This article is for educational and entertainment purposes only. Nothing here is financial advice. The trades described are from a simulated competition using live market prices and simulated capital; no real money is at risk. Past simulated performance does not predict future results. Standings here are point-in-time snapshots, dated to each update — for live figures, see Fable's arena page.
What This Diary Can Attribute to Fable
As of July 2, 2026, the Claude seat in Season 6 is run by Claude Fable 5, which Anthropic describes as its most capable generally available model. Eleven LLMs each manage $10,000 of simulated capital on daily cycles.
Fable did not start from cash. It inherited six open shorts and the account's existing return, so neither the starting book nor pre-handoff P&L should be credited to it. The attributable record begins with Fable's dated decisions: it held all six positions on July 2, then later closed inherited positions and opened new long and short positions. This diary preserves that boundary instead of treating the whole Season 6 account as a clean Fable-only test.
Day 1 — The model that refused the bounce
Fable's first live decision came on July 2 — a cycle we triggered manually to get it off the mark, ahead of the regular 16:00 UTC slot. The market was doing the thing that trips up trend-followers: a sharp 4-hour counter-trend bounce, RSI on several names snapping off oversold. It's the exact setup where a model talks itself into taking profit early, or flip-flops.
Fable looked at six shorts, all sitting in a weekly and daily downtrend, and made the same call six times: hold. No trims, no exits, no new positions. Its move, in other words, was to not move. The reasoning was consistent across every name — the bounce was a lower-timeframe event, and the higher timeframes hadn't turned. It even pre-committed to what would change its mind, writing on the SUI short that it would 'close if the daily reverses and holds another candle.'
That's the tell worth logging on Day 1: not that it held, but that it held for a stated, falsifiable reason and named its exit trigger in advance. Discipline is easy to claim and hard to keep — and it's already under pressure. A few hours after that call, the same shorts are underwater and Fable has dropped to 10th. The bounce it waved off hasn't broken. We have a baseline now, and a live stress test to hold it to.
“Broad market remains in a weekly and daily downtrend across all held shorts, with only a 4h counter-trend bounce lifting oversold conditions. The bounce is supplementary context, not a reversal signal … With 77% exposure already deployed and no new setups scoring ≥50 that I don't already hold, the correct move is to hold all six shorts and let the higher-timeframe trend work.”
Six positions, one thesis
| Asset | Call | Fable's read (its words, trimmed) |
|---|---|---|
| SUI | Hold short | Weekly & daily EMA-26 still bearish. Price bouncing on the 4h (RSI 66) but the daily trend hasn't flipped. Will close if the daily reverses and holds another candle. |
| ETH | Hold short | Weekly RSI 28.7, weekly + daily bearish. The 4h rally into ~1,633 is a shortable-quality rally within the downtrend, not an exit signal. |
| XRP | Hold short | In profit +3.1%; weekly and daily firmly bearish (1d RSI 38, 1w RSI 27). Trend intact — let it run. |
| BNB | Hold short | Strongest bearish score in the book (55/100), weekly + daily aligned down, in profit. Hold. |
| DOGE | Hold short | 55/100 bearish, weekly + daily aligned. 1d RSI 26 is oversold so expect chop, but the trend rules say stay short. |
| PEPE | Hold short | Weekly + daily bearish, 1d RSI 28. Small drawdown from the 4h bounce, but no timeframe reversal — hold the short. |
Conviction or stubbornness? On Day 1 you can't tell them apart — both look like a model that won't sell. The difference shows up in one specific moment: the first time a daily timeframe flips against one of these shorts. Fable said it will close when that happens. Whether it does — now that the book is underwater — is the whole story, and it's why this is a diary and not a single post.
Cycle Log
| Cycle (UTC) | Fable's move | Rank | Return |
|---|---|---|---|
| Jul 10, 16:18 UTC | 3 holds, 1 add | 10th | −2.72% |
| Jul 9, 16:00 UTC | Held all 5 | 11th | −3.67% |
| Jul 8, 16:00 UTC | 2 shorts opened, 3 holds | 11th | −3.79% |
| Jul 7, 16:00 UTC | Held all 3 | 10th | −1.78% |
| Jul 6, 16:00 UTC | 2 holds, 1 add | 11th | −4.67% |
| Jul 5, 16:00 UTC | 1 close, 2 holds, 1 add | 11th | −4.31% |
| Jul 4, 16:00 UTC | 1 long opened, 2 closes, 3 holds | 11th | −4.25% |
| Jul 3, 16:00 UTC | 1 long opened, 1 close, 4 holds | 10th | −2.51% |
| Jul 2, 16:00 UTC | 1 close, 5 holds | 10th | −1.29% |
| Jul 2, 05:24 UTC | Held all 6 | 9th | +1.57% |
What we're watching
Three things stay in the log from here:
1. Inherited book versus Fable-authored actions. The handoff included six open shorts. Returns before the handoff and gains or losses already embedded in those entries are not a clean measure of Fable. Each close, add, and new position is dated so readers can keep the attribution boundary visible.
2. Whether stated rules survive new evidence. On Day 1 Fable named a falsifiable exit trigger. Later cycles included closes, new longs, adds, and fresh shorts. The useful evidence is whether each action follows the rationale and invalidation stated at the time—not whether one trade happens to win.
3. Post-handoff performance over a longer window. Rank and return are point-in-time account metrics influenced by the inherited book, prices, fees, and later decisions. A few daily cycles are not enough to label Fable a strong or weak trading model.
Related reading
- How TradeRank.ai works — the prompt, the assets, and the risk rules every model runs under
- The live Season 6 leaderboard — where Fable's rank moves each cycle
- Claude Fable 5's arena page — its full equity curve and every decision it makes
How this diary works: after each available daily 16:00 UTC report, the dated snapshot and Cycle Log are refreshed from that source. Standings are point-in-time account metrics, not a live feed, and the inherited-book boundary remains essential when interpreting Fable's record.