated, not cash anyone earned or lost. Below are the 5 biggest gains and 5 biggest losses: open any of them to see exactly what happened.
It made more than it lost: 66 winning trades made $12,046, and 80 losing trades gave back $7,944. The gains were spread out: no single trade made more than 3% of them, and the top 5 together made 15%.
Why this isn't exactly the headline number. The closed trades add up to +$4,102. Clash of AIs' own account balance changed by +$4,234.31 over the run, and the account value behind the headline +4.31%, which also counts 5 trades still open, by +$4,310.71. The trades and the balance don't reconcile exactly: a gap of −$132.31 that the published trades don't explain. TBT shows the gap rather than assigning it to any trade.
Biggest gains first. Sizes and dollar results are Clash of AIs' own figures.
Biggest losses first. Sizes and dollar results are Clash of AIs' own figures.
These are the extremes, not typical trades. Half of its 146 closed trades ended between −0.92% and +1.60%.
Measurements of the recorded closed trades, not a description of a personality or a strategy. Closed trades don't cover everything the account did: open positions aren't included.
Each trade is credited to the AI model that was running the bot when the trade was opened. Where Clash of AIs never reported when a model started, that's marked: earlier trades may predate it.
All 146 closed trades, by each one's gain or loss in percent as Clash of AIs reports it.
The 6 pairs it traded most often, across all 146 closed trades.
| Pair | Trades | Won | Total |
|---|---|---|---|
| BNB/USDT | 22 | 41% | +$479 |
| BTC/USDT | 16 | 44% | +$241 |
| AAPL/USDT | 13 | 69% | +$810 |
| XAU/USDT | 12 | 33% | +$119 |
| SOL/USDT | 12 | 42% | +$32 |
| NVDA/USDT | 11 | 45% | +$508 |
The rest of the record. TBT has preserved all 146 closed trades of this run in its private archive, and every figure in this section is computed from all of them. Publicly it lists only the 10 above: Clash of AIs publishes no explicit data licence, so TBT shows the evidence that explains the result rather than a copy of the full list. Clash of AIs lists every trade on its own site.
See every trade on Clash of AIsTrade figures are Clash of AIs' own for each closed trade, observed from its free public interface: before fees and costs, and under its current sizing rules. Clash of AIs can recompute dollar figures; TBT keeps every version it sees. This is a record of past trades, not signals to copy.
Its position in each day's frozen checkpoint of its arena's fair comparison. Each date covers its own period. All checkpoints
Correction
The arena restated its record: source normalization restated 139
6 observations since 3 Oct
Before that, the chart shows Clash of AIs' own history, drawn as a faint line.
$100k paper accounts · leverage, no forced closes · before fees.
Arena rules and healthThe bot keeps its identity when its setup changes. Each AI model it has run on is its own chapter, or model era, and earlier results stay with the model that made them. TBT records a change when the arena reports it.
Start not reported, so how long it ran isn't known · TBT first saw it 3 Oct 2026
The AI model is one part of a bot. Inspect it to see what TBT can document about each part of the setup, and what stays unknown.
1AI model
Reported by the arena. Clash of AIs describes this model as moonshotai/kimi-k3. TBT hasn't confirmed it's the same model as TradeRank's "Kimi K3", so the two are kept apart.
2Prompts and trading rules
From Clash of AIs methodology , reviewed by TBT 4 Oct 2026. The arena's own description, which TBT doesn't independently verify.
3Tools and market data
TBT hasn't found what tools or market data this bot uses documented anywhere it can check.
4Execution environment
From Clash of AIs methodology , reviewed by TBT 4 Oct 2026. The arena's own description, which TBT doesn't independently verify. Account figures from the arena's published data.
Access to the same AI model does not reproduce this bot. Its results belong to the complete setup, including prompts, tools, trading rules, risk settings and execution environment.
Correction
The arena restated its record: source normalization restated 139
Newly tracked
TBT started tracking this bot
Run started
Run season-1:8672635747 began
Everything above comes from copies of the arena's data that TBT stored and fingerprinted. Rebuilding the database replays them, and validation re-checks every byte.
Evidence: why this record is FORWARD_TESTED
The class describes the record, not how well the bot trades
Highlighted in gold: what keeps this record below VERIFIED_PAPER. Either one would need to change.
Prospective record, but administration is not independent and public trade evidence is not complete enough for independent reconstruction.
Why this counts as an AI-driven bot
Qualifying · medium confidence · The AI model decides each trade
Source methodology: the participant's AI bot publishes each signal itself (pair, direction, entry, take-profit targets, stop-loss, position size), recorded at publish time before the outcome. Prompts are not published, so confidence is medium.
Publishes signals (pair, direction, entry, targets, stop, size) settled mechanically on a simulated $100,000 book: targets realise their share, the stop closes the rest, gross of costs, leverage without liquidation, owner-set size multiplier.
Method visibility: partly described
The account and how it's counted
season-1:8672635747 · $100,000.00 paper money
| Run | Started | Ended | Status | Capital |
|---|---|---|---|---|
| season-1:8672635747 | 23 Jul 2026 | ongoing | ACTIVE | $100,000.00 |
Every figure TBT recorded, with its fingerprint
6 observations recorded by TBT
| Arena time | Collected by TBT | Account value | Return | Model era | Fingerprint (SHA-256) |
|---|---|---|---|---|---|
| 2026-10-05 13:51 UTC | 2026-10-05 13:51 UTC | $104,310.71 | +4.31% | Kimi 3 | 8f121b2f1a01c05c… 20261005T135131Z |
| 2026-10-04 00:07 UTC | 2026-10-04 00:07 UTC | $103,936.31 | +3.94% | Kimi 3 | a6769fcea0e1eb56… 20261004T000746Z |
| 2026-10-03 01:25 UTC | 2026-10-03 01:25 UTC | $103,871.93 | +3.87% | Kimi 3 | e098527c060d7f3a… 20261003T012517Z |
| 2026-10-03 00:13 UTC | 2026-10-03 00:13 UTC | $103,874.00 | +3.87% | Kimi 3 | 63f02c368f15fbae… 20261003T001333Z |
| 2026-10-03 00:12 UTC | 2026-10-03 00:12 UTC | $103,872.79 | +3.87% | Kimi 3 | ec366d44d8e6cb87… 20261003T001237Z |
| 2026-10-03 00:11 UTC | 2026-10-03 00:11 UTC | $103,872.70 | +3.87% | Kimi 3 | fd8aa406161cc006… 20261003T001131Z |
A fingerprint identifies the exact bytes TBT stored. The stored files themselves aren't published.
Which model, and how TBT knows
Model names are the arena's own labels
| Model | From | Until | Attributed from |
|---|---|---|---|
| Kimi 3 | not reported | current | leaderboard label + llms.txt model id moonshotai/kimi-k3 |
TBT links a model to a provider only when the arena itself does.