Choosing a Denominator for Daily Portfolio Returns
Summary
The document compares two ways to scale daily trading profit and loss: dividing by the previous day’s gross portfolio value or by the account’s initial equity. These choices describe different things. The prior-day value expresses each day’s gain relative to the capital then in the portfolio, which helps assess return consistency as the account grows. Initial equity instead expresses daily profit and loss relative to the starting investment, making it useful for tracking how that original amount changes over time.
The discussion is conceptual and does not establish that either method is universally correct. It notes that the appropriate denominator depends on the performance question, and that a fixed initial-equity denominator can make daily returns appear to rise as the portfolio grows. A second response suggests checking the software package documentation to confirm what its function means by initial account equity. The exchange provides no numerical comparison, formal return series, or detailed guidance on annualization, compounding, or risk-adjusted statistics.
Key ideas
- A prior-day portfolio value denominator measures daily profit relative to the capital held at the start of that day.
- An initial-equity denominator relates daily profit and loss to the original account size.
- The denominator should match whether the goal is to assess return consistency or growth from the initial investment.
- Software functions may define account equity differently, so their documentation matters.
Tags
Full text
# Ways to calculate daily returns # Ways to calculate daily returns A complete rookie here. I'm currently reading Ernie Chan's 'Algorithmic Trading' and trying to recreate his results with quantstrat in R. Everything seems to be fine except for portfolio return calculation part. In his book EC uses the following formula for daily returns: `daily_return = net_daily_p&l / yesterdays_gross_portfolio_value` while PortfReturns function in R definitely returns something like this: `daily_return = net_daily_p&l / initial_account_equity` which makes impressive EC's plots and Sharpe-values not so impressive at all. Are these different methodologies for return calculation or am I missing something important? ## Answer by Tanmay (score 0, accepted) https://quant.stackexchange.com/a/37689 Yes, they're certainly different ways of describing performance. Both may be better suited depending on your requirements. If I had to illustrate a portfolio's consistency-of-return aspect, I would use `yesterdays_gross_portfolio_value`. The main reason is that as the portfolio grows and so might the absolute daily return values and if you used `initial_account_equity` as the denominator, you'd end up getting an increasing function, wherein the returns are only bigger because of more capital that is "invested", not because of actual alpha. On the other hand, if you were only concerned with showing how your initial investment grew over a period of time rather than how it got there, you may want to use `initial_account_equity`. ## Answer by Giladbi (score 1) https://quant.stackexchange.com/a/37675 the initial_account_equity might refer to the base capital amount. in this case, it will not be "yesterdays" value... you need to check the packs help docs.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.