Quick Summary (TL;DR)
Start with Forecast.Solar if you want a solar-production prediction inside Home Assistant with minimal setup. Consider Solcast if you need more detailed time-series data, two differently oriented rooftop arrays, and a choice of forecast scenarios for battery or EV charging decisions. Neither service measures electricity generated by your inverter. Both use remote forecasting data, and neither is proof that a costly battery will pay for itself.
The free tiers are different in ways that matter. Home Assistant’s built-in Forecast.Solar integration supports a single panel plane on its public service with hourly-resolution forecasts and hourly polling; paid access can add finer resolution, more planes and a longer horizon. Solcast’s residential hobbyist programme currently supports up to two rooftop sites at one location, 30-minute periods and a maximum of ten API requests per UTC day for new accounts. Home Assistant users typically connect Solcast through a third-party HACS custom integration, not an official core integration. Check current terms before building a control strategy around either service.
Forecast.Solar vs Solcast: the practical differences
Solar forecasting is useful when a home has flexible loads: a heat-pump water heater, immersion diverter, dishwasher, electric vehicle or battery that can be charged at a more sensible time. What you need is not necessarily a pretty graph. You need a forecast in the correct units, for the correct period, updated often enough and reliable enough for the decision.
| Feature | Forecast.Solar in Home Assistant | Solcast residential plus HACS integration | Why it matters |
|---|---|---|---|
| Integration route | Official, built into Home Assistant | Community-maintained Solcast PV Forecast custom integration, commonly installed through HACS | Installation and maintenance burden |
| No-cost option | Public API without an account or API key | Free registered residential hobbyist account, subject to eligibility and terms | Setup friction and access |
| Free plan roof geometry | One panel plane | Up to two rooftop sites at one residential location | East/west arrays may need separate modelling |
| Free forecast periods | Hourly native resolution | 30-minute periods | Fine-grained scheduling |
| Typical Home Assistant updates | Free service polled hourly; API-key accounts every 30 minutes | Limited by allocated daily API calls and integration polling configuration | How quickly new weather information reaches controls |
| Free API requests | Public service has provider-side limits; avoid aggressive manual polling | Currently up to ten per UTC day for new residential accounts | Prevent exhausted quotas |
| Forecast horizon | Public plan covers today and the following day | Hobbyist rooftop forecast can extend up to fourteen days | Day-ahead versus longer-term planning |
| Uncertainty scenarios | Primarily estimated production; tuning via damping and inverter cap | Estimated and alternative scenario data, exposed according to integration version | Conservative decisions during changeable weather |
| Energy dashboard | Supported by official integration | Supported by community integration | Forecast overlays and comparison |
| Offline forecasting | No; external service needed for new forecasts | No; external service needed for new forecasts | Cloud dependence |
Important: these are the published service conditions checked in October 2026, not permanent promises of price, quota, support or performance. Solcast’s site and API documentation distinguish hobbyist rooftop endpoints from commercial products; do not assume a commercial trial and a residential account have identical limits.
What a PV forecast actually predicts
A solar-production forecast is an estimate of electrical output from a modelled photovoltaic system. It normally combines your location and array orientation with weather inputs and assumptions about the modules and conversion losses. Some services incorporate satellite-derived irradiance and cloud data; the actual methodology and model updates are provider-specific.
Four quantities must not be mixed up:
- Instantaneous power (W or kW): the predicted production rate at a particular time, or average rate across an interval if documented that way.
- Energy in a period (Wh or kWh): the predicted amount generated during an hour or a half-hour. Energy is power integrated over time, not an instantaneous reading.
- Cumulative energy today (kWh): the modelled output for the whole day; do not add this value to period-by-period predictions.
- Remaining energy today (kWh): output expected from now until the end of the forecast day. This is especially useful for decisions at lunchtime.
