Net return measures what remains after specified costs. Reconcile the cash ledger, identify the fee base and timing, and avoid charging embedded execution costs twice.
Net return
Net return is the investment gain or loss, relative to the stated capital base, after deducting a specified set of costs. For a systematic strategy, it answers an economic question: how much of the investment result remains after the modelled costs of earning it?
The qualification matters. “Net” alone does not promise that every tax, funding charge, advisory fee or other investor expense has been deducted. Read it alongside the cost boundary, return period and cash-flow convention. The companion gross return must use a compatible boundary for the comparison to mean anything.
Reconcile the wealth before comparing the rate
Consider a synthetic account over one stated holding period. Starting wealth is 100 index units. Investment gains, including all income retained in the account, bring ending wealth to 108 before a flat end-period account charge of 1 unit. Commissions and the effects of spread and market impact are already reflected in that 108. There are no external deposits or withdrawals. Funding, securities borrowing, other advisory charges and taxes are not modelled.
The specified final charge leaves wealth of 108−1=107. The net return is therefore
Rnet=100107−100=7%.
Before this charge, the return is 8%; the charge consumes 1 percentage point of return on starting capital. This is a return for the stated holding period, with no annualisation. It is neither a personal after-tax return nor evidence that omitted real-world costs are zero.
With positive starting wealth W0, pre-charge ending wealth W1, and a separate flat cash cost C paid only at the end, the same calculation is
Rnet=W0W1−C−W0=Rbefore−W0C.
Here W1 includes investment income and any costs already recorded in the account. The formula deducts only the remaining specified charge. A cost is not an external investor withdrawal to remove from performance: removing it would erase the cost drag one is trying to measure.
A fee rate needs a base and a clock
The flat 1-unit charge is not interchangeable with a fee of 1% on pre-charge ending wealth. Under that alternative rule, still paid only at the end, the charge would be 0.01×108=1.08 units. Wealth would finish at 106.92 and net return would be 6.92%.
Two synthetic accounts start at 100 and reach 108 before costs. A flat charge of 1 leaves 107 and 7% net return; a charge of 1% of 108 is 1.08, leaving 106.92 and 6.92%.
Original synthetic fee-base comparison. Both rows use the same starting capital, pre-charge wealth and end-period timing; only the cost rule changes.
For an end-period fee fraction f applied to pre-fee ending wealth, the exact relation is
1+Rnet=(1+Rbefore)(1−f).
Subtracting the fee rate directly from the return misses its different base. Fees charged earlier or repeatedly also change the capital left to earn subsequent returns. Follow their actual timing rather than applying an end-period shortcut. The SEC’s fee bulletin explains why both transaction fees and ongoing expenses reduce investment wealth, including through the capital available to compound.
Charge each economic cost once
A backtest should identify where each transaction cost enters the accounting. Buying at an adverse fill price already spends more cash; selling at an adverse fill price already receives less. Subtracting the same spread or impact estimate again understates net performance. A separate report may attribute the shortfall to execution without making another cash deduction.
Conversely, a marked portfolio value need not contain every bill. Explicit commissions, financing accruals, securities-borrow charges or management fees may require separate entries. “Included in the cost model” is only meaningful when those entries actually reach the reported return.
Financing, Borrow and Transaction Costs develops the distinction between costs embedded in fill prices and separate trading or holding charges. The public preview introduces the lesson; the full treatment requires membership.
Make two net figures comparable
Check the same holding period, currency, valuation basis, capital denominator and income treatment before ranking two reported returns. Net of costs is a different choice from total return, which includes income. If income is paid out rather than retained, include it consistently in the return calculation. If investors contribute or withdraw capital, use a suitable time-weighted return or money-weighted return calculation rather than the simple no-flow wealth ratio above.
Reporting standards can define a specific cost boundary. Under the GIPS Standards for Firms, composite net-of-fees returns deduct transaction costs and investment management fees; disclosure identifies additional deductions and whether actual or model management fees are used. That defined reporting measure is not a guarantee of each investor’s final after-tax experience. See the GIPS handbook, provisions 2.A.38 and 4.C.7.
Net performance can also change which strategy is attractive. Qian, Hua and Sorensen show that allowing for turnover-related transaction costs can shift the preferred alpha model toward less costly trading. Comparing paper returns first and adding a cost disclaimer later can therefore miss the economic decision itself.
Edward E. Qian, Ronald H. Hua and Eric H. Sorensen, Quantitative Equity Portfolio Management: Modern Techniques and Applications (Chapman & Hall/CRC, 2007), Chapter 8, pp. 265–266, Table 8.6 and Figure 8.7: net returns and model choice under stated transaction-cost assumptions.
