The workings

How we get our numbers.

The savings figure on our front page isn’t a marketing estimate - it’s the output of a simulation you can read the source code of. Here’s exactly what we ran, and what it found.

~£530/yr
simulated saving at our first member site (solar + EV) - a real £105/mo bill
~£410/yr
simulated saving for a typical home (no solar, no EV) vs a standard tariff
5-9 yrs
payback on ~£3,800 installed (~£2,700 for founding members), depending on your home

How the simulation works

We took a full year of real Octopus half-hourly prices (Agile, July 2025 - July 2026, London region, straight from Octopus’s public API) and replayed it, day by day, through the same software that runs on the Raspberry Pi in a member’s garage. It is one codebase, not a spreadsheet:

Being open about the split: roughly £180 of the ~£410 comes from the tariff switch itself, the rest from the battery. We count both, because half-hourly prices carry real peak-price risk that makes them a gamble for a home without a battery - the battery is what makes the cheap tariff safe to hold (it never buys at the peak, in any year), so the two only work as a pair.

The real-world anchor: our first member site

Simulations need a reality check, so here is our one real datapoint - the York family home (our template install) behind the front-page figures. It has 4kWp of solar and an EV, and its real bill today is £105/month, standing charges and its 2p solar export included. We simulated that same home through the full year of half-hourly prices with the standard kit: the EV smart-charging into each night’s two cheapest slots, the battery grid-charging on the cheapest overnight slots while leaving headroom for the day’s forecast sun, storing ~1,100kWh/yr of solar surplus that earns pennies as export today, and never exporting.

The same home, over a yearper month
Real bill today (flat tariff, solar, EV)£105
Simulated: kit + half-hourly pricing (same standing charge)~£61
Saving~£44

That’s ~£530/yr - about £3,700 of kit paying back in ~7 years. The assumptions we can’t verify until it’s live: standing charges unchanged (~60p/day), ~9,000 EV miles/yr charged at home, and our weather model standing in for real York sunshine. It’s a simulation of a real baseline, not yet a measured result - the install is underway, and we’ll publish the live ledger, penny by penny, as it runs.

The prices, over the whole year

Half-hourly pricing makes no promises - it tracks the wholesale market. But look at the floor it actually delivered. This is the cheapest half-hour of each day across our simulation year, the slots your battery fills up on, against the flat ~25.5p a standard tariff charges around the clock:

cheapest half-hour that day 30-day average standard tariff (25.5p) paid-to-charge day

The cheapest slot averaged ~10p against the 25.5p a standard tariff charges all day - and on 39 days the price went negative: the grid paid the battery to charge. The catch is that none of it is guaranteed: the cheap slots land at a different time each night, and the evening peaks carry real risk in a bad winter. It’s a market only a battery with software can navigate safely - the battery means you never buy at the peak, and finding each day’s floor is exactly the job of the optimiser, which re-plans every morning when the day’s prices publish.

What we’re not claiming

The controller, the optimiser, the simulator and the raw price data are all in our repository - the figure on the front page is kwc simulate’s output, nothing more.