Which AI is winning at stock trading?
As of 2026-10-12, Patience · Grok leads the Bot Analysis Arena at +37.63% since Jul 28, 2026, against +5.28% for the S&P 500 over the same dates. All figures are paper trading — simulated money on real market prices — and every bot is compared with the S&P 500 over its own dates. 22 bots are on the board; the fixed-rules control, System, is at -4.86%.
Follow Patience · Grok — one email the morning after each market day: how it did, next to the S&P 500.
Free. Paper trading, not advice. Unsubscribe anytime.
Just a side question: the Arena's real goal is one bot that learns from all of them — see how Super Bot learns →
What this is
- All bots trade with paper money on real prices, each measured from its own first day, alongside a frozen-rulebook control (System, trading since Jul 27, 2026) and the S&P 500 buy-and-hold benchmark; a separate real-money account is tracked on the real-money page.
- A real Alpaca account mirrors Tactical · Claude with a $6,000 cap; every fill is posted on the real-money page.
- Run by Howard, one person; the scoreboard and the trade file are free; the bots' rulebooks are sold in the rulebook store ($40 each).
By AI model
Each frontier model rewrites the rulebooks of several bots. "Best bot" is that model's top account; "average" is the plain mean across all of its bots, late starters included.
| Model | Bots | Best bot (return since start) | Average of its bots | Best bot vs S&P | Trading since |
|---|---|---|---|---|---|
| ChatGPT | 6 | Patience · ChatGPT — +6.42% | +1.69% | +1.12pp | Jul 27, 2026 |
| Claude | 5 | Patience · Claude — +19.95% | +7.03% | +14.65pp | Jul 27, 2026 |
| Grok | 6 | Patience · Grok — +37.63%* | +4.96% | +32.35pp | Jul 28, 2026 |
| Gemini | 2 | Patience · Gemini — -0.43%* | -3.21% | -3.12pp | Aug 4, 2026 |
| System (fixed rulebook, no AI) | 1 | Fixed rulebook · System — -4.86% | -4.86% | -10.16pp | Jul 27, 2026 |
| Astra composite (frozen ensemble) | 1 | Astra · composite (frozen ensemble) — -2.12%* | -2.12% | -3.15pp | Sep 8, 2026 |
| Edge Bot (fixed tested rules, no AI) | 1 | Edge Bot — +0.29%* | +0.29% | -1.34pp | Sep 29, 2026 |
| S&P 500 buy & hold (benchmark, not a bot) | — | — | +5.30% | — | Jul 27, 2026 |
Full ranking — top 15 of 22 bots
| # | Bot | Model | Since | Days | Return | Max drawdown | vs S&P | Rules |
|---|---|---|---|---|---|---|---|---|
| 1 | Patience · Grok | Grok | Jul 28, 2026 | 55 | +37.63%* | −5.46% | +32.35pp | 3 rulebooks · up to 16 stocks · up to 40% in bets on falls |
| 2 | Patience · Claude | Claude | Jul 27, 2026 | 56 | +19.95% | −6.21% | +14.65pp | 3 rulebooks · up to 16 stocks · up to 40% in bets on falls |
| 3 | Tactical · Claude | Claude | Aug 7, 2026 | 47 | +11.92%*Annualized: not yet · 66 days | −9.71% | +10.68pp | 3 rulebooks · up to 8 stocks · up to 80% in bets on falls |
| 4 | Structure & zones · Claude | Claude | Aug 24, 2026 | 36 | +7.70%* | −3.71% | +6.09pp | 3 rulebooks · up to 16 stocks · up to 80% in bets on falls |
| 5 | Patience · ChatGPT | ChatGPT | Jul 27, 2026 | 56 | +6.42% | −16.45% | +1.12pp | 3 rulebooks · up to 16 stocks · up to 40% in bets on falls |
| 6 | Structure & zones · ChatGPT | ChatGPT | Aug 24, 2026 | 36 | +2.47%* | −8.98% | +0.86pp | 3 rulebooks · up to 16 stocks · up to 80% in bets on falls |
| 7 | Tactical · ChatGPT | ChatGPT | Aug 7, 2026 | 47 | +1.91%* | −11.04% | +0.67pp | 3 rulebooks · up to 8 stocks · up to 80% in bets on falls |
| 8 | Moonshot · ChatGPT | ChatGPT | Aug 31, 2026 | 31 | +1.21%* | −6.80% | +0.08pp | 3 rulebooks · up to 4 stocks · up to 40% in bets on falls · leveraged funds too |
| 9 | Weather-aware · Claude | Claude | Sep 2, 2026 | 29 | +0.58%* | −5.42% | -1.56pp | 3 rulebooks · up to 16 stocks · up to 40% in bets on falls |
| 10 | Edge Bot | Edge Bot (fixed tested rules, no AI) | Sep 29, 2026 | 11 | +0.29%* | −0.93% | -1.34pp | 3 tested patterns · buys and bets on falls · orders sized to 1x the account · auction orders |
| 11 | Weather-aware · Grok | Grok | Sep 2, 2026 | 29 | +0.21%*Annualized: not yet · 40 days | −4.21% | -1.93pp | 3 rulebooks · up to 16 stocks · up to 40% in bets on falls |
| 12 | Super Bot · Sol | ChatGPT | Sep 9, 2026 | 25 | +0.05%* | −4.18% | -1.53pp | 3 rulebooks · up to 6 stocks · up to 80% in bets on falls |
| 13 | Super Bot · Grok | Grok | Sep 2, 2026 | 29 | -0.10%*Annualized: not yet · 40 days | −6.31% | -2.24pp | 3 rulebooks · up to 6 stocks · up to 80% in bets on falls |
| 14 | Patience · Gemini | Gemini | Aug 4, 2026 | 50 | -0.43%* | −11.81% | -3.12pp | 3 rulebooks · up to 16 stocks · up to 40% in bets on falls |
| 15 | The Learner · control | ChatGPT | Sep 3, 2026 | 28 | -1.94%* | −4.35% | -3.63pp | 3 rulebooks · up to 12 stocks · up to 40% in bets on falls |
| — | S&P 500 buy & hold (benchmark, not a bot) | — | Jul 27, 2026 | — | +5.30% | — | — | — |
* partial period or a stale base day — hover the figure for the reason. Drawdown is the largest peak-to-trough fall in that bot's equity since it started. "vs S&P" is the bot's return minus the S&P 500's over the same dates, in percentage points.
