Run 5: Run 4 Repeated All Runs Run 7: Largest Gap Run 6 4 arms

First 4-Arm Run

Sep 1, 2025 - Mar 6, 2026, 187 ticks, Opus 4.6

Treatment

+3.18%

Control

-9.51%

Control-WS

+0.94%

Placebo

-1.55%

Treatment vs Control

+12.69pp

Treatment vs Control-WS

+2.24pp

BTC Buy & Hold

-34.3%

Key Finding

Added a web search arm. The briefings arm finished 2.24 points above it and 12.69 above price only.

What Changed vs Run 5

Added 4th arm (Control-WS) that receives web search results instead of PreReason briefings. Tests whether the value is in the analysis, not just having information.

Market Context

Same window as Run 5: Sep 2025 through Mar 6, 2026. BTC fell 34.3%, from $108K to $71K. This run added a fourth arm, given web search results in place of briefings.

Briefings Used

Same briefings, but added a new Control-WS arm that gets web search results instead of briefings

btc.energy

Bitcoin energy model: hash rate, difficulty, miner economics as fundamental signals

cross.regime

Cross-asset regime classification: stocks, bonds, volatility, DXY correlation signals

cross.breadth

Market breadth analysis: how many assets confirm the current trend

btc.momentum

Bitcoin momentum indicators: trend strength, mean reversion signals, volume profile

See the exact preamble and briefing content for all arms

Observations

The web search arm returned +0.94%, 10.45 points above price only and 2.24 below briefings.

In the tick log, the web search arm made 37 trades, the briefings arm 23 and the price only arm 7.

Run 5: Run 4 Repeated All Runs Run 7: Largest Gap

Related Content

How We Test

The simulated backtest design and its controlled variables

Research Deep Dive

Full research analysis and findings across all runs

Context Engineering for Crypto Agents

Why context format affects agent decisions