Diversification raises the P90
Across 136 European wind deals, each point of diversification benefit raises the portfolio's P90 by about 1 GWh per TWh of gross yield.
The value is 0.96 GWh of P90 per point per TWh (90% interval 0.93 to 1.00), and it holds both ways: the 69 deals that added points gained 388 GWh of P90, the other 67 lost 292 GWh. In the 2025 low-yield year, the acquired farms outperformed the buyer's other farms in 10 of 14 deals that added more than 1 point, against 20 of 64 deals that lowered the benefit.
Key takeaways
- In the 2025 low-yield year, the acquired farms outperformed the buyer's other farms in 10 of 14 deals that added more than 1 point of benefit, and in 20 of 64 deals that lowered it.
- A point of diversification benefit is worth about 1 GWh of P90 per TWh, gained or lost.
- 69 of 136 deals added points of diversification benefit and gained 388 GWh of P90; the other 67 lost 292 GWh.
- Jädraås, 213 MW in a second wind system, added 26 GWh of P90 to TRIG; Tauron's 180 MW in the same wind system added -0.08 GWh.
Free white paper · 21 pages · Dr. Joaquin V. Ferrer · October 2026
Will your next wind deal make the portfolio steadier, or more exposed?
Two wind farms with the same megawatts and the same P50 can do opposite things to a portfolio's P90. We tested this on 136 European wind deals.
What you'll take away
- What a point of diversification is worth: about 0.96 GWh of P90 per TWh of gross yield.
- Why 67 of the 136 deals made the buyer's portfolio less steady, and what set the other 69 apart.
- How the deals held up in the weak 2025 wind year.
- 136European wind deals
- 0.96GWh of P90 per TWh, per point
- 2025weak-year check
The P90 is the yearly gross yield that a portfolio reaches or exceeds with a 90% chance in any one year. We measured each deal on the buyer's portfolio at the end of 2025, over 30 years of wind resource.
In 2025, deals that added diversification softened the low-yield year
2025 was a low-yield year for most buyers. In the 14 deals that added more than 1 point of diversification benefit, the buyer's other farms were 2.5% below their median gross yield, and the acquired farms 1.1% above theirs (medians of the deals). These acquired farms softened the low-yield year. In the 64 deals that lowered the benefit, the acquired farms were 5.9% below their median, against 2.5% for the other farms. These acquired farms deepened the low-yield year.
Diversification raises the P90/P50 ratio
Wind assets in one wind system have their low-yield years together. Wind assets across independent wind systems do not, so part of their year-to-year variation cancels out. This share is the diversification benefit.1,10 A higher benefit raises the P90 and leaves the P50 unchanged.
7.2% below for the median buyer, 5.7% for well-diversified buyers
Downside spread (P50 − P90)
at a 40% benefit
Each dot is the European wind portfolio of 1 buyer at the end of 2025 (37 buyers). The curve uses the median year-to-year variation of the farms in our sample.
1 Wind resource only: the P50 and the 1-year P90 of yearly gross yield. Long-term, modelling, availability and loss uncertainty are not included, so a full P90/P50 is lower.
A point is worth about 1 GWh of P90 per TWh
This table turns points of diversification benefit into P90. Find the row for the P50 of your portfolio, its median yearly gross yield, and the column for the points that a target adds. The cell gives the P90 gained per year: for example, 1 point on a 5 TWh portfolio adds about 4.8 GWh.
Each point is worth about 1 GWh of P90 per TWh, from the 136 deals in this paper (see Method and data). A target that lowers the benefit costs the same amount, so read the cell as a loss.