Suppose a forecast predicts an average of 2.4 kW between 12:00 and 12:30. That period contributes approximately 1.2 kWh of energy, because 2.4 kW × 0.5 hours = 1.2 kWh. Summing the 30-minute average power numbers directly as though each were an hourly kWh reading would double the estimated daily energy.
A forecast of 18 kWh for tomorrow also does not mean that 18 kWh will be available for a battery. Household loads consume part of the production; battery round-trip losses, inverter limits, grid export rules, state of charge and generator behaviour affect what remains.
Choosing a service by your intended automation
| Your goal | Sensible starting point | What to verify before automating |
|---|---|---|
| See tomorrow’s predicted PV output in Home Assistant | Forecast.Solar public integration | Correct roof orientation, array size, inverter cap |
| Schedule a dishwasher in the sunniest couple of hours | Either; Forecast.Solar is simpler | Predicted interval energy exceeds expected appliance demand with a margin |
| Model an east/west roof as separate arrays | Solcast hobbyist’s two sites, or a qualifying paid Forecast.Solar plan | Separate tilt, azimuth and capacity inputs |
| Prepare a home battery for a cloudy tomorrow | Either, but prefer conservative controls and measured state of charge | Load forecast, tariffs, backup reserve and forecast uncertainty |
| Schedule EV charging from afternoon excess solar | Either with real-time grid export as a final condition | EVSE minimum charging current and vehicle needs |
| Detect short cloud passages and respond in seconds | Neither is sufficient on its own | Live inverter or grid metering and appropriately designed control |
| Compare forecast models over a month | Run both without double-counting sensors | Store issued forecasts and actual yield for comparable time windows |
A detailed forecast is not the same as a better one for your roof. Trees, chimney shadows, local sea fog, snow, module soiling and clipping can dominate short-period errors. Test both services with actual data before subscribing or allowing an automation to take expensive action.
Set up Forecast.Solar: the lowest-friction route
The official Home Assistant Forecast.Solar documentation describes a UI-based setup. You do not need an API token to start with its public service.
- Open Settings → Devices & services → Add integration and search for Forecast.Solar.
- Enter the panel location. Home Assistant proposes your home coordinates; change them if the array is installed somewhere else.
- Enter declination (panel tilt from horizontal, in degrees) and azimuth (compass facing: 0° north, 90° east, 180° south, 270° west). A south-facing array is 180°, not 0°.
- Enter total Watt-peak (Wp) for the panels in that one orientation group. Sixteen 425 Wp modules are 6,800 Wp, not 6.8 Wp.
- Complete the integration, let the forecast sensors populate, and compare a few clear days against the inverter’s actual generation.
- Under Settings → Dashboards → Energy, choose the existing solar production entry and enable its production forecast using Forecast.Solar. Keep the inverter’s real cumulative-kWh sensor as the source of actual solar generation.
- In the integration’s options, adjust inverter size if it is lower than the DC array rating, and use morning/evening damping only if your real measurements show systematic shading error.
For a roof with two separate azimuths, do not enter the combined east/west capacity with a single south-facing azimuth and expect an accurate output curve. On the public plan, consider which plane is most useful or use a suitable service tier that explicitly supports multiple planes. Home Assistant documentation says Personal Plus and above can support up to four planes. Different panel groups only belong in the same plane when orientation is sufficiently similar.
Forecast.Solar sensors worth using
| Sensor type | Practical meaning | Appropriate use |
|---|---|---|
| Estimated energy production – today | Forecast total for the present day | Dashboard overview |
| Estimated energy production – remaining today | Energy still expected this day | Battery and flexible-load decisions later in the day |
| Estimated energy production – tomorrow | Next day’s estimated generation | Day-ahead tariff or charging plan |
| Estimated energy production – this hour | Production during the current hour | Hourly scheduling, not instantaneous grid-export control |
| Estimated energy production – next hour | Output expected in the following hour | Delaying a load briefly |
| Estimated power production – now | Modelled present power | Informational only; not a replacement for a live inverter sensor |
| Highest power peak time – tomorrow | Approximate forecast time of peak power | A possible scheduling hint, not guaranteed sunny conditions |
Your exact entity IDs depend on how Home Assistant names the installation. Select the entities from the UI rather than pasting assumed IDs from a tutorial. Extra forecast-power sensors can be disabled by default and may need enabling in the integration’s entity list.
