Introduction to Saturnalia

Predict price variation in November thanks to scores computed after the harvest
Main features
Our focus moves from macro areas (e.g. Burgundy region) to single crus (e.g. Chambertin)
Thanks to machine learning, the algorithm will continue to improve vintage after vintage 
Price prediction tool at 3 years after release of a panel of about 100 investment grade wines. Check the list HERE
Saturnalia scores can be computed and released soon after the harvest. This will facilitate decisions on the previous vintage.
Increase of transactions thanks to the additional information provided
The Saturnalia algorithm can process several areas, such as Burgundy, Rhone, Barolo, Montalcino, Bolgheri, etc.
Not influenced by taste and personal opinions
Who we are
Exclusive Saturnalia early score estimates of selected investment wines 
Easy integration of maps for the appellations and producers within websites. Check the list HERE
Last available market price directly retrieved from Liv-ex 
Continuous monitoring of vineyards using satellite and weather data
Easy integration of maps and data in websites 
Exclusive and detailed early harvest reports for the main regions and communes relevant for wine investment 
Bred in an academic laboratory boasting more than 20 years of research and development in geospatial and Earth observation data processing and analysis, a spin-off company from the University of Pavia was founded in year 2014 under the name of Ticinum Aerospace.
The team is composed of highly educated professionals, characterized by different expertise, and driven by their ambition to develop global-scale, game-changing geospatial solutions.
Saturnalia is a European award-winning service, ingesting big geospatial data collected from both satellites and weather stations, with the final aim to unlock insights about any single vineyards worldwide.
Our starting point is data
Keep customers updated on how was the vintage in different key locations and on recent market prices


SVI map
Latest score
Evolution during season
Last Available Price
The image on the right is the map of the Saturnalia Vigour Index. It is a measure summarizing vigour of the vineyards. On the right, the video shows the evolution of the vineyards in time described by our modified vegetation index; the video is particularly helpful to highlight differences among the plots.
Fair Price
The figure above represents temperatures and precipitations collected by the weather station in Merignac during the 2019 growing season. Below, the post-processed results of the GPM measurements over the Bordeaux area. In particular, they represent the total of precipitation (measured in millimeters) in 4 different key periods: from November 2018 unitl March 2019, April and May 2019, June to mid-August 2019 and August until mid-October. This kind of processing helps in better undestanding the precipitation distribution.
Saturnalia estimates wine scores by using a proprietary algorithm considering satellite imagery, climate and terrain data. Using this unique feature it is possible to estimate the most recent vintage weeks after the harvest.
Saturnalia, at the moment of writing, is the only provider of the 2019 score due to the cancellation of the En Primeur event.
The fair price is our estimate of the release price according to the given score and previous vintages.

This is a comparison of the predicted prices (computed by our model, used for the monthly price prediction) and the actual price. The x-axis reports the vintage (e.g. V2009) and the year when our price prediction stops (3 years after its En Primeur). For example, for vintage 2015 our prediction stopped in June 2019. The y-axis refers to the percentage of variation from the release price.
A monthly update -included in the subscription- will report the expected trend for the following month.
The charts below report the last prices by Liv-ex for vintages from 2007 to 2018 for a 12 bottles case and the release prices (ex-London). The graph on the right shows the difference between last price and release price for each vintage.
SVI distribution across vintages gives an idea of the relative quality that is to be expected. The wider the distribution, the more heterogeneous are the plots; on the contrary, a narrow distribution represents a homogeneous behaviour across the property. 2019 is centered towards a lower vigour - clue of good quality - but also shows a wider distribution in respect of previous vintages.


Comparison of the Saturnalia Vigour Index over the last vintages.
The vigour is computed as the result of the continuous monitoring of vineyards during the growing season. The greener is the colour, the higher the estimated vine vigour. 
The Saturnalia score is computed soon after the harvest by our proprietary algorithm. It automatically combines data acquired from satellite and from ground. The final score can be tuned according to different score providers as input. The values currently displayed refer to the Liv-ex benchmark (Robert Parker until 2013, Neal Martin after). 
The chart on the left shows the Saturnalia scores from 2015 to 2019: the latter is available exclusively from Saturnalia as no En Primeur event hasn't been held yet. On the right, a comparison of Saturnalia scores and prices is shown. 
These measures refer to the weather measured during the growing season compared with the previous seasons. On the left the amount of rain measured in different periods; we took account of the amount of rain during dormancy starting from November of the previous year.
The bottom-left refers to the Growing Degree Days instead. A summary of these figures is also presented in the chart below.

By working on the behaviour of prices in time, we designed a model able to explain 60% of the variation of prices for wines up to 3 years from their En Primeur. Until now, the model works on a panel of 100 Bordeaux wines. It takes into account the effect of new vintages on already available ones. 
We update our prediction every month taking into account the last market price release by Liv-ex. A trend symbol is provided describing the expected behaviour in 5 different ranges from very positive to very negative.
The table below shows a sample of the wines available. The analysis is limited to 4 vintages as the model works best in the short-medium term.