What exactly is turnover?
Define the formula before comparing methods. You may sum purchases and sales, or use the smaller of the two amounts relative to average portfolio value. These measures differ. External contributions and withdrawals also need separate treatment: buying with new money is not necessarily the same restructuring of the existing composition as selling and buying again.
Suppose the fictional portfolio’s average value for the period is 10,000 conventional units. To restore target weights, 600 of assets are sold and 600 are bought. Gross transaction volume is 1,200, or 12% of average value. The smaller-side measure is 6%. Neither figure should be published under the word “turnover” without a definition.
The invisible part of execution cost
Suppose the combined cost on each side of the transaction is 0.2% of that side’s volume, including the explicitly specified components. Selling and buying 600 each costs 2.4 units. That is 0.024% of 10,000. If the same operation is repeated many times, total cost needs a separate calculation; it cannot be counted only once at year end.
In reality, a broker commission is not the whole cost. Bid–ask spread, the effect of one’s own volume on execution and the inability to trade at an observed average price can change the result. The same tariff does not mean the same total cost for a liquid small order and a difficult large order.
Calendar and threshold are different rules
A calendar check creates a regular organisational moment. A threshold rule reacts to deviations but may trigger frequently when values move around the boundary. Execution frequency depends on threshold width, data and return rules, not only on the method’s name. Distinguish percentage points from relative deviation from the target weight.
Moving 100 units at a total cost of 0.4% costs 0.4 units. A small cost does not by itself prove that the action is useful: risk may barely change. Conversely, avoiding a trade to save money is not always justified when the structure has materially departed from the stated task. Show financial and organisational consequences together rather than declaring a universal winner.
How to compare rules honestly
For a historical test, set rules before analysis, use the same data, external flows, costs and starting conditions. Do not let one method trade at a price unknown when the decision was made. Compare ending value, structural deviation, number of transactions, turnover and drawdowns. A successful result over one interval does not prove future superiority; the SEC distinguishes a hypothetical test from actual performance.[27](#source-market-performance)
Assess the use of new contributions and distributions separately. They may change composition without a sale, but free cash cannot always be assumed. Tax consequences also cannot be inserted universally as one percentage: the account, jurisdiction and specific transactions require review. Unknown costs are better labelled than silently set to zero.
The rule card should state the objective, deviation measure, trigger, size and execution cost, plus reasons for revision. It can test the price of discipline without turning rebalancing into a promise of extra return. Educational arithmetic does not set suitable weights or instruct anyone to buy or sell each time a weight changes.
Keep two versions of the result
For comparison, keep the result before costs and the result after an explicitly specified set of costs. The difference shows what implementation cost in the model. If some costs are unestimated, call the after-cost result partial rather than final. Execution uncertainty matters especially when the method’s advantage is small.
Separate checking from acting. A composition can be observed more often than trades are made; observation itself creates no turnover. If a report combines these events, many observations can be mistaken for high trading activity, or the reverse.
For every transaction, retain the deviation before action, volume, execution condition and residual deviation after costs. This shows whether the rule was implemented rather than merely drawn in a final chart, and helps identify repeated trades caused by oscillation around the same boundary.
Related foundations
[Rebalancing and behaviour: restore the rule, not a market guess](/guides/rebalancing-and-behavior/); [Financial records: prepare data, do not guess tax](/guides/tax-records-basics/).