A Breakdown on an account's Performance tab tells you how each group did. "Equity +20.2% over a year" still means little until it sits beside what a plain index did over the same days, so each group can carry a benchmark.
What each group is compared with
- Model slice: the slice as the model defines it, the slice's own funds at their split, using the model's current version for the whole history. It needs no setup. Your actual dollars in the slice against the fund the model says to hold is the cost of drift, timing, substitutions and cash drag in one number. When the slice's only fund is the one you held all along, the two differ only by timing and cash, and that is the honest answer.
- Asset class: the ticker you map to that class (Equity → VTI, say).
- Fund category: the ticker you map to that category. Nothing is suggested here; most categories have no obvious index.
- Cash: the ticker you map for cash (a short-Treasury ETF such as SGOV or BIL), in every breakdown. Its vs row appears only when Cash has a return of its own, that is, when it holds a money-market fund; settlement cash alone has no return to compare (see what TWR leaves out).
- Unclassified and Not in target model have no benchmark.
Measured over the dates the group was held
A group's benchmark covers exactly the span the group's own figure covers, never the window's full span. For a 1Y cell it runs from the window's start if the group was held then, or from the day the group first held something inside the window, to the last day it held anything. A group sold six months ago is compared with six months of index, not a year. A group sold and bought back later is compared over both stretches, and the benchmark earns nothing in between, just as the group did not. A window the group did not hold shows a dash for the benchmark too, and so does one whose start comes before the benchmark's own price history begins (hover the dash and it says so). 3Y and 5Y annualize over the window's years, as the group's own cell does; Since inception runs over the group's first and last held dates, and Annualized spreads that same return over the days the group's own annualized figure covers.
The comparison is fair because both sides are total return: a group's time-weighted return credits a cash dividend as return, and a benchmark is read from adjusted closes, which fold distributions back in. One honest caveat: a dividend your broker pays out of the account, or posts with no fund symbol, leaves the group's return without it while the benchmark keeps it, so that group lags by roughly its yield.
In the table and in the chart
Under each group row that has a benchmark, a muted vs row shows the group's excess return, the group's figure less the benchmark's, in percentage points, signed and coloured like the return cells. The small control above the table swaps every vs row to the benchmark's own return; hovering a cell shows the benchmark's figure and the two dates it was measured between. A benchmark with no stored price history says so instead of showing a figure: price it in Securities & prices or add it to the comparison list, which registers and prices a new symbol.
In TWR % and TWR %/yr, show one group alone (Alt-click its legend entry) and the chart's overlay becomes that group's benchmark, drawn over the same dates the group's line is drawn; the legend's benchmark entry takes its name. Show the other lines again and the chart's own vs choice comes back.
Value mode: the allocation view
In Value mode a breakdown no longer draws crossing dollar lines. It draws each group's share of the account, stacked to 100%, so the account's mix through time is one picture. In Model slice mode each slice's target share is a dashed line at the level the top of its band would sit at on target: the gap between the band's edge and the line is the drift. Hover for each band's share and its dollars; hide a band from the legend and the others close up, keeping their own shares.
Choosing the benchmarks
Benchmarks… beside the Breakdown control opens the mapping: one ticker per asset class, per fund category and for Cash, one list for the whole app. Pick any priced ticker. On first open the editor suggests VTI · BND · VNQ · SGOV for Equity, Fixed Income, Real Assets and Cash when those tickers have price history; one button fills the empty rows, and nothing is saved until you press Save. A model slice needs no entry. A slice fund with no stored history is replayed on its reference ticker the way a backtest is, all or nothing; a fund whose history begins later joins from its next rebalance, and one whose quotes stop early gets no new weight after that. The vs row says so with an asterisk you can hover.