Getting the full hourly forecast for an automation
The official forecast_solar.get_forecast action returns mappings named watts and wh_period. The latter holds watt-hours per period; it is suitable for estimating the energy available in a scheduled window. The documented action supports config_entry, optional time bounds and resolution: hourly or raw.
# Example action sequence for the Home Assistant automation editor.
# Replace the configuration-entry ID with your own.
actions:
- action: forecast_solar.get_forecast
data:
config_entry: YOUR_FORECAST_SOLAR_CONFIG_ENTRY_ID
start: "{{ now() }}"
end: "{{ now() + timedelta(hours=3) }}"
resolution: hourly
response_variable: solar_forecast
- condition: template
value_template: >-
{{ (solar_forecast.wh_period.values() | sum) >= 2500 }}
- action: notify.notify
data:
message: "Forecast suggests at least 2.5 kWh in the next three hours."
This example notifies rather than starting a mains load. That is intentional: a battery, EV or heater deserves a second decision based on actual grid power and device state. The returned values are production estimates, not the guaranteed surplus. The sample checks the forecast as supplied; for strict boundaries inspect the action’s output timestamps and available periods, particularly near sunrise, sunset and daylight-saving changes. See Home Assistant’s Get forecast action reference.
Set up Solcast with Home Assistant
Solcast offers a free, personal non-commercial residential tier for people modelling their own home PV system. It requires registration, rooftop-site configuration and acceptance of the service terms. Accounts may require residential-location verification. The published hobbyist limit for new accounts is ten API requests per UTC day, with up to two rooftop sites within one kilometre of each other, forecast data in 30-minute periods and a horizon of up to fourteen days.
- Register on Solcast’s home PV programme, selecting the residential home-system option rather than the commercial API product.
- Create one rooftop site for each genuinely separate PV plane, up to the allowed limit. Enter the relevant installed capacity, tilt, azimuth and other requested configuration accurately.
- Open the Solcast Toolkit and copy the API key. Keep it private; do not expose it in blog snippets, frontend dashboards, screenshots or a public Git repository.
- Read the current documentation for the community-maintained Solcast PV Forecast integration. Install the repository through HACS if compatible with your Home Assistant version, then restart Home Assistant if prompted.
- Add Solcast PV Forecast through Settings → Devices & services. Supply the API key and correct quota. The free residential limit described here is ten calls per UTC day for new accounts, although older accounts may have different allowances.
- Choose the integration’s automatic update option or configure its documented forecast-update action. Don’t stack manual and automatic polling without counting both against the same budget.
- Verify that its forecast entities populate and then select the Solcast forecast in the Home Assistant Energy dashboard if supported by the installed integration version.
Solcast azimuth requires particular care. The community integration documentation describes an east/west signed convention (negative for east, positive for west), which differs from Forecast.Solar’s 0–360° compass convention. Check the Toolkit’s own orientation hints; do not simply copy the integer 180 or 90 from another service. Also distinguish the API key, which authenticates requests, from each rooftop site’s resource ID, which identifies the modelled installation.
Solcast is not a built-in Home Assistant integration
The widely used Solcast PV Forecast connector is a custom component, maintained independently of Home Assistant core and independently of Solcast. It supports rich forecast entities and an Energy dashboard view, but its installation, breakage risk and update cadence need managing like any third-party integration. Follow its current README and release notes rather than an old blog post about a different maintainer or an obsolete HACS repository. If your priority is minimal maintenance, this is a real advantage for the built-in Forecast.Solar option.
