[v9.0] feat: add short selling support to vector backtest engine

[MDUYN]

Summary

Add short selling support to the vector backtest engine. This enables strategies to profit from price declines by selling an asset they don't own (opening a short) and buying it back later (covering).

The vector engine is the ideal starting point because it bypasses order validation and database sync — orders are created as in-memory domain objects marked immediately as CLOSED. This makes the implementation self-contained and low-risk.

Motivation

Many trading strategies rely on short selling (mean reversion, pairs trading, market-neutral, etc.). Currently the framework only supports long-only strategies. Adding shorts to the vector engine first covers the most common use case: backtesting short strategies at high speed.

Design

Signal API

Add two new signal methods to TradingStrategy, following the existing pattern:

class MyStrategy(TradingStrategy):
    def generate_short_signals(self, data) -> Dict[str, pd.Series]:
        """Return boolean Series indicating when to open a short position."""
        ...

    def generate_cover_signals(self, data) -> Dict[str, pd.Series]:
        """Return boolean Series indicating when to close a short position."""
        ...

Both default to None (no short trading), maintaining full backward compatibility.

Signal Priority (extended from current)

for each bar, for each symbol:
    ├─ has open orders? → SKIP
    ├─ in cooldown? → SKIP
    │
    ├─ SELL signal AND has long position → close long (existing)
    ├─ COVER signal AND has short position → close short (NEW)
    │
    ├─ no position:
    │   ├─ BUY signal → open long (existing)
    │   └─ SHORT signal → open short (NEW)
    │
    └─ has position → scale-in/scale-out logic (existing)

Short Order Mechanics

Opening a short (_open_short_trade):

• Create a SELL Order with OrderSide.SELL and metadata {"is_short": true}
• Amount = capital / fill_price (same sizing as long entries, respects PositionSize)
• Fill price = TradingCost.get_sell_fill_price(price) (includes slippage)
• Cash effect: unallocated += proceeds (you receive cash when selling short)
• Track as negative position amount internally via last_trade

Covering a short (_close_short_trade):

• Create a BUY Order with metadata {"is_cover": true}
• Fill price = TradingCost.get_buy_fill_price(price)
• Cash effect: unallocated -= cost (you pay cash to buy back)
• P&L = (open_price - close_price) * amount (inverted from long)

Trade Model

• Trade.cost = proceeds received when opening short
• Trade.net_gain = (open_price - cover_price) * amount - fees
• No new fields needed — use existing orders list on Trade (first order = short, second = cover)

Position Tracking

• Use the existing last_trade dict in the vector loop
• Add a is_short boolean flag per symbol: sym_data['is_short'] = True
• Position.amount already accepts negative values

Files to Change

File	Change
investing_algorithm_framework/app/strategy.py	Add generate_short_signals() and generate_cover_signals() stub methods
investing_algorithm_framework/infrastructure/services/backtesting/vector_backtest_service.py	Add SHORT/COVER branches to the main loop, _open_short_trade(), _close_short_trade()
investing_algorithm_framework/domain/models/trade/trade.py	Ensure P&L calculation handles inverted short trades (or add direction field)

Scope & Non-Goals

In scope:

• Short/cover signals in vector backtest
• Correct P&L calculation for short trades
• Slippage and fees via TradingCost
• Position sizing via PositionSize
• Cooldown support for short entries

Out of scope (separate issues):

• Event backtest engine short support → [v9.0] feat: add short selling support to event backtest engine #434
• Live trading short support → [v9.0] feat: add short selling support for live trading #435
• Stop loss / take profit for short positions
• Margin requirements / leverage limits
• Short borrowing costs

Estimated Effort

~300-400 lines of code, 1-2 weeks. Low risk — changes are isolated to the vector loop with no database or validation impact.

Related Issues

• [v9.0]: Defer trade creation to order fill — order lifecycle refactor #431 — v9.0 architectural refactor (defer trade creation)
• Add market order support (estimated price + reconciliation) #430 — Market order support (merged via feat: add market order support (estimated price + reconciliation) #432)