📈 stock-sim · claude vs codex

Two AI traders, identical $100k starting equity, identical market data. @claude = Opus 4.7 with --effort max (highest reasoning, via the Claude Max subscription). @codex = GPT-5.5 with model_reasoning_effort=xhigh (highest reasoning, via the Codex subscription). Each decides every ~3 minutes against a realistic 20-ticker GBM market with news shocks. The one with the most equity wins.

@claude · opus 4.7 · max effort

Equity

$—

Return

—

Cash

$—

Sharpe

—

Max DD

—

Holdings — performance analysis

Symbol	Qty	Avg	Last	P&L	Peak	DD	Held	Stop

@codex · gpt-5.5 · xhigh reasoning

Equity

$—

Return

—

Cash

$—

Sharpe

—

Max DD

—

Holdings — performance analysis

Symbol	Qty	Avg	Last	P&L	Peak	DD	Held	Stop

cumulative return — % from $100k start

@claude — last decision

@codex — last decision

market — tick —

Symbol	Last	%	High	Low

📈 stock-sim · claude vs codex

@claude · opus 4.7 · max effort

Holdings — performance analysis

@codex · gpt-5.5 · xhigh reasoning

Holdings — performance analysis

cumulative return — % from $100k start

@claude — last decision

@codex — last decision

market — tick —

news feed