Position Sizing¶

Position sizing determines how much to trade when signals arrive. Wu provides four approaches, each with different characteristics.

Position Sizing Types¶

WU_POSITION_SIZE_ABS¶

Absolute sizing trades a fixed quantity regardless of capital or prices. Setting size_value = 100.0 means always trading exactly 100 shares.

When to use: Fixed quantity strategies, replicating specific portfolios, or testing strategies where position size shouldn't vary.

Limitation: Doesn't scale with capital. If your portfolio grows from $100k to $1M, you're still trading 100 shares. Similarly, if a stock splits or reprices, your absolute quantity might become inappropriate.

WU_POSITION_SIZE_PCT¶

Percentage sizing allocates a fixed percentage of available cash per trade. With size_value = 0.5, each buy signal consumes 50% of current cash.

When to use: Most single-asset strategies. The position size scales naturally as capital grows or shrinks. If you double your money, position sizes double automatically.

Multi-asset behavior: Each asset uses the specified percentage of available cash independently. With two assets at 50% each, if both signal buys simultaneously, you allocate 100% of cash (50% + 50%). Cash splitting handles this (see below).

Cash management: "Available cash" means current cash in the account. If you have $100k and buy $50k worth, the next buy at 50% will use $25k (50% of remaining $50k), not $50k.

WU_POSITION_SIZE_PCT_EQUAL¶

Equal allocation divides portfolio value equally across all assets. Setting size_value = 0.9 with three assets gives each asset 30% of total portfolio value (90% ÷ 3).

When to use: Balanced portfolios, index replication, or strategies that want equal risk across assets (assuming similar volatilities).

Calculation: Uses total portfolio value (cash + positions), not just cash. This provides automatic rebalancing. If one asset grows faster than others, its allocation naturally exceeds its share. On the next trade, it gets allocated less to bring things back into balance.

Multi-asset requirement: Only makes sense with multiple assets. For single-asset portfolios, use PCT instead.

WU_POSITION_SIZE_STRATEGY_GUIDED¶

The strategy controls allocation by setting each signal.quantity field to the desired portfolio percentage (0.0 to 1.0). The portfolio allocates that percentage of total portfolio value to the trade.

When to use: Sophisticated allocation strategies where sizing depends on signal strength, volatility, correlation, or other factors. Risk parity, inverse volatility weighting, and momentum-based sizing all fit this model.

Implementation: Your strategy calculates desired allocations and sets them in signals:

signals[0].quantity = 0.6;  // 60% to first asset
signals[1].quantity = 0.4;  // 40% to second asset

The portfolio converts these percentages to actual quantities based on current prices and portfolio value.

Cash Splitting¶

When multiple buy signals arrive simultaneously with PCT sizing, cash splitting prevents over-allocation. Without splitting, each signal would try to use 50% of cash, potentially allocating 100% or more total.

How it works: The portfolio counts buy signals, divides available cash equally among them, and processes each signal with its cash allocation. If you have $100k and two buy signals at 50% each, each signal gets $50k to work with.

When it applies: Only for WU_POSITION_SIZE_PCT when multiple buy signals arrive in the same update. Other sizing modes don't need splitting because they have different allocation semantics.

Strategy-guided exception: If any signal in the batch has non-default quantity, cash splitting is disabled. The strategy is assumed to be managing allocations explicitly.

Fractional Shares¶

Wu supports fractional shares. Position quantities are double, not integers. If you have $1000 cash, 50% sizing, and a stock at $333.33, you buy 1.500015 shares.

Most brokers now support fractional shares for many stocks, making this reasonably realistic. For assets that don't support fractional trading (options, futures), you'd need to implement quantity rounding in a custom portfolio.

Position Limits¶

The portfolio doesn't enforce position count limits. You can accumulate multiple positions in the same asset (buy multiple times without selling between). The portfolio tracks them as separate position records with different entry prices and timestamps.

This matches real trading where you might scale into positions over time. When selling, the portfolio closes all positions in that asset simultaneously.

Margin and Leverage¶

For long positions, you can spend cash you have. Borrowing (borrow_params) only applies to short positions. The borrow_params.limit caps the total dollar value of short positions, effectively limiting leverage.

There's no leverage on the long side. If you want to model margin trading for long positions (borrowing cash to buy more), implement a custom portfolio with that logic.

Design Evolution¶

Position sizing is an area under active exploration. Current implementation handles common cases but has rough edges:

• Cash splitting logic is heuristic
• No sophisticated risk-based sizing (Kelly criterion, etc.)
• Rebalancing is implicit, not explicit
• No transaction cost awareness in sizing decisions

These limitations are acceptable for a learning project but would need refinement for serious backtesting.