Add hypothesis property-based tests for statistical invariants; fix sortino ZeroDivisionError

[Copilot]

Adds property-based tests using hypothesis to verify mathematical invariants that must hold for any valid returns series. In doing so, exposes and fixes a latent ZeroDivisionError in sortino() when no negative returns are present.

New: property-based tests (tests/test_jquantstats/test_properties.py)

Six invariants tested across arbitrary generated return sequences:

Test	Invariant
test_sharpe_nonnegative_for_all_positive_returns	Sharpe ≥ 0 when all returns > 0
test_max_drawdown_always_nonnegative	Max drawdown ≥ 0 (any returns)
test_max_drawdown_zero_for_nonnegative_returns	Max drawdown = 0 when all returns ≥ 0
test_win_rate_in_unit_interval	Win rate ∈ [0, 1]
test_volatility_nonnegative	Volatility ≥ 0
test_sortino_nonnegative_for_all_positive_returns	Sortino ≥ 0 (or ±∞) when all returns > 0

Tests use @pytest.mark.property (already registered in pytest.ini) and @settings(max_examples=100). The module uses pytest.importorskip("hypothesis") so tests are automatically skipped when hypothesis is not installed, and run normally when it is available — no changes to pyproject.toml are required.

Bug fix: sortino() ZeroDivisionError (_stats.py)

When every return is non-negative, downside_deviation = 0. Previously this raised ZeroDivisionError; now returns the mathematically correct value:

# Before: raises ZeroDivisionError
data.stats.sortino()  # with all-positive returns

# After: returns inf (no downside risk)
# mean > 0  → +inf
# mean < 0  → -inf  (degenerate)
# mean == 0 → nan

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Add property-based tests — Use hypothesis to test statistical invariants (e.g. Sharpe ≥ 0 when all returns positive, max drawdown ≤ 0)</issue_title>
<issue_description></issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes Add property-based tests — Use hypothesis to test statistical invariants (e.g. Sharpe ≥ 0 when all returns positive, max drawdown ≤ 0) #374

---

⚡ Quickly spin up Copilot coding agent tasks from anywhere on your macOS or Windows machine with Raycast.

[tschm]

@copilot: no need to install hypothesis

[Copilot]

@copilot: no need to install hypothesis

Removed hypothesis from pyproject.toml. The test module now uses pytest.importorskip("hypothesis") so tests skip gracefully when hypothesis isn't installed and run automatically when it is available. Fixed in commit 034a946.