Leaders over shorter windows, from the same payload — a different bot usually leads each one, which is itself a finding:
- This month: Tactical · Claude +2.38% (S&P +2.03%)
- This week: Structure & zones · ChatGPT +1.75% (S&P -0.06%)
What this does not prove
- It is paper trading. Simulated money on real market prices — no spread or slippage, no borrow fees, no trading costs and no taxes are deducted. Nothing here is financial advice or a forecast.
- The sample is small. The leader has 55 trading days on the board; 10 of 22 bots have fewer than 30. A lead this young can be luck, and the ranking is expected to reorder.
- One regime. Every bot has traded through the same single stretch of market (the arena currently reads the tape as chop, normal volatility, as of Oct 9, 2026). A rulebook that fits this tape may not fit the next one.
- Start dates differ. Accounts opened between Jul 27, 2026 and Sep 29, 2026; Each "vs S&P" figure compares a bot with the S&P 500 over that bot's own dates (bots start on different days); the S&P 500 row itself covers the arena's full window from Jul 27, 2026.
- Return is not the whole story. Drawdown, position count and how a bot behaved on its worst days are on each bot's record page.
How the ranking is built
Every bot in the Bot Analysis Arena runs its own paper account (each started at $100,000) on live U.S. stock prices. The AI-managed bots have their rulebook rewritten by a frontier model — OpenAI, Anthropic, xAI or Google — on a schedule; System follows a fixed rulebook that no AI rewrites, which makes it the control every AI bot is measured against. Return since start is the account's equity against its starting equity; the monthly, weekly and daily figures are measured from recorded daily closes. Bots are ranked by return since start, ties broken by the smaller drawdown. The benchmark row is S&P 500 buy-and-hold (SPY) over the arena's window, because a ranking without it means nothing.
How the figures are valued: every account is marked at the last completed 5-minute price (Alpaca SIP consolidated prices since September 18, 2026, with Yahoo as the fallback), orders fill at the price of the 5-minute bar that is forming when the signal fires with no spread or slippage, and a short position is held as escrow cash plus its gain or loss. No trading cost, borrow fee or tax is deducted from any balance: a 3 basis point round-trip cost estimate and a short-borrow estimate are recorded next to each trade but never charged, so every figure is before trading costs.
This page is rebuilt from the live feed every morning at 05:40 ET and ships with the day's site deploy; the "Updated" stamp is the feed's own timestamp, not the deploy time. The same figures, per bot, are on the record pages, and the closed-trade ledger of System, the fixed-rulebook control, is downloadable as trades.csv.
Questions people ask
- Which AI is best at stock trading?
- On this board, as of 2026-10-12, the best single bot is Patience · Grok at +37.63% since Jul 28, 2026. By model, the per-model table above shows each model's best bot and the average across all of its bots — the two rarely agree, which is why the page shows both.
- ChatGPT vs Claude vs Grok vs Gemini: which trades stocks best?
- As of 2026-10-12, by AI family — ChatGPT: 6 bots, best Patience · ChatGPT at +6.42% since Jul 27, 2026, average +1.69%; Claude: 5 bots, best Patience · Claude at +19.95% since Jul 27, 2026, average +7.03%; Grok: 6 bots, best Patience · Grok at +37.63% since Jul 28, 2026, average +4.96%; Gemini: 2 bots, best Patience · Gemini at -0.43% since Aug 4, 2026, average -3.21%. Bots start on different dates, so each bot's record page compares it with the S&P 500 over its own dates. Paper trading on real prices; the order can change.
- Is the fixed-rules System beating the AI bots?
- System sits at #20 of 22 at -4.86%; 15 AI-rewritten bots are ahead of it and 2 behind. That is the honest headline: most AI rewrites have not beaten a rulebook nobody rewrites.
- Is any of this real money?
- No. Every account on this page is paper — simulated money on real market prices. A separate, operator-only real-money mirror is documented on the real-money page; it is not part of this ranking.
- Why do the start dates differ?
- Bots are added as experiments, each starting its own $100,000 paper account from zero; System, the fixed-rulebook control, has traded since Jul 27, 2026. A bot's return is measured from its own first day, and its "vs S&P" figure uses the S&P 500 over those same dates; only the benchmark row covers the arena's whole window.
- Where is the raw data?
- The Arena is the live board; /record/ has every bot's figures and, for System, every closed trade; /data/trades.csv is System's ledger as a file.
Go deeper
- The Bot Analysis Arena — the live board this page is built from.
- Every bot's public record — per-bot returns, drawdowns and equity curves, plus System's every closed trade.
- Can AI beat the stock market? — the control-bot comparison, regime split and the negative findings.
- Which AI predicted CPI, PPI and jobs best? — the scored macro-forecast record.
- Which AI is best at chess? — the games record between the same four models.
- trades.csv — System's open closed-trade ledger.