feat(backtest): default crypto research wrapper to orchestrator

[Pigbibi]

Summary\n- Default crypto research wrapper to orchestrator/profile\n- Keep legacy path only for compatibility\n\n## Test Plan\n- python3 -m py_compile on modified script\n- --help smoke test

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

[chatgpt-codex-connector]

💡 Codex Review

CryptoStrategies/src/crypto_strategies/backtest/orchestrator_runner.py

Line 222 in 4947a29

sliced,

Trim lookback rows out of fold metrics

When start_date is supplied by BacktestOrchestrator.walk_forward, _slice_history prepends min_history_days + 5 days as indicator lookback, but this line feeds that entire slice into run_combo_backtest and then returns result.metrics unchanged. The reported start_date is reset to the requested fold start, so CAGR, drawdown, total return, and Trading Days for combo walk-forward folds include pre-window returns, corrupting the fold results. Please keep the lookback for warm-up but compute/report metrics only on returns dated >= start_date.

---

CryptoStrategies/src/crypto_strategies/backtest/combo_simulator.py

Line 149 in 4947a29

return equity.pct_change().fillna(0.0)

Remove DCA cash deposits from reported returns

For any combo backtest, portfolio_values grows from the new dca_amount_usd contribution added each day, and this line converts that externally contributed cash into investment returns for _performance_metrics. Even with flat BTC/ETH prices this reports positive CAGR/Sharpe from deposits rather than market performance, so orchestrator results are materially overstated; compute returns on capital excluding cash flows or report a money-weighted metric explicitly.

---

CryptoStrategies/src/crypto_strategies/backtest/orchestrator_runner.py

Lines 223 to 226 in 4947a29

	combo_config=CryptoComboBacktestConfig(
	combo_mode=cast(ComboMode, combo_mode),
	min_history_days=min_history_days,
	),

Honor combo allocation params in runner

When callers pass research params such as btc_weight, trend_weight, dca_amount_usd, or dynamic_trend_cut through the orchestrator, those values are recorded in the BacktestResult params but are not applied here because the config is built only from combo_mode and min_history_days. This makes parameter sweeps or single runs appear to test different allocations while every result still uses the dataclass defaults.

---

CryptoStrategies/src/crypto_strategies/backtest/combo_simulator.py

Lines 161 to 164 in 4947a29

	if len(close) < int(combo.min_history_days):
	raise ValueError(
	f"market_history requires at least {int(combo.min_history_days)} overlapping trading days"
	)

Validate overlapping BTC/ETH history before reporting metrics

This check counts dates in the pivoted frame, not dates where both required assets have usable closes. If ETH is missing, all-NaN, or non-overlapping for an otherwise long BTC history, the check passes, _combo_daily_returns returns an empty series, and the backtest reports all-zero metrics instead of rejecting invalid input; validate the overlapping non-null BTC/ETH index here.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.