Key Takeaways
- Expana launches 500+ new forecasts
- Proof: An October 2024 mahi mahi model was left untouched for 2 years — no retuning — and correctly called both winter lows, the record $11/lb summer 2025 high, and continued climb through April 2026.
- How it works: Trend and seasonality are modeled together, not as fixed layers — so a winter dip on a rising trend ≠ a rally break, and a seasonal high ≠ a market top. Shocks (disruptions, demand spikes) are recognized separately so they don’t get mistaken for normal seasonal moves.
- Why it matters: Misreading trend or season costs procurement teams real money — chasing a market that keeps rising, or mistiming coverage around dips/spikes that were never what they looked like.
- Rollout: Mahi mahi was the proving ground; Expana IQ Forecasts are now live across the Expana platform.
Season and trend are buried in years of price data, easy to miss on their own. Put them together, and they point to where a price could be heading next.
In previous articles we have looked at what seasonality and trend are: the calendar a market repeats every year, and the slower multiyear story sitting underneath it. This is where they earn their keep. Read together, they beat chasing whatever headline moved the market last week, because by the time news reaches a forecast, it is already priced in.
Expana IQ Forecasts are a new feature on the Expana platform: model-driven price projections that learn a market’s own trend, seasonality, and shorter-term dynamics, and how those components shift over time, then project them across an 18-month horizon.
The feature only reached the platform recently, but the work behind it started years earlier. In October 2024, before any of this was built for release, a model was trained on mahi mahi and then left untouched: no retuning, no quiet correction with the benefit of hindsight. That was deliberate. The only way to know whether a forecast holds up is to test it the one way that actually means something, against prices nobody could have known yet. What you see in the chart below is what a user would have gotten had Expana IQ existed and been handed to them back then. By the time the feature reached the platform, it already had a two-year track record, not just a backtest. Two years on, price has tracked the mahi mahi forecast in direction, magnitude, and timing, across both the seasonal and trend components.
That result is worth breaking down. For two decades, mahi mahi prices have climbed on a schedule tight enough to set a calendar by. Since 2023, that seasonal climb has ridden on top of a renewed uptrend, one that pushed the 2025 summer high past $11 a pound, a level the market had never touched in twenty years of data. The forecast ran 18 months, through April 2026, projecting lows in both winters, a high in summer 2025, and a market still climbing as it closed. All three have held, and prices have kept trending higher since.
How the forecast worked
Every year, mahi mahi prices follow the same rhythm: a dip each winter, a climb into summer, a cycle that has repeated with consistency, though the size of the swing shifts some from year to year. That rhythm is what the forecast leaned on, anticipating a low each winter and a high each summer, not by comparing this year to the last ten, but by checking whether the rhythm itself was still holding true, and whether its dynamics, how large the winter dip and summer climb ran, how sharply they turned, were shifting from one year to the next.
Mahi mahi’s overall trend bottomed out in the late 2000s, climbed for years to a peak in the middle of the 2010s, then rolled over as the drivers faded. A second climb began in 2023, and the forecast read it as still underway rather than a plateau, which is why it projected the market to keep pushing higher through the 2025 summer high and stay on a rising path all the way to the April 2026 close.
Because both seasonality and trend were moving at once, the underlying models had to read them together, letting each component’s read shift as the other moved rather than holding either one fixed. A price swing is a step off whatever baseline the trend has set, not a fixed number: the same seasonal percentage produces a different dollar swing as the baseline itself rises or falls.
That let the forecast treat the winter dips as normal pauses on a rising floor, not a break in the rally. It read the summer run toward $11 as the expected seasonal high, not a sign the market had peaked, all while keeping the underlying uptrend intact. That is why its own path was still climbing, not leveling off, when the window closed in April 2026.
That same discipline is what keeps the read clean when something else entirely hits the market. A shipping disruption, a disease outbreak, a sudden pull from a new buyer: none of these sit on a calendar or build for years, but each leaves its own signature, how far it knocks a price off that baseline and how long the gap takes to close. Expana IQ learns to recognize that signature on its own terms, so a real shock does not get mistaken for a seasonal dip or a trend break, and the seasonal and trend reads underneath it stay intact.
What the forecast got right
Put together, that is what the October 2024 forecast for mahi mahi got right: a seasonal rhythm still holding after two decades, an uptrend still climbing rather than plateauing, and no shock along the way to throw either off course. Both winters came in low as expected, summer 2025 peaked on schedule, and the market was still climbing when the 18-month window closed two years later. This is the kind of result the approach aims for when season and trend are read as they actually move, together, rather than as two fixed assumptions bolted onto a chart.
What the forecast says now
As of today, the same models are still tracking mahi mahi, refreshed on the latest data, and the read hasn’t changed. The models continue to read the 2023 uptrend as intact, with the seasonal rhythm repeating on top of it: a winter low, then a summer high, then another winter low, each one landing on a floor that keeps rising rather than leveling off.
Why it matters for procurement
Getting a forecast wrong is expensive, and nowhere more than in procurement, where sizing and timing both depend on it. Misread the trend, and a buyer books coverage assuming prices have plateaued, then spends two years chasing a market that keeps climbing past levels it had never touched before. Misread the season, and a buyer locks in during what looks like a normal winter dip but is actually a pause on that same rising floor, or waits out a summer spike that was never going to reverse. Coverage built on that kind of reading is less likely to be thrown by a normal winter dip or summer spike, because both reads are tracking dynamics that are still genuinely moving rather than guessing at a fixed target. How much coverage to take, and when, remains each buyer’s own decision, based on their own policy and risk profile.
How Expana approaches this
Expana IQ applies statistical and machine learning approaches to create forecasts driven by data. For each commodity, Expana IQ selects whichever configuration best fits that market’s own behavior and performs best against its historical data. The model learns how a market normally moves within a season and over a longer trend, how those patterns shift over time, and what a shock does to them when one hits, all learned together rather than pieced together after the fact. That is what turns a forecast into something that can inform a decision on sizing or timing, rather than a description of what already happened.
The mahi mahi forecast from one of those early models is one example of what this approach can do. The models behind it are now becoming available on the Expana platform, putting the same forecasting power within reach for anyone making a sourcing decision.
For more information on Expana’s forecasting methodologies and insights, request a demo.
This commentary is prepared by Expana and its group of companies, neither of which is an investment firm. We have no positions in the commodities or derivatives referenced. Expana IQ Forecasts are model-driven projections provided for information and educational purposes only; they are not a guarantee of future prices or results, and the mahi mahi example is a single illustration that is not representative of all forecasts. The views expressed are for information only. See our disclaimer for more information: https://www.expanamarkets.com/disclaimer/
Written by Andrei Rjedkin
