council-book
PAPER · NO BROKER

Agents / Decision

Single-agent control LLM

One agent with the same facts, no analysts and no debate, as a comparison; it never trades.

Latest: same as the council → the core held

PAPER Decision #4

Its last 5 meetings, newest first

  1. #4usable
  2. #3usable
  3. #2usable
  4. #1usable
  5. 25 Sep, 14:40 UTCusable

✓ a usable reply · ✗ the call failed · — it did not run

What it does

Published so readers can see what the council's analysts and debate change.

Kind
a language model
Model
deepseek-v4.1-flash:cloud

Prompts in force: council-single_agent/v1 a66e23487d21. Each run records the hash of the prompts it used; a prompt change is a versioned policy change.

Its numbers

Usable replies
15 of 15 calls
Failed calls
0
none · 1 fixed by a correction
Median time
1.0 s
per call
Meetings
5
4 paper · 1 live or rehearsal
Differs from the council
1 of 5
runs

Failed calls

No failed call in the published runs.

History

Newest first. The latest is open; each other meeting opens on a click and links to its run or decision page.

#4 · 2 Oct 2026, 18:40 UTC PAPER ok same as the council

Open decision #4 →

  1. Bull opened: add Crude oil size 0.00 → 0.25 rebuttal: keep the reference
  2. Bear asked to keep the reference
  3. Manager decided to lever up Nasdaq-100 35% → 35.7%; lever up Semiconductors 15% → 15.4%; lever up S&P 500 15% → 15.3%; add Gold 3% → 3.1% (sided with the reference, 3 of 3 attempts)
  4. Risk engine 32/32 checks passed, held back Gold, Bitcoin, Ether
  5. Human not needed: a paper run
Single-agent control ×3 — 3 calls

call single_agent --attempt 1 --prompt council-single_agent/v1

sha a66e23487d21 · 5,480 tokens in · 137 out · 1.2 s

ok

call single_agent --attempt 2 --prompt council-single_agent/v1

sha a66e23487d21 · 5,480 tokens in · 135 out · 1.2 s

ok

call single_agent --attempt 3 --prompt council-single_agent/v1

sha a66e23487d21 · 5,480 tokens in · 143 out · 1.1 s

ok

3 calls · median 1.2 s · 16,440 tokens in / 415 out

Control — one agent, no analysts or debate: lever up Nasdaq-100 from 35% to 35.7%, lever up Semiconductors from 15% to 15.4%, lever up S&P 500 from 15% to 15.3% and add to Gold from 3% to 3.1% of the portfolio — same as the council. It never trades.

Attempt 1

ok 1.2 s 5,480 in / 137 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: Core lines frozen with bands pinned; overlay adds (OIL, GBPUSD) fail on negative short-term momentum and carry cost, so the reference stands everywhere.
Decisive fact
All core lines are frozen (market closed) with bands pinned to current levels; OIL and GBPUSD overlays show negative 10d momentum (-3.41%, -1.08%) against their uptrends, so no add is justified. Crude oil10-day change: −3.41%

Attempt 3 DISCARDED

ok 1.1 s 5,480 in / 143 out council-single_agent/v1

Discarded by the auditor: decisive fact: cited evidence that is not in the fact pack.

Attempt 2: same call as attempt 1 (ok · 1.2 s)

Attempt 2

ok 1.2 s 5,480 in / 135 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: Core lines frozen with bands pinned; OIL/GBPUSD adds fail on negative short-term momentum and carry cost, so the reference stands everywhere.
Decisive fact
All core lines are frozen (market closed) with bands pinned to current levels; OIL and GBPUSD are the only addable lines but both show negative 10d momentum (-3.41%, -1.08%) against their uptrends. Crude oil10-day change: −3.41%
#3 · 2 Oct 2026, 14:40 UTC PAPER ok after a correction differs on GOLD

Open decision #3 →

  1. Bull opened: keep the reference rebuttal: cut Gold 3% → 0%
  2. Bear asked to cut Gold 3% → 0%
  3. Manager decided to cut Gold 3% → 0% (sided with the bear, 3 of 3 attempts)
  4. Risk engine 32/32 checks passed, held back Nasdaq-100, Semiconductors, S&P 500, Gold, Bitcoin, Ether
  5. Human not needed: a paper run
Single-agent control ×3 — 3 calls

call single_agent --attempt 1 --prompt council-single_agent/v1

sha a66e23487d21 · 5,418 tokens in · 136 out · 1.7 s

ok

call single_agent --attempt 2 --prompt council-single_agent/v1

sha a66e23487d21 · 5,418 tokens in · 149 out · 1.5 s

ok

call single_agent --attempt 3 --prompt council-single_agent/v1

sha a66e23487d21 · 11,068 tokens in · 318 out · 4.5 s

ok after a correction (one correction turn)

3 calls · median 1.7 s · 21,904 tokens in / 603 out

Control — one agent, no analysts or debate: held the reference — differs from the council on Gold. It never trades.