Budget the API calls explicitly
A quota of ten calls per UTC day does not mean that updating every ten minutes is safe. A day has 144 ten-minute periods. If you plan eight weather-refresh calls, leave some budget for configuration, retries or occasional troubleshooting. On systems with two rooftop sites, verify how the particular client and API count requests. The community connector supplies quota diagnostics; its update controls should be configured for your actual account allowance.
| Polling schedule (illustration) | Forecast calls per day | Consequence for a ten-call cap |
|---|---|---|
| Every 15 minutes throughout the day | 96 | Exceeds free residential quota |
| Every hour, 24 hours | 24 | Exceeds quota |
| Every two hours, 24 hours | 12 | Usually exceeds quota |
| Eight deliberate daylight updates | 8 | Leaves limited headroom |
| Four updates: morning, midday, afternoon, evening | 4 | Greater headroom, less frequent refresh |
The figures above assume one chargeable request per scheduled refresh and no additional calls. They are illustrations, not a guarantee of how a particular integration version counts multi-site operations. A quota generally resets at midnight UTC, which need not equal midnight in the local time zone. The Solcast documentation and the integration’s actual API Used sensor are the sources of truth.
Forecast accuracy: measure it before spending money
Marketing claims about accuracy cannot tell you whether your own roof gets value from a particular service. Compare forecasts made at the same decision time against actual AC energy measured by your inverter, across clear, mixed and overcast conditions.
An honest test stores each forecast when issued; looking at today’s final forecast after the weather has already changed is not a fair test of what you could have known yesterday evening. Compare a day-ahead forecast made at 19:00 with tomorrow’s actual output; separately compare forecasts made at 09:00 for the next three hours against the same three-hour actual result. Both services must face the same timestamps and selection rules.
| Day | Forecast.Solar predicted (kWh) | Solcast predicted (kWh) | Inverter actual (kWh) |
|---|---|---|---|
| Monday | 18 | 17 | 16 |
| Tuesday | 8 | 9 | 10 |
| Wednesday | 21 | 23 | 20 |
| Thursday | 12 | 11 | 12 |
| Friday | 15 | 14 | 11 |
| Five-day total | 74 | 74 | 69 |
Illustrative data only — not a measured comparison or performance benchmark.
For Forecast.Solar the absolute daily errors are 2, 2, 1, 0 and 4 kWh: mean absolute error (MAE) = 1.8 kWh/day. For Solcast they are 1, 1, 3, 1 and 3 kWh: MAE = 1.8 kWh/day. The five-day total error is 5 kWh for both, but that’s a poor substitute for interval-by-interval analysis because positive and negative errors can cancel out.
A better buying evaluation is to calculate MAE across 30–60 days, separate clear and cloudy days, and check whether using the forecast actually changes a decision. For example, did it avoid an unnecessary overnight battery charge? Did it delay a hot-water cycle without leaving you short? Did it cause EV charging to miss a departure deadline? The most detailed forecast is not necessarily the most useful.
Common accuracy pitfalls
- Panel azimuth or tilt entered incorrectly: shifts the estimated production curve. Verify the provider’s angle convention independently.
- Array capacity entered in kW rather than Wp: can produce a thousandfold error in Forecast.Solar’s input. A 5.4 kWp array needs 5400 Wp in its relevant field.
- DC nameplate compared with AC inverter output: clipping and conversion losses can make midday predictions high. Configure inverter power limits where supported.
- Battery charging or curtailment: the inverter may produce less electricity than unconstrained solar potential; comparing modelled potential against curtailed AC output can make the weather model look unfairly optimistic.
- Hidden shade and soiling: shading, dirt and seasonal trees alter actual output independently of the cloud forecast. Where offered, damping can address a repeatable pattern; it cannot cure sudden obstruction.
- Incorrect meter boundary: a house grid import sensor is not a solar-generation sensor. A battery’s discharge is not additional PV production.
