Two common turnover numbers describe the same trades
Portfolio turnover is incomplete without its convention. In a fully invested long-only rebalance without external cash flows, the sum of absolute weight changes measures buys plus sells. Half that sum measures one-way replacement. Both can be useful, but multiplying a cost quoted per dollar traded by the smaller number understates the charge unless the cost convention already includes both sides.
Begin with drifted weights immediately before trading, not the targets from the previous rebalance. Price changes alter the amount that must be bought or sold. Define the net asset value used to convert weights into dollars and record whether cash, derivatives and shorts are included. Comparisons across studies are misleading when identical labels use different exposure or capital denominators.
Reconcile a three-name rebalance
Assume a fictional $10 million long-only portfolio has pre-trade weights of 40%, 30% and 30%. New targets are 30%, 50% and 20%. The trades are a $1 million sale, a $2 million purchase and a $1 million sale. Absolute traded notional is $4 million, giving 40% two-way turnover and 20% one-way replacement.
If every traded dollar incurs an illustrative all-in one-side cost of 15 basis points, the charge is $4 million multiplied by 0.0015, or $6,000. Relative to $10 million of capital, that is six basis points. Multiplying 20% by 15 basis points would produce only three basis points and count half the trading activity.
Keep the annual aggregation honest
Twelve identical monthly rebalances would sum to 480% two-way turnover. With constant capital and the same assumed rate, their direct charges sum to 72 basis points of that capital. This is a simplified cost-budget calculation, not an exact annual return adjustment: actual capital, trade sizes and costs change, and costs affect subsequent compounding.
External subscriptions and withdrawals complicate the half-sum identity because buys and sells need not balance. A portfolio initially funded from cash also differs from replacing existing holdings. For long-short books, report traded notional and the denominator directly; a net-zero allocation cannot serve as the capital base. Include mandatory corporate-action trades consistently rather than omitting inconvenient activity from the turnover figure.
Explain where trading comes from
Decompose turnover into entries and exits, weight changes among retained names, cash flows and exceptional events. This shows whether a buffer, a wider rebalance interval or a different weighting rule addresses the actual cause. A stable symbol list may still trade heavily. Conversely, an index membership change may involve little capital if the affected position is small.
Frazzini, Israel and Moskowitz's trading-cost research examines how costs vary with trade and security characteristics; a constant rate is therefore an assumption to stress, not a universal market fact. Use higher rates for adverse scenarios and compare the net effect with any claimed signal benefit. The numbers here are invented inputs illustrating accounting conventions, not observed strategy turnover or a validated net return.
Inspect the example
| Fictional holding | Pre-trade weight | Target weight | Trade on $10m NAV |
|---|---|---|---|
| U | 40% | 30% | Sell $1m |
| V | 30% | 50% | Buy $2m |
| W | 30% | 20% | Sell $1m |
| Total traded | Two-way: 40% | One-way: 20% | $4m |
| Assumed cost | 15 bps per traded dollar | 6 bps of NAV | $6,000 |
Sources & reproduction
Primary references support the definitions and method. Numerical examples are illustrative unless a source record is explicitly identified.