Welcome to the Dairy Analytics GDT Forecaster. The purpose of this page is to increase transparency around the quantity forecasts that Fonterra issues prior to GDT auctions, and link those to their impact on price movements in the GDT auction. We're also trying to forecast prices, but for now we're still developing our models so take these forecasts with a grain of salt (especially the ones more than a month or two out!).

This page is new and a bit rough around the edges. If you love it/hate it/don't understand anything send an email with your thoughts/criticisms to oscar@dairyanalytics.co.nz and we'll try to improve things.
We'd like to thank AgriHQ for sharing the historic forecast data prior to 7 March 2016. They have been collecting this data and making it available to subscribers of the AgriHQ Pulse since November 2011. If you want more data and analysis, go check out their website (they also have a nifty Milk Price Forecast Calculator) and subscribe to AgriHQ Pulse.
  • info_outlineHelp! How do I read these graphs?

    There's a lot going on but hopefully it's not too hard to comprehend! We're going to step through slowly.

    The top graph is the average winning price of the product in each GDT auction over time. The Actual Price shows what actually happened in the past. The shaded band of Price Forecast is where we think there is an 8 in 10 chance of the future price ending up in over time.

    The middle graph is the total quantity of the GDT product sold (or forecasted) across all contract periods, over time. The Actual Quantity shows what actually happened in the past. The Short-term Quantity Forecast is based on Fonterra's latest quantity estimates for the next four GDT auctions. There is a 8 in 10 chance that the actual quantity offered by Fonterra will be within the shaded band. The Long-term Quantity Forecast is based on the long run average quantity sold by Fonterra. There is a 8 in 10 chance that the actual quantity offered by Fonterra will be within the shaded band. In the future we're going to try and improve this forecast. You shouldn't rely on it for anything more than a rough guide.

    The bottom graph is the difference between the total quantity offered in the auction compared to the quantity forecasted by Fonterra prior to the auction, over time. If the value is greater than 0, it means that Fonterra sold more product than they said they would prior to the auction. If the value is smaller than 0, it means that Fonterra sold less product than they said they would prior to the auction.

    The 1 Auction Prior line relates to the forecasts that Fonterra issued one auction prior to the actual event. For instance, "1 Auction Prior" to Trading Event 183 (7 March 2017) was the quantity forecasted by Fonterra for Trading Event 183 when they released the actual quantities for Trading Event 182 (i.e. 16 February 2017). You can see the actual quantities they forecasted on the middle graph as well as the error between the forecat and actual quantities on the bottom graph. The same is true for 2 Auction Prior, 3 Auction Prior, and 4 Auction Prior.

Whole Milk Powder

We're trying to forecast the average winning price of Whole Milk Powder, as well as the total quantity offered by Fonterra, for each GDT auction over the next 12 months.


Forecast for GDT 187 - 2 May 2017

$ /90   $ /50   $ /10

Historically, Fonterra have been pretty consistent with their forecasts for WMP one auction prior (ignoring pre-2013 when Fonterra were still getting to grips with the forecast process). The most significant abberation occured in the second auction of August 2015, when Fonterra removed 9050 tonnes from the market at late notice.

Skim Milk Powder

We're trying to forecast the average winning price of Skim Milk Powder, as well as the total quantity offered by Fonterra, for each GDT auction over the next 12 months.


Forecast for GDT 187 - 2 May 2017

$ /90   $ /50   $ /10

Historically, Fonterra have been pretty consistent with their forecasts for SMP one auction prior (ignoring pre-2013 when Fonterra were still getting to grips with the forecast process). The worst abberations were in March 2014, and March 2017, when Fonterra added 1900 and 1200 tonnes to the auction respectively.