Can vintages be finally scored scientifically?
Ticinum Aerospace, an Italian aerospace start-up, thinks so.
Welcome to In the mood for wine — a weekly newsletter on wine for the next gen of wine lovers and investors.
Today I’ll treat you to a deep dive into science, aerospace & satellites. Yes, you read that right!
If you are new, you can join here to receive straight to your inbox updates about this project, WineLeaks, a curated overview of the wine market, and other in-depth reports.
What does aerospace and wine got to do with one another?
Nothing.
Or so I thought, when Daniele De Vecchi, project manager of an Italian start-up called Ticinum Aerospace, contacted me.
I was puzzled, yet intrigued.
Daniele mentioned that he had read my newsletter, specifically the one where I write about the Barolo 2018 vintage, and that his company developed a platform (Saturnalia) that could help me. We planned a call for that following week. We chatted for longer than the scheduled 30 minutes and then we chatted three more times after that.
Once I better understood their models and the technology behind them, I was so excited that I talked about them to every single one of my friends. Alas, as Alder Yarrow quite poignantly reminded us in his article Pix – the inside story for jancisrobinson.com, the wine industry isn’t always open or forthcoming to change and new technology.
“Wine isn’t keeping up with the digital technology revolution, and it hasn’t been for a long while. I’m not talking about the surprising resistance of some wineries to invest the time and energy required to engage customers on social media. I’m talking about a lack of the deep-seated digital transformations that have occupied most other industries for the last decade.”
Nevertheless, or perhaps because of it, I decided that it was crucial to spread the word about Ticinum Aerospace’s technology — already widely adopted by the insurance sector.
The company
Ticinum Aerospace is a start-up based in Pavia (Italy), and their wine platform Saturnalia was conceived in 2016 on the back of the Space Camp App, an accelerator program organised by the European Space Agency (ESA).
Their idea?
Leveraging the latest advances in machine learning to make sense of the so-called Earth Observation technology gathered by ESA & NASA satellites. Earth Observation technology, through the use of satellites that carry imaging devices, is used to monitor and assess the status of, and changes in, the natural and manmade environment.
Ticinum Aerospace already successfully employs this technology to support insurance companies, by enabling them to rapidly assess risk of extreme weather events or other events such as human conflict.
And because of this success, the European Space Agency has continued to support Saturnalia with two grants, the latest one in 2020 for the development of a new platform, Saturnalia, this time solely designed for fine wines.
And here we are, wine and aerospace.
The technology
“Our problem“ said Daniele during one of our calls, “is that I don’t think we are explaining our technology very well.“
Indeed, I found myself scrambling through papers and papers worth of scientific jargon trying to piece together the information, sometimes puzzled, sometimes resigned.
And then, as I sat down to write about it for this piece, I did have the feeling that if I made it too simple it would sound like some mad conspiracy theory. So I’ll just explain it as I understand it hoping it won’t.
The Earth Observation satellite images used by Saturnalia capture photographs of the Earth with more information than standard photographs (in chart below, ‘Visible’), such as the presence of absorbed water by the vines (in chart below, ‘Near-Infrared’ and ‘Shortwave Infrared’).
Thanks to this data, the team behind Saturnalia have developed a suite of tools and mathematical models to scientifically analyse vintages, down to single vineyards and crops, with the ultimate goal of assessing the quality of a vintage and a wine.
How?
Using a set of tools available on their platform that, when considered altogether, give a picture of the wine and its journey from the vine (with maps, climatic data and their proprietary SVI and SEI indices) to its market price dynamics (fair price analysis and comparables).
Saturnalia and I agreed on a project whereby I’ll explore what their technology can do for wine investors when picking fine wines. Going forward, I’ll use their models and data when assessing wines in my WineLeaks or other pieces and below you can find the ones that I find most useful.
🔵 Climate Data
If you have read the latest edition of WineLeaks — The Nebbiolo Edition, you’ll be familiar with some of the data here.
In my pamphlet in favour of the 2013 vintage in Barolo, I use data such as the Precipitation, the Diurnal Variation and the Growing Degree Days to compare 2013 to exceptional vintages such as 2016.
This climate data offers a holistic view of vintages in a very specific area for specific producers, while at the same time contextualising it with average 10-year data.
And such data can be compared with other vintages or wines (you can see an example below in “Comparables“).
Why is it useful?
When assessing a wine or a vintage, Saturnalia climate data offers a very localised overview of the plots that go into making the wine, down to precipitation, land temperature and diurnal variation.
🔵 The Maps
The maps are at the essence of what Saturnalia does. While other companies can only offer static drone views, the data offered here is very detailed and, more importantly, constantly updated from the satellite into the platform.
Why is it useful?
First and foremost, I use it as a visual aide for my studies to understand and visualise wine regions to individual crus.