- Timestamps and daylight saving: an interval beginning at 11:30 local time cannot simply be matched to 11:30 UTC. Use timezone-aware values and the interval start time.
- Unfair revision timing: compare the provider’s predictions as they stood at the same time before the event, not their final retrospective updates.
For more on making inverter or meter data trustworthy, see our guides to ESP32 Modbus energy meters and Home Assistant Energy dashboard troubleshooting.
Forecast scenarios and uncertainty: what Solcast adds
Solcast provides central and alternative PV scenarios at each 30-minute step in its rooftop forecast dataset; the community integration can expose different scenario selections. This can be useful for risk-aware planning. Do not assume a ‘low’ scenario is a guaranteed floor or that a ‘high’ scenario guarantees a specific wattage. The percentile definitions, calibration and field names must be read from the data/API and integration documentation rather than inferred from the labels alone.
For example, a battery controller might use the conservative scenario for the next morning when deciding how much emergency reserve to preserve. A hot-water controller could use a central forecast but wait for live measured export before actually energising a heater. An EV charging controller might use the forecast to plan the charging window while maintaining a firm minimum charge deadline independent of sunshine.
Forecast.Solar’s simpler view is still enough for many households. A reliable ‘remaining today’ sensor plus a conservative threshold may outperform a complicated script nobody maintains. The question is whether the additional scenarios improve your actual operating outcome, not whether they look impressive on a chart.
A worked home-battery and EV-charging decision
Consider a notional home with a 6 kWp solar array, a 10 kWh nominal battery and an EV that must gain 8 kWh before 07:00. At 18:00, the forecast for tomorrow is 16 kWh, while the house normally consumes 11 kWh per day. A simplistic automation might calculate 16 − 11 = 5 kWh surplus and promise that an EV will be charged entirely from solar.
That would be unreliable. The 11 kWh of loads may not occur when the sun shines; overnight heating and cooking may fall outside production hours. Battery charging and discharging waste some energy; inverter and EV charging power may be constrained. Forecast error might be several kWh. Most importantly, tomorrow’s surplus does not guarantee an 07:00 departure requirement if the sun has not risen sufficiently before then.
A sound control hierarchy is:
- Meet the immovable deadline and backup reserve first. Plan grid charging if necessary; don’t gamble vehicle range or critical loads on tomorrow’s predicted irradiance.
- Use the forecast to choose a flexible window, such as whether to heat water at 11:00 or 14:00, not to promise energy that has not arrived.
- Read real grid import/export and inverter/battery state immediately before switching a load. This is where Shelly, a supported Modbus meter or the inverter’s approved meter matters.
- Apply hysteresis and minimum run-time rules. A passing cloud should not cycle a compressor or contactor several times per minute.
- Respect manufacturer control interfaces, export restrictions and electrical protections. Forecast sensors must never bypass certified device protections, BMS limits or utility rules.
That division of responsibility is the central reason to invest in good local measurement before paying for more forecasting granularity.
Can you run Forecast.Solar and Solcast together?
Yes, for comparison. Install Forecast.Solar as the official integration, add the Solcast custom component if appropriate, and expose both forecast series in your own comparison dashboard or recorder. Keep the actual PV production entity tied to your measured inverter output. Do not sum the two forecasts and present the result as a larger solar array. And do not include modelled future kWh in a cumulative energy sensor that Home Assistant treats as a physical meter.
For a fair trial, use matching array capacity and orientation assumptions wherever possible, then compare both forecasts against the same inverter’s AC-energy totals for at least a month. Document any separately modelled east/west planes and clear out obsolete entities before interpreting charts.
Data privacy, resilience and operational limitations
Both forecasting systems depend on remote data. A local Home Assistant installation can keep running its previously calculated automations when disconnected from the internet, but it cannot obtain newly issued weather forecasts if its external provider is unreachable. The Solcast HACS component may cache data; do not assume cached values are fresh without checking the last-successful-poll sensor and timestamps.
