Skip to content
Husfarm
EN
Add offer
Esta página también está en español Ver esta página en español →
Husfarm price forecasts · data of 13 Nov 2026

Wholesale price forecasts for the next 12 weeks

Statistical forecasts of 811 wholesale price series from 23 markets in 5 countries, from one to twelve weeks ahead, each with an 80% range, and seasonal price calendars for 133 products. A series gets a forecast only after the method beat “no change” on its own history.

How the forecasts workAccuracy
Series with a forecast
811
of 1,156 tested
Markets
23
in 5 countries
Seasonal calendars
133
products, 256 country rows
Typical error, 4 weeks
10.6%
“no change”: 12.7%
Typical error, 12 weeks
15.0%
“no change”: 22.1%

Typical error: the median difference between forecast and actual price in tests on the last 52 weeks of the published series. Over all tested series, including those that did not get a forecast: 8.6% against 9.4% at 4 weeks and 12.6% against 15.7% at 12 weeks. Full accuracy record

Where prices usually go over the next 12 weeks

Products with at least two forecast series and a move larger than half the typical error. Change: median of the series’ most likely paths from the latest week to 12 weeks ahead, each in its own currency.

Seasonal rise expected

Products with an expected seasonal rise
ProductMost likely changeTypical errorSeries
Soyabean Green (Edamame)+57.5%±27%5
Watermelon+36.5%±24%17
Potato+27.0%±25%6
Sweet Corn+24.9%±30%2
Asparagus+21.5%±18%19
Eggplant+20.9%±11%20
Lemon+20.3%±7%5
Burdock+20.2%±18%11
Melon+19.5%±18%20
Lettuce+18.7%±15%11

Seasonal fall expected

Products with an expected seasonal fall
ProductMost likely changeTypical errorSeries
Pea−32.8%±32%18
Strawberry−25.1%±34%13
Tomato−15.4%±18%40
Kiwifruit−13.8%±6%10
Yam−5.8%±6%10
Banana−5.4%±10%2

Forecasts and seasonal calendars by product

134 products. Forecast series and countries with a seasonal calendar for each.

Vegetables 55

Fruits 26

Other 11

Fishes 10

Grains and cereals 8

Dairy and eggs 7

Herbs 5

Plants 4

Meat and poultry 4

Processed food 2

Spices 1

  • Ginger 11 forecasts · 3 calendars

Aquaculture 1

How the forecasts work

Starting point
The latest weekly average price of the series, in the market’s own currency and unit, as published.
Seasonal step
For each week ahead, the change the same series showed over the same weeks in each of up to three previous years. The median of those changes is taken, and half of it is added: past seasons repeat only in part, and the half step proved more accurate than the full one.
Test on its own history
Before publishing, the method is run from each of the last 52 weeks as if that week were today, and compared with what happened. A series gets a forecast only if, at 4, 8 and 12 weeks together, its errors were smaller than those of the simplest forecast: “the price will not change”. 811 of 1,156 tested series passed at the latest update.
The range
The 80% range spans the 10th to the 90th percentile of the series’ own forecast errors in that test, for each horizon. In a series with stable prices it is narrow; in a volatile one it is wide, and it never narrows further ahead.
What it cannot see
Weather, harvest size, trade measures, animal disease, transport costs and currency moves are not in the model. A forecast repeats what the season usually does, so it is least reliable when this year is unusual.
Updates and record
Forecasts are recalculated with every data update. Each week, the forecasts issued are archived unchanged and later compared with the prices that followed; the results are published on the accuracy page.

A statistical forecast, not advice

The forecasts extend each series’ own seasonal pattern; they do not know about this year’s weather, harvests, policy or currency moves. Use the range, not only the central figure, and do not base trading, purchasing or credit decisions on them alone.

More: Accuracy record · Methodology · Price indices · API

Questions and answers

Can wholesale food prices be forecast?

In part. Prices of fresh produce follow seasons: harvests, imports and demand repeat each year. A forecast that extends this pattern beats “no change” for many series, which is the test every Husfarm forecast has to pass. Weather, policy and currency shocks cannot be forecast this way, so every forecast comes with a range.

How far ahead do the forecasts go?

Up to 12 weeks, week by week. Beyond three months the seasonal pattern of past years says little about this year, and the errors grow beyond use.

Why do some products or markets have no forecast?

A series needs at least two years of weekly or ten-day quotes and a current quote, and it has to pass the test on its own history. Many series are younger than two years, quoted monthly or not seasonal enough; they get no forecast, but the product may still have a seasonal calendar.

What does the 80% range mean?

In tests on the series’ own history, 8 out of 10 forecasts ended inside a range of this width. The actual price can fall outside it, about once in five times by design, more often in unusual years.

How often are the forecasts updated?

With every data update, at least once a day. The week’s forecasts are archived every Monday and compared with the prices that followed on the accuracy page.

Can I use the forecasts?

Yes, under CC BY 4.0 with credit to Husfarm and a link. They are also in the free API as JSON. They are statistical estimates, not offers or advice.

Forecasts are calculated from Husfarm price data and change with every update. Data licensed under CC BY 4.0.