Advanced Price Forecasting

Follow

SmartShift™️ has to forecast 3 things to create a personalised plan for your battery:

  1. The price for you site
  2. Your fixed load usage - how much power you / your house consumes
  3. Your solar generation

Together, these 3 inputs, with some additional guiding parameters, go into a cost optimisation model that generates your personalised battery plan, with the objective to have the lowest cost (or highest earnings, as earnings are negative cost) possible given all the things it knows at each point in time.

This article will focus primarily on Price Forecasting, but all 3 work together so there is a brief summary of fixed load and solar forecasting at the bottom of this article.

Advanced Price Forecasting

In the past we have relied on price forecasts from AEMO (grey line) to make a plan for your battery system. The forecasts that AEMO generates can differ from the prices that actually materialise for a number of reasons.

We now generate Advanced Price Forecasts (pink dotted line / shaded area) using a number of inputs, such as the AEMO price, weather, demand and supply forecasts, in order to provide a more accurate prediction of what future prices will be.

By using this in the SmartShift™️ algorithm we can better plan for the future, and therefore make smarter decisions for your battery system.

Learn more about the partnership that kicked off this work here.

Why is this better than the previously used AEMO forecasts

The AEMO price forecasts are based on current pre-dispatch price bids.

Forecasting the wholesale price like this is a bit like trying to predict the final price a house on the auction block will sell for based on all the bids so far. The prediction will start off less accurate, and become more accurate over time, as more bids are placed, and more information is available.

Generators will often wait until the last possible minute to put in their most competitive bids. They're all jostling to get the highest possible price for their energy.

For example: generators will often place an extremely high bid early in the day (e.g. equivalent to $15/kWh), expecting to lower the bid over time. Even when the bids for a certain period are virtually guaranteed to become more competitive as the period draws closer, the market operator will forecast a price spike as long as those high bids are still in place.

A key issue with pre-dispatch prices is that they do not account for market dynamics. For example, if the pre-dispatch price is high, it might incentivise generators to come online, increasing supply, which could lead to prices falling. Due to these dynamics, AEMO forecasts regularly show spiky pricing, and periods of high prices that we know are unlikely to occur.

Why do I see different prices in different parts of the app?

Our Advanced Price Forecasts are currently only used for SmartShift™️ - as such the Live prices screen still uses AEMO’s forecasts.

As we work on the changes required to roll out meter native billing (i.e. billing customers with 5 minute meters based on 5 minute prices instead of a half hour average), we will look at ways we might be able to bring our Advanced Price Forecasts to this screen and to all customers.

To make it easier to connect the dots between the live price screen and your SmartShift™️ Plan we have included a line for both the new Advanced Price Forecast (pink) and the AEMO forecast (white/grey) in the SmartShift™️ plan screen chart.

The Buy/Sell price shown in the pink pill in the top section of the Devices tab is set to match the large live price on the Live Prices Screen. This may differ slightly from the current price being used by SmartShift™️ as it uses AEMO forecast prices for future 5 minute intervals in the current half hour period.

If you an API user, the new forecasts are already in there - look for the ‘advancedForecast’ attribute

We are continuing iterations of how new forecast prices show in the app and hope to make incremental improvements soon. Additionally, in late 2024 we expect to be able to bill all customers with 5 minute meters based on actual 5 minute prices and will adjust our app UI accordingly. When this happens the live price shown will be the actual 5 minute price for relevant customers and will therefore match across all systems.


Fixed load forecasting

Our forecast of your household usage is produced by a model that learns of your historical usage. This is updated each week, so as your behaviour changes throughout the year so too will our forecasts (e.g. as you use more power heating in colder months, or less as it warms up going into spring).

Unfortunately SmartShift™️ is not a mind reader or fortune teller, so it can only do it’s best to predict your usage and make a plan of how much to charge/discharge/preserve your battery given the data it has to work with. Sometimes people change their plans and routines, or do something out of the ordinary, and this can affect how well your battery plan performs.

Solar forecasting

Our solar forecasts are based on regional solar forecasts provided by AEMO. These are scaled to the size of your solar system based on the highest solar generation reading we have ever recorded on your site.

Ever wondered why its cloudy outside but we are still forecasting production at your house. Well due to the regional nature of the forecasts, there may still be a lot of sun, just not in your localised area.

We are currently in the process of transitioning to a new solar forecasting method, which will provide more accurate and localised forecasts. This change will roll-out over the next few weeks to all SmartShift™️ customers, so you may notice some changes in the forecasts shown to you in the Amber app during this time. 

Was this article helpful?
25 out of 27 found this helpful

Comments

0 comments

Article is closed for comments.