Forecast requests expose or derive geographical and PV-system information. Solcast additionally requires an account and API key. Treat keys as secrets, review each service’s privacy notices and use only the commercial/data rights appropriate to your intended application. Solcast’s free hobbyist access is specifically for personal non-commercial use; do not silently republish its forecast feed to a public website, product or hosted multi-user service. Check present terms directly.
A local inverter connection, separate UPS for essential network equipment, and a fail-safe device policy are more important than the cloud’s forecast refresh rate when your electricity or internet fails. No consumer forecast service should be the only source of permission for essential loads or safety-critical operations.
Cost and upgrade considerations
| Spending option | What you get | When it may make sense | Watch for |
|---|---|---|---|
| Forecast.Solar public | Built-in HA integration, one plane, hourly forecast | Simple single-orientation roof, dashboard and non-critical scheduling | Short forecast horizon and slower data updates |
| Paid Forecast.Solar tier | Finer native periods, extra horizon, some plans support multiple planes | East/west roof or forecasts feeding several flexible loads | Select a tier for specific required features |
| Solcast residential hobbyist | 30-minute forecasts and longer horizon at zero fee if eligible | Enthusiasts with two orientations or wanting comparison and scenario data | Ten daily API calls, account and HACS maintenance |
| Commercial forecasting/API product | Rights and quotas for a business workflow | Professional energy service or public/commercial deployment | Contract, attribution, rate limits and data rights |
| Local inverter/grid meter | Measured energy and power, not forecasts | Reliable real-time excess-solar control | Installation cost, compatibility and electrical safety |
Forecast.Solar’s public website displays plan pricing and product tiers, but service fees can change. Solcast hobbyist registration terms and quotas can also change independently of its commercial subscriptions. Recheck the live pricing pages before purchase; the right way to compare cost is over the expected lifetime of the system, including time maintaining HACS, accuracy testing and any extra hardware.
Is a paid forecast financially worthwhile?
A subscription may be economically sensible if it consistently prevents grid imports or improves tariff scheduling; it is rarely justified by ‘more data’ alone. Suppose a change in schedule saves an additional 0.5 kWh/day of paid grid electricity for 200 days a year. That is 100 kWh/year. At an illustrative marginal tariff of £0.25/kWh, the annual gross saving is £25, before accounting for additional operating costs or mistaken predictions. This calculation is not a forecast of savings for your house.
If the proposed subscription and extra maintenance cost more than the measurable benefit, use the simpler system. Use your own import/export tariffs and billing structure; exported solar already has an opportunity cost if your energy supplier pays for it. Automation savings are not necessarily the full retail rate on every redirected kilowatt-hour.
Troubleshooting checklist
| Symptom | Most likely cause to check | Practical response |
|---|---|---|
| Forecast predicts much more than inverter generates on clear days | Wrong DC size, shade, clipping or curtailed inverter | Correct Wp, tilt and azimuth; configure clipping; review actual conditions |
| Forecast peak appears at the wrong time | Wrong orientation sign, time zone or panel grouping | Check both providers’ angle conventions and interval timestamps |
| Forecast.Solar only models one roof side | Public service single-plane limit | Choose a plan supporting multiple planes or use a different model setup |
| Solcast reports 429 or quota exhaustion | Too many update requests | Review API Used and scheduled/forced update calls; wait for UTC reset |
| Solcast custom integration vanished after an update | HACS/custom component compatibility issue | Check current repository release and logs; do not expose credentials in support logs |
| Tomorrow’s forecast sensor shows unavailable | Provider outage, incorrect config or new-site validation | Check provider account status, entity diagnostics and last successful refresh |
| Energy dashboard shows two solar production streams | Forecast mistakenly configured as actual generation | Use the inverter kWh sensor for actuals; choose one forecast display appropriately |
| EV or immersion heater cycles constantly | Forecast used as a live switch condition | Add measured export thresholds, hysteresis and minimum on/off time |
| Numbers differ strongly between providers | Distinct weather models, orientation/capacity or polling times | Align inputs and record forecasts issued at the same timestamp |
| Yesterday’s forecast seems unusually accurate | Comparing revised/recent forecast instead of original day-ahead issue | Store forecast snapshots before each day begins |
Frequently asked questions
Is Forecast.Solar included with Home Assistant?
