Data / Modelling assumptions
Modelling assumptions
Everything the model assumes rather than sources: panel wattage, degradation, loan terms, escalation.
- Records
- 23
- With their own source
- 2
- Carrying a caveat
- 21
- Verified
- 2026-09-06
What this decides
Everything. These are the values a reader is most entitled to disagree with.
Every one of these is adjustable in the calculator or published on the methodology page. Nothing here is a secret, and nothing here is a measurement.
Every record
| Record | Value | Group | Verified | Source |
|---|---|---|---|---|
| panel watts modelling choice | 400 | The system | 2026-09-04 | — |
| sqft per panel modelling choice | 17.5 | The system | 2026-09-04 | — |
| offset target modelling choice | 1 | The system | 2026-09-04 | — |
| degradation per year modelling choice | 0.005 | The system | 2026-09-04 | — |
| O&M cost per year modelling choice | 150 | The system | 2026-09-04 | — |
| inverter replacement cost modelling choice | 1,800 | The system | 2026-09-04 | — |
| inverter replacement year modelling choice | 12 | The system | 2026-09-04 | — |
| analysis years modelling choice | 25 | The system | 2026-09-04 | — |
| min installed kW sourced | 5 | The system | 2026-09-06 | escholarship.org |
| The smallest system a household is realistically sold. LBNL Tracking the Sun (2024 edition, systems installed in 2023) reports a median residential size of 7.4 kW with the 20th-to-80th percentile band at 5-11 kW, so four in five installed systems are at least 5 kW. It is not a hard floor - one in five is smaller - but below it a buyer is outside what most installers routinely quote, and the per-watt price rises because permitting, design and a day of labour do not scale down. The engine applies this only when a caller asks for it (overrides.minSystemKw); it is not imposed on the default sizing. | ||||
| median installed kW modelling choice | 7.4 | The system | 2026-09-04 | — |
| Median US residential system size, LBNL Tracking the Sun 2024 edition, systems installed in 2023. Recorded so the page can quote it rather than restate it. | ||||
| p80 installed kW sourced | 11 | The system | 2026-09-06 | escholarship.org |
The 80th percentile of US residential system size. Same source and edition as the median and the floor: LBNL Tracking the Sun reports the 20th-to-80th percentile band at 5-11 kW for systems installed in 2023, so three in five installed systems sit between min_installed_kw and this value. Recorded as a figure rather than left inside a note so a page can place a size in the real distribution instead of asserting one. | ||||
| fixed monthly fee modelling choice | 12 | Your bill | 2026-09-04 | — |
| Non-bypassable customer charge; solar does not remove it. | ||||
| rate escalation per year modelling choice | 0.025 | The market | 2026-09-04 | — |
| without battery modelling choice | 0.4 | Self-consumption | 2026-09-04 | — |
| Share of generation consumed on site rather than exported. Irrelevant under 1:1 net metering, decisive under NEM 3.0. | ||||
| with battery modelling choice | 0.75 | Self-consumption | 2026-09-04 | — |
| Share of generation consumed on site rather than exported. Irrelevant under 1:1 net metering, decisive under NEM 3.0. | ||||
| APR modelling choice | 0.07 | Loan | 2026-09-04 | — |
| term years modelling choice | 20 | Loan | 2026-09-04 | — |
| down payment modelling choice | 0 | Loan | 2026-09-04 | — |
| PPA rate vs retail modelling choice | 0.7 | Lease or power purchase agreement | 2026-09-04 | — |
| Third-party owner keeps the 48E credit and prices below the retail rate. Zero down, lower savings. | ||||
| annual escalator modelling choice | 0.029 | Lease or power purchase agreement | 2026-09-04 | — |
| Third-party owner keeps the 48E credit and prices below the retail rate. Zero down, lower savings. | ||||
| usable kWh modelling choice | 13.5 | Battery | 2026-09-04 | — |
| installed cost modelling choice | 12,000 | Battery | 2026-09-04 | — |
| immediate min share of bill modelling choice | 0.1 | How results are worded | 2026-09-04 | — |
| A zero-upfront scenario is called immediate only if its year-1 net cash flow is at least this share of the annual bill. Below it: cash-flow neutral. A one-dollar margin must never read as a win. | ||||
Source column links the page or endpoint the value was read from. Where a row has no link of its own, it inherits the file-level source.
Our own choices, except where a row cites a source of its own.
Take it
- assumptions.json
The file the engine actually reads, unaltered, with the documentation blocks it carries.
- assumptions.csv
The 23 records flattened for a spreadsheet: value, verified date, source, caveats, note.
Our compilation is free to reuse with a link back. The underlying data is not ours to license — follow the source on any row for the authoritative version and its own terms. The full position.
The rest of the data
- Electricity prices
What a residential kilowatt-hour costs, by state. The single biggest driver of whether solar pays.
- Solar yield
How much electricity a kilowatt of panels makes in a year, by place. The only pure physics on the site.
- Installed cost
What a system costs installed, on both published bases: competitive quotes, and what buyers actually paid.
- Export rules
What the utility pays for a kilowatt-hour you send out. Decides payback more than sunshine does.
- State incentives
What a buyer can actually claim from a state, read from the statute or programme that grants it.
- Federal policy
The federal credit as data, versioned by year, with the dated events that changed it.
Published 6 September 2026. Last verified 6 September 2026. A figure here that looks wrong is the mail we most want: tell us, and we publish corrections. The formulas that consume these records are on the methodology page.