Attempt 1

ok 1.7 s 5,418 in / 136 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: No qualifying card exists; all bands pin the reference or offer no cost-justified deviation. Holding the mechanical reference is the correct action.
Decisive fact
Every equity and crypto line is in an uptrend with bands pinned at the reference; GOLD is the only line with a cut option but its -3.4% vs 50-day average and -3.9% 10-day momentum are already expressed by ref +0.25. Goldtrend: down
Attempt 2: same call as attempt 1 (ok · 1.5 s)

Attempt 2

ok 1.5 s 5,418 in / 149 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: No qualifying cards; bands pin NDX/SEMIS/SPX/BTC/ETH/EURUSD, and GOLD/OIL/GBPUSD deviations fail the cost gate against their small expected moves.
Decisive fact
Every equity and crypto line is in an uptrend with volatility at or below its one-year median (NDX 0.88x, SPX 0.80x, SEMIS 0.98x); no qualifying card exists to justify a cut. Nasdaq-100volatility vs its 1-year norm: 0.883
Attempt 3: same call as attempt 1 (ok after a correction · 4.5 s)

Attempt 3

ok after a correction 4.5 s 11,068 in / 318 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: No qualifying card exists; bands pin NDX, SEMIS, SPX, BTC, ETH, EURUSD, and GOLD's downtrend is already at ref +0.25. OIL and GBPUSD adds fail the cost gate on carry.
Decisive fact
All equity and crypto lines are in uptrends with bands pinned at reference; GOLD is the only cut direction open, but its -3.4% vs 50-day average and -3.9% 10-day momentum are already expressed by ref +0.25. Goldtrend: down
#2 · 1 Oct 2026, 18:40 UTC PAPER ok same as the council

Open decision #2 →

  1. Bull asked to keep the reference
  2. Bear asked to keep the reference
  3. Manager decided to lever up Nasdaq-100 35% → 36%; lever up Semiconductors 15% → 15.4%; lever up S&P 500 15% → 15.4%; add Gold 3% → 3.1% (sided with the reference, 3 of 3 attempts)
  4. Risk engine 32/32 checks passed, held back Gold, Bitcoin, Ether
  5. Human not needed: a paper run
Single-agent control ×3 — 3 calls

call single_agent --attempt 1 --prompt council-single_agent/v1

sha a66e23487d21 · 5,463 tokens in · 133 out · 1.0 s

ok

call single_agent --attempt 2 --prompt council-single_agent/v1

sha a66e23487d21 · 5,463 tokens in · 149 out · 1.0 s

ok

call single_agent --attempt 3 --prompt council-single_agent/v1

sha a66e23487d21 · 5,463 tokens in · 127 out · 1.0 s

ok

3 calls · median 1.0 s · 16,389 tokens in / 409 out

Control — one agent, no analysts or debate: lever up Nasdaq-100 from 35% to 36%, lever up Semiconductors from 15% to 15.4%, lever up S&P 500 from 15% to 15.4% and add to Gold from 3% to 3.1% of the portfolio — same as the council. It never trades.

Attempt 1

ok 1.0 s 5,463 in / 133 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: Core lines frozen with bands pinned; OIL and GBPUSD adds lack momentum confirmation and would pay carry for a weak signal. No qualifying cards exist.
Decisive fact
All core lines are frozen (market closed) with bands pinned to current levels; OIL and GBPUSD overlays show no momentum confirmation (OIL 10-day momentum -6.7%, GBPUSD 10-day momentum -0.8%). Crude oil10-day change: −6.73%
Attempt 2: same call as attempt 1 (ok · 1.0 s)

Attempt 2

ok 1.0 s 5,463 in / 149 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: Core lines frozen with bands pinned; OIL's 10-day momentum is negative (-6.7%) despite uptrend, and GBPUSD sits below both averages with negative 10-day momentum, so no add is justified.
Decisive fact
All core lines are frozen (market closed) with bands pinned to current levels; OIL and GBPUSD are the only addable lines but both carry 0.96 basis points/day and OIL 10-day momentum is -6.7%. Crude oil10-day change: −6.73%
Attempt 3: same call as attempt 1 (ok · 1.0 s)

Attempt 3

ok 1.0 s 5,463 in / 127 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: Core lines frozen with bands pinned; overlay adds lack momentum confirmation and would pay carry for no edge. Reference stands everywhere.
Decisive fact
All core lines are frozen (market closed) with bands pinned to current levels; OIL and GBPUSD overlays show no momentum confirmation (OIL 10-day momentum -6.7%, GBPUSD 10-day momentum -0.8%). Crude oil10-day change: −6.73%
#1 · 1 Oct 2026, 14:40 UTC PAPER ok same as the council