Yes. Forecast.Solar is an official Home Assistant integration and can be configured through Settings → Devices & services. The public Forecast.Solar service can be used without an account or API key, subject to its single-plane, horizon and resolution limits.
Is Solcast free for homeowners?
Solcast offers free residential hobbyist access for eligible personal non-commercial home PV systems. Its published current terms allow up to two rooftop sites at a home location and ten API requests per UTC day for new accounts. Sign-up, verification and terms apply. Commercial use has different requirements.
Does Solcast have an official native Home Assistant integration?
The Solcast PV Forecast component commonly used for this purpose is maintained by the community and installed through HACS, rather than being the official built-in Forecast.Solar integration. Confirm its current maintainer, releases and compatibility before installing.
Is Solcast more accurate than Forecast.Solar?
Neither provider can be declared universally more accurate without a fair measurement campaign for your roof. Compare similarly timed forecasts with actual production across different weather conditions and judge whether the difference changes a useful automation decision.
Can I use a solar forecast as my electricity meter?
No. A forecast is a model, not a reading from an inverter, revenue meter or approved energy sensor. Use local measurement for import/export accounting, energy billing, live load control and verification. See our ESP32 energy-monitoring methods comparison.
Will the forecasts work if my broadband connection fails?
Previously cached forecasts may remain visible to an integration, but new forecasts need the provider’s remote service. Use live local meter values and robust default settings to keep essential automation safe when the cloud is unavailable.
Can two forecasts be combined for a better prediction?
You can compare them or design a carefully tested ensemble, but simply averaging two vendor numbers does not automatically improve accuracy. Different lead times, roof geometries, clouds and bias can make a naive average worse. Start by recording issued forecasts and testing the error against your own PV output.
Verdict: which should you use?
Choose Forecast.Solar first for a straightforward Home Assistant Energy dashboard and simple household planning. Its official integration, no-key public option, usable estimated-energy sensors and documented forecast action make it an excellent baseline. If one-plane hourly forecasts work for your decision, paying for more complexity is unnecessary.
Try Solcast when the additional capability serves a specific requirement: two differently oriented rooftop sites, 30-minute forecast periods, longer planning horizon or optional scenario values for conservative scheduling. Accept that you are using an account-bound service plus a third-party Home Assistant component, and plan within the residential API limit.
For either service, validate the predicted kWh against your inverter before allowing the forecast to influence charging costs or heating comfort. For responsive excess-solar control, the decisive information still comes from actual grid and battery sensors, not the weather model.
Official documentation and source links
- Home Assistant — Forecast.Solar integration: setup, single-plane free service, sensors, damping, update cadence and Energy dashboard use.
- Home Assistant — Forecast.Solar Get forecast action: action inputs, resolution and
watts/wh_periodresponse format. - Forecast.Solar — service plans and documentation: current plan terms, forecast horizons and available resolutions.
- Solcast — free rooftop solar forecasting: residential programme, use restrictions and API quota.
- Solcast — hobbyist rooftop API reference: resource IDs, forecast endpoints, 30-minute resolution, two-site maximum and forecast horizon.
- Home Assistant Solcast PV Forecast integration — repository and setup: community HACS installation, sensors, quota handling and limitations.
- Home Assistant Solcast PV Forecast FAQ: common API and update-scheduling problems.
Editorial check: October 2026. Feature availability, account limits, integrations and plan prices are subject to change. Any numerical savings and forecast data used above are illustrative, not measurements or product tests.