Snapshot
Free edition: live nowcasts and satellite data are shown with a 14-day delay (data to 07 Nov 2026).Free edition · 14-day delay Real-time access → ·
Case studies

What Wheat Nowcast would have told you about wheat, in euros per tonne

Five examples from recent seasons, written for the people who buy wheat for mills and pasta makers. For each one: what our model said on the day, what USDA said, what the crop turned out to be, and what Italian cash prices did next. The last one is about where we would not have helped.

These are backtests, not live forecasts. Each season is rebuilt as it would have stood on the day: the model is trained on earlier seasons only and uses only the data published by that date. Our live, time-stamped record started on 10 November 2026.

Case 1 · Soft wheat · Mills

2024: Europe's small crop, seen in May

In mid-May 2024 our model put the European Union's wheat yield −5% against a normal year, and −8% by 1 July. The crop came in −7%. In its July report USDA still had the EU yield 7% too high, and 6% in August.

Our call, 15 May 2024
−5%
EU crop against a normal year
Final crop
−7%
against a normal year
USDA, 12 Jul 2024
+7%
too high against the final crop
Milling wheat, Bologna
+43 €/t
Jul 2024 → Mar 2025

Milling wheat, Bologna, delivered to first buyer

What it meant for a buyer

Bread-making wheat in Bologna went from 222 €/t in July 2024 to 266 €/t in March 2025. For a mill buying 10,000 tonnes over the season, 43 €/t is about 434 thousand euros. Knowing in May that Europe's crop was short was an argument for covering more of the season at harvest instead of waiting.

Prices also follow Russia, the euro, demand and much else: we do not claim that the EU crop alone explains the rise, nor that every buyer could have bought at the July price.

Case 2 · Durum wheat · Pasta makers

2021: the Canadian drought, seen in early July

On 1 July 2021 our model put the Canadian durum yield −16% against a normal year, and −27% by 15 July. The drought turned out even worse: −54%. For all Canadian wheat, on 15 July we were at 23.3 million tonnes against USDA's 31.5; USDA cut to 24 only on 12 August. The final crop was 22.4.

Our call, 01 Jul 2021
−16%
Canadian durum yield against a normal year
Final crop
−54%
against a normal year
Canadian wheat, 15 Jul 2021
23.3 / 31.5 Mt
ours / USDA's (final 22.4)
Durum wheat, Bologna
+230 €/t
Jul 2021 → Nov 2021

Durum wheat, Bologna, delivered to first buyer

What it meant for a buyer

Durum in Bologna went from 308 €/t in July 2021 to 539 €/t in November 2021. A pasta maker who saw in early July that the Canadian crop was falling well below normal, weeks before USDA cut its number, had a strong argument for securing durum before the rise.

The model saw a bad crop, not how bad: the final yield was about twice as far below normal as our mid-July call. Other origins and the Italian harvest also moved the price.

Case 3 · Durum wheat · Pasta makers

2025: a good Canadian durum crop, seen on 1 August

Canada is the largest durum exporter and a main origin of Italy's durum imports. On 1 August 2025 our model put the Canadian durum yield +6% against a normal year (2.71 t/ha against 2.57). The final figure from Statistics Canada was +8% (2.79 t/ha).

Our call, 01 Aug 2025
+6%
Canadian durum yield against a normal year
Final crop
+8%
against a normal year
Durum wheat, Foggia
−23 €/t
Jul 2025 → Dec 2025
Per 1,000 tonnes
23,300 €
difference between buying in July and in December

Durum wheat, Foggia, farm gate

What it meant for a buyer

Durum in Foggia went from 296 €/t in July 2025 to 272 €/t in December 2025. A pasta maker who knew in early August that Canada's crop was above normal had a reason not to rush the purchases of the autumn.

Canada is one origin among several, and the Italian harvest matters as much for Foggia. Our Italian durum model does not beat a normal-year forecast yet, and we say so on the durum page.

Case 4 · Exporters · Everyone

2025: Kazakhstan's big crop, five months before USDA

On 15 June 2025 our model put Kazakhstan's crop at 17.3 million tonnes, against USDA's 14.5. USDA made half of its upward revision only on 14 November 2025, 152 days later. The final crop was 19.3 million tonnes.

Kazakhstan 2025: our numbers and USDA's

Why it matters

USDA's numbers move the whole wheat market on report day. Kazakhstan is not a direct origin for most Italian mills, so we show no Italian price here: this case is about being on the right side of USDA months early. Our number was still below the final crop: the direction was right, the size only in part.

Across 82 large USDA revisions since 2011 we were on the right side before USDA got half way in 63; see Ahead of USDA.

Ahead of USDA →

Case 5 · Our misses

Two times we got it wrong

Russia 2024: USDA was faster

On 1 June 2024 our figure for Russia was 87.0 million tonnes, close to USDA's 88. A late spring frost cut the crop and USDA lowered its number to 83 on 12 June, before we did. We reached 79.5 on 1 July; the final crop was 81.6.

EU 2025: we called the crop too small

On 1 July 2025 our EU yield was −8% against the final figure; USDA was at −5% and a normal year at −6%. The crop came in +6% above normal. A buyer who had bought early for fear of a short crop would have paid more: milling wheat in Bologna went from 246.5 €/t in June to 238.5 in September, in Rouen from 200.6 to 190.8.

A very dry spring in northern France, drier than anything in the model's history, pushed the call too low. We are reviewing the French part of the model before the 2027 season.

How to read these cases

No hindsight, no cherry-picking we did not show

  • Backtest. Every forecast on this page is rebuilt as it would have stood on the day, with a model trained on earlier seasons only. It was not published at the time.
  • Normal year. The yield on the long-run trend line: the forecast anyone can make without our model.
  • Prices. Monthly averages of the weekly cash prices that Italy and France report to the European Commission (DG AGRI), in euros per tonne: Bologna and Rouen from July 2018, Foggia from July 2024.
  • Not advice. The cases show what was knowable, not what a buyer should have done. Prices depend on far more than one crop.

See the next season as it happens

Two weeks of full Pro access, no commitment.

↑ ↓ to choose · Enter to open · Esc to close