Open decision #1 →

  1. Bull asked to keep the reference
  2. Bear asked to cut Gold 3% → 0%
  3. Manager decided to keep the reference (sided with the reference, 2 of 3 attempts)
  4. Risk engine 32/32 checks passed
  5. Human not needed: a paper run
Single-agent control ×3 — 3 calls

call single_agent --attempt 1 --prompt council-single_agent/v1

sha a66e23487d21 · 5,414 tokens in · 142 out · 1.0 s

ok

call single_agent --attempt 2 --prompt council-single_agent/v1

sha a66e23487d21 · 5,414 tokens in · 147 out · 1.0 s

ok

call single_agent --attempt 3 --prompt council-single_agent/v1

sha a66e23487d21 · 5,414 tokens in · 121 out · 0.8 s

ok

3 calls · median 1.0 s · 16,242 tokens in / 410 out

Control — one agent, no analysts or debate: held the reference — same as the council. It never trades.

Attempt 1

ok 1.0 s 5,414 in / 142 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: No qualifying cards exist; every band is pinned at its reference (equities +1.00, crypto reference-only, FX pinned). Nothing material changed, so the reference stands.
Decisive fact
All equity lines are in uptrends with bands pinned at ref +1.00 and no qualifying cards; NDX vol 0.91x, SPX 0.83x, SEMIS 1.00x of 1y median. Nasdaq-100volatility vs its 1-year norm: 0.908
Attempt 2: same call as attempt 1 (ok · 1.0 s)

Attempt 2

ok 1.0 s 5,414 in / 147 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: No qualifying cards exist; every band pins the reference (equities at +1.00, GOLD [0,0.25], overlays 0). Nothing material changed, so the reference stands.
Decisive fact
All equity lines are in uptrends with bands pinned at ref +1.00 and no qualifying cards; NDX vol 0.91x, SPX 0.83x, SEMIS 1.00x of 1y median. Nasdaq-100trend: up
Attempt 3: same call as attempt 1 (ok · 0.8 s)

Attempt 3

ok 0.8 s 5,414 in / 121 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: Every line's band and reference already express the trend; no card, no cost-gate-clearing deviation, and no material change justifies a move.
Decisive fact
All equity lines are in uptrends with bands pinned at ref +1.00; no qualifying card exists to justify a cut, and no line clears the cost gate for an add. Nasdaq-100trend: up
25 Sep 2026, 14:40 UTC REHEARSAL ok same as the council

Open in the run →

  1. Bull asked to keep the reference
  2. Bear asked to keep the reference
  3. Manager decided to keep the reference (sided with the reference, 3 of 3 attempts)
  4. Risk engine 32/32 checks passed
  5. Human not needed: rehearsal
Single-agent control ×3 — 3 calls

call single_agent --attempt 1 --prompt council-single_agent/v1

sha a66e23487d21 · 5,002 tokens in · 134 out · 1.0 s

ok

call single_agent --attempt 2 --prompt council-single_agent/v1

sha a66e23487d21 · 5,002 tokens in · 141 out · 1.0 s

ok

call single_agent --attempt 3 --prompt council-single_agent/v1

sha a66e23487d21 · 5,002 tokens in · 145 out · 0.9 s

ok

3 calls · median 1.0 s · 15,006 tokens in / 420 out

Control — one agent, no analysts or debate: held the reference — same as the council. It never trades.

Attempt 1

ok 1.0 s 5,002 in / 134 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: No qualifying cards exist; every line's band pins or already matches its reference, and no deviation clears the cost gate with a citable reason.
Decisive fact
All equity lines are in uptrends with bands pinned at ref +1.00 and no qualifying cards; NDX vol 0.99x, SPX 0.88x, SEMIS 1.12x of 1y median. Nasdaq-100trend
Attempt 2: same call as attempt 1 (ok · 1.0 s)

Attempt 2

ok 1.0 s 5,002 in / 141 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: No qualifying cards; every line's reference already expresses its trend state. Overlays (OIL, GBPUSD) show negative 10d momentum and carry costs, so no adds are warranted.
Decisive fact
All equity and crypto lines are in uptrends with volatility at or below 1y median (NDX volatility vs its 1-year norm 0.99x, SPX 0.88x); no qualifying card exists to justify any cut. Nasdaq-100volatility vs its 1-year norm
Attempt 3: same call as attempt 1 (ok · 0.9 s)

Attempt 3

ok 0.9 s 5,002 in / 145 out council-single_agent/v1

Sided with
the reference
Changes
keep the reference: All equity lines uptrend with bands pinned at reference and no cards; GOLD mixed ref 0.75 is neutral; overlays ref 0; crypto reference-only. Nothing material changed.
Decisive fact
NDX and SEMIS are in uptrends with bands pinned at ref +1.00 and no qualifying card; SEMIS 3-month momentum -9.4% vs 10-day momentum +9.5% is a bounce, not a cut trigger. Semiconductors3-month change