Here’s a detailed yet simple example @dregnier
TWR calculation
Basically, the concept of TWR resembles calculating the performance of a stock (or fund / portfolio) based on the market-price of a single share.
(i.e. it ignores the total volume of “assets under management” of the fund/portfolio, but highlights the mix of its components)
As highlighted here,
The slight differences … in the beginning is caused by the fees of the purchase and the slightly different buying price [used for the share] regarding the close price of the day [used for the benchmark].
If calculated on day-basis over a reporting period of months or years, such “slight differences” should be negligible.
The most significant divergence occurs when a dividend is paid on December 15, 2022. The performance of the actual Share-1 position benefits from this additional outbound transfer, resulting in a significant increase in the daily performance of around 9%.
If the dividend payout is modeled to happen “overnight”, i.e. reduces the share value at start of the period, this will push that period’s performance. (implementation details here: adding the dividend outflow to the subsequent MVE, basically makes it an “overnight” outflow)
If the same payout was modeled to happen during that same period, it would decrease the period’s performance. Which would distort the goal of analysis in TWR.
At the moment I’m not fully clear what “cumulative , true TWR” means in PP as opposed to TWR as linked above.
-
concerning “cumulative”: as to my understanding by the TWR concept it is meant to be a single measure for a compound portfolio across several periods. i.e. the measure is compound / cumulative by nature.
-
concerning “True”: it’s meant to say that PP’s day-based calculation avoids approximation margins found in month- or quarter-based calculations. (see here)