Skip to content

Self-Financing Hedge: Rebalancing Without External Cash

For educational purposes only; not investment advice.

A trading strategy is self-financing when every rebalance is paid for by assets already inside the portfolio: selling one holding, spending cash, or borrowing finances purchases of another holding. After inception there is no external cash contribution or withdrawal. This constraint prevents a replication argument from quietly creating money whenever its hedge ratio changes.

Self-financing does not mean costless, profitable, fully hedged, or free from funding needs. A portfolio can have a negative cash balance, incur interest and transaction costs, breach margin limits, and finish with a hedging loss while still satisfying the theoretical accounting identity.

Let Δ_t shares of stock priced S_t and a cash account B_t form a hedge with value:

V_t = Δ_t S_t + B_t

In the idealized continuous-time model, the self-financing condition is:

dV_t = Δ_t dS_t + r B_t dt

The term S_t dΔ_t is not a free gain or loss. When the stock holding changes, the cash account changes by the opposite purchase amount. With no costs at a discrete rebalance:

B_after = B_before − (Δ_after − Δ_before)S

If the hedge buys shares, cash falls or borrowing rises. If it sells shares, cash rises. Dividends, borrow fees, financing spreads, commissions, and bid-ask costs require additional cash-flow terms; omitting them is a model assumption, not self-financing proof.

In Black-Scholes replication, a sufficiently regular option value C(S,t) has hedge ratio Δ = ∂C/∂S. Under continuous trading, the stock and cash portfolio can reproduce the option payoff within the model. Real hedges rebalance at discrete times and therefore retain Gamma, jump, volatility, and execution risk.

Suppose a one-option hedge initially holds 40 shares because the option Delta is 0.40 and the standard multiplier is 100. Stock is $100, so the shares are worth $4,000. Let the cash account be −$3,200; portfolio value is:

V_before = 40 × $100 − $3,200 = $800

Delta rises to 0.60, requiring 60 shares. Buying 20 shares at $100 costs $2,000, funded entirely by the cash account:

B_after = −$3,200 − 20 × $100 = −$5,200

Immediately after the rebalance:

V_after = 60 × $100 − $5,200 = $800

The composition changed but the value did not jump. A backtest that adds the 20 shares without reducing cash would report a false $2,000 wealth increase.

Now include a $0.05 per-share spread-and-fee cost. Trading 20 shares costs another $1, leaving B_after = −$5,201 and V_after = $799. The strategy can still be tracked with complete cash accounting, but friction makes exact costless replication impossible.

  • Discrete rebalancing: price moves between hedge times, leaving Gamma-related error.
  • Jumps and gaps: the underlying can cross many hedge levels before a trade is possible.
  • Volatility model risk: realized dynamics may differ from the volatility used to compute Delta.
  • Funding: cash borrowing and stock financing occur at real rates, not one frictionless rate.
  • Trading frictions: spreads, fees, taxes, market impact, and stock-borrow costs accumulate with turnover.
  • Constraints: share indivisibility, position limits, margin calls, short-sale rules, and liquidity can prevent the desired trade.
  • Contract events: dividends, early exercise, assignment, and adjusted deliverables alter cash flows.

A robust hedge ledger records shares, cash, interest, dividends, fees, option value, and every trade at executable prices. The residual difference from the liability is hedging P/L, not an unexplained plug.

  • “Self-financing means no borrowing.” Borrowing inside the portfolio is allowed; external injections are not.
  • “Rebalancing never changes wealth.” It does not create wealth instantaneously in the frictionless identity, but costs reduce wealth.
  • “Delta neutrality eliminates risk.” Delta is local; Gamma, Vega, jumps, time, and model error remain.
  • “A perfect terminal payoff proves a valid backtest.” The intermediate cash account and financing must also be feasible.
  • “Continuous replication is an executable trading plan.” It is an idealization requiring assumptions real markets do not satisfy exactly.