In addition, I find invaluable details of vineyards and their ‘vicinity bonus‘ information for property value. For example, by observing the map above, it becomes clear that the vineyeards of Château Beausejour Duffau Lagarosse in St Emilion (Premier Grand Cru Classé B) are intertwined with those of Château Angélus, making the former an interesting château to watch (given the much lower price).
🔵 Indices (SEI & SVI)
Here is where it gets a bit complicated.
In short, combining data from the satellite images with machine learning, Saturnalia has developed two mathematical indices which monitor the quantity of chlorophyll, mesophyll, and the amount of water in the vine as a distribution (SVI) and through a time series for a specific vintage (SEI), with the ultimate goal of summarising the data from the satellites.
The goal is to evaluate a vintage as impartially as possible, translating the weather data and the images received from the satellites into a score.
The visual satellite images such as the one above is then translated into a distribution of values such as the two below from Lafite’s 2013 (one of the worst vintages in the last decade) and 2016 (one of the best).
The SVI index, as exemplified below, shows that:
better vintages are centred around lower SVI values (or ‘positively skewed’), and
better vintages are more narrowly distributed around a single point.
The second index, the Saturnalia Evolution Index (SEI), charts the yearly time series evolution of the SVI, and shows how “green“ a vine is throughout the year. With that information we can deduce whether it has suffered from water stress or conversely, if rain has diluted flavours. More importantly, one can visually compare the evolution of the vines across vintages.
The summer of 2021 in St Estephe, for example, was characterised by heavy rains and conversely, the previous year 2020 was a particularly dry one.
Why are they useful?
Although they may not feel very intuitive at first, these two indices make for detailed vintage analysis. Crucially, it become easy to compare one vintage with another one or with the average to understand:
if / when there was too little or too much rain. The timing of rainfall, is crucial when analysing the chart: spring rain is welcomed — not so much just ahead of harvest.
effects of spring frosts and hail stones.
how homogeneus a vintage was.
🔵 Fair Price
Saturnalia has also developed a few indicators to help understand the price dynamics of specific wines. In addition to Release price and Liv-ex latest price, one can compare it to an “Estimated Fair Price”.
The model behind it is a statistical regression model that, at stated level of confidence (see model confidence below), calculates a “Fair Price” by aggregating critics’ scores and the price performance of previous vintages.
The Fair Price model also considers the “Price per Point”. It explains which brands command a stronger market price ‘for each point’ of wine critics — a concept similar to that of ‘price per square meter’ in real estate.
Angélus commands a unit price much higher than both Canon and Figeac (see table below). Perhaps the gap between Angélus and newly promoted Figeac will narrow? If this thesis is correct, Figeac 2019 is currently very undervalued.
Why is it useful?
While the model doesn’t consider the market and momentum or mean reversion, it can help identify bargains based on critics’ scores and undevalued brands.
🔵 Comparables
To find pockets of value, another useful tool for wine collectors is the ability to compare two (or more) producers, two (or more) wines or two (or more) vintages side by side.
Most data is summarised in this table below — from the release price to the composite average score to the climate data and fair price estimate.
For example, if analysed carefully, the table below makes the case for investing in ‘off’ vintages as the châteaux and négociants seem to be coming out with a higher release price in ‘on’ vintages, eating into the profits of secondary market investors.
Why is it useful?
A snapshot, comparing a wine with ‘benchmark vintage‘ and/or with a ‘benchmark estate‘.
The project — A Model Portfolio
The idea behind the partnership with Saturnalia is to put together a fine wine model portfolio with the help of the tools offered by their platform, which are powerful but can sometimes appear daunting and technical in nature.
My mandate is to show how I would go about using the Saturnalia platform alongside other tools to build my own wine portfolio. I made it available via GoogleSheets here: 🔗 Link to the portfolio
Asset under Management (AUM): £50,000
Universe: Bordeaux, Burgundy, Barolo, Bolgheri & Montalcino.
Names currently in the portfolio:
Ausone, 2016 and 2019 (before 2022 reclassification)
Figeac, 2019 (before 2022 reclassification)
Canon, 2014, 2016 and 2019 (yes, I was expecting it to get upgraded!)
Clos Dubreuil, 2019 (after their reclassification)
Larcis Ducasse, 2020 (WineLeaks #12)
Lynch-Bages, 2016 (WineLeaks #12)
What’s Next?
This project is both something new and a work-in-progress.
The plan is to provide an update on a monthly basis (the first Thursday of the month) with a small monthly update on buys/sells and big movers (positive & negative) as well as a more in-depth quarterly update (first one due on Jan, 5th).
Some of the ideas behind my buys are discussed in the WineLeaks or other long-form pieces; others will be discussed separately.
Please share with me what you’d like to see/know about Saturnalia.
As someone who's about to test a toe in wine investment, this is a really interesting tool to keep watch on. Naturally the market is going to fluctuate based on demand and uncontrollable factors, but hopefully gives a bit more granularity outside of a brokerage! Thanks for sharing