| Portfolio gross yield ↓ | +1 point | +2 points | +5 points | +10 points |
|---|---|---|---|---|
| 1 TWh1.000 GWh/yr | +1.0GWh/yr (P90) | +1.9GWh/yr (P90) | +4.8GWh/yr (P90) | +9.6GWh/yr (P90) |
| 2 TWh2.000 GWh/yr | +1.9GWh/yr (P90) | +3.9GWh/yr (P90) | +9.6GWh/yr (P90) | +19GWh/yr (P90) |
| 5 TWh5.000 GWh/yr | +4.8GWh/yr (P90) | +9.6GWh/yr (P90) | +24GWh/yr (P90) | +48GWh/yr (P90) |
| 10 TWh10.000 GWh/yr | +9.6GWh/yr (P90) | +19GWh/yr (P90) | +48GWh/yr (P90) | +96GWh/yr (P90) |
In three deals, the wind system set the P90 gain, not the MW
Lesson 1 · A second wind system adds P90
In 2019, TRIG bought Jädraås, a wind farm in central Sweden. Its other farms are mostly in the UK and Germany, where the good and poor wind years come at other times. So Jädraås made the portfolio steadier, and raised its P90 by 26 GWh a year.11 2025 showed it: the other farms stayed close to their long-term median, while Jädraås was 14% above its own. A farm in a second wind system adds P90 that a valuation of the farm on its own does not see.8
Lesson 2 · A small farm can soften a low-yield year
In 2022, Greencoat Renewables bought Ersträsk North, a wind farm in northern Sweden. It is a small part of the portfolio, and most of the other farms are in Ireland and Germany, in other wind systems. The farm raised the P90 by 13 GWh a year. In 2025 the other farms fell below their long-term median, while Ersträsk North was 18% above its own and covered 14% of the shortfall. A small farm in another wind system softens a low-yield year.6
Lesson 3 · The same wind system adds MW, not P90
In 2018, Tauron bought five In.ventus farms in Poland, where its other farms are too. The added farms share their wind system, so their good and poor years come at the same time. The deal added MW, but almost no P90. In 2025 all Tauron farms fell below their long-term median together, the other farms by 9%. MW in the same wind system add gross yield, not a steadier P90.7
Enlight and Naturgy repeat lessons 1 and 3: Malarberget adds 1.9 points to an already diverse portfolio; ASR Wind adds MW in the same wind system.5,9
Access the diversification analysis for the 136 deals in this paper
For each deal, the analysis shows the farms on the map, the points of diversification benefit, the P90 effect and the 2025 result, as in Exhibits 5 to 7. It covers 37 buyers, 2010 to 2024.
Your portfolio or a deal. Book a call for a deep dive. We return your benefit, your place among the 37 buyers, and the P90 effect of each target — public farm data is enough to start, and a non-disclosure agreement covers anything you add.
Compute the diversification benefit of your portfolio.
Compute the benefit with the target farms added. The difference is the points gained.
Multiply: 0.96 GWh (about 1 GWh) x portfolio TWh x points.
Rank targets by P90 gain per 100 MW added, next to their P50.
“After a low-yield year, ‘hopefully next year is better’ isn’t a strategy. Wind owners need to know what actually happened, because it shapes every expansion and risk decision they make next.”Dr. Joaquin V. Ferrer
Key numbers at a glance
| Measure | Value | Note |
|---|---|---|
| Deals tested | 136 European wind deals | 37 buyers, deals from 2010 to 2024 |
| Value of a point of diversification benefit | 0.96 GWh of P90 per TWh | About 1 GWh; 90% interval 0.93 to 1.00 |
| Deals that added points | 69 deals, +388 GWh of P90 | Gained or lost: the value is symmetric |
| Deals that lost points | 67 deals, −292 GWh of P90 | |
| 2025 check: deals that added more than 1 point | 10 of 14 (71%) | Acquired farms outperformed the buyer's other farms |
| 2025 check: deals that lowered the benefit | 20 of 64 (31%) | Fisher test p = 0.01 |
| P90 below the P50, median buyer | 7.2% | Diversification benefit 24% |
| P90 below the P50 at a 40% benefit | 5.7% | Wind resource only, 1-year P90 |
| Example: 1 point on a 5 TWh portfolio | +4.8 GWh of P90 a year | |
| Wind years modelled | 1996 to 2025 (30 years) | ERA5 with the Global Wind Atlas adjustment |
Modelled gross yield, wind resource only, 1-year P90, at 100% of each farm. Buyer portfolios as held at the end of 2025. Source: Repath analysis.
Frequently asked questions
How much P90 does a point of diversification benefit add to a wind portfolio?
About 1 GWh of P90 per TWh of yearly portfolio gross yield. Across 136 European wind deals the value is 0.96 GWh per point per TWh (90% interval 0.93 to 1.00). On a 5 TWh portfolio, 1 point adds about 4.8 GWh of P90 a year. A point lost costs as much as a point gained adds.
What is the diversification benefit of a wind portfolio?
It is the share of the farms' year-to-year variation in gross yield that cancels out in the portfolio: 1 − σ portfolio / Σ σ farm. Wind assets in one wind system have their low-yield years together; wind assets across independent wind systems do not. A point is one percentage point of benefit. A higher benefit raises the P90 and leaves the P50 unchanged.
Did diversifying deals do better in the 2025 low-yield year?
Mostly, yes. In 10 of 14 deals that added more than 1 point of diversification benefit (71%), the acquired farms outperformed the buyer's other farms in 2025. In deals that lowered the benefit, they did so in 20 of 64 (31%). One year cannot prove the effect and the groups are small, so the paper reads the direction, not the exact shares.
Is this P90 the same as the P90 in an energy yield assessment?
No. The P90 here is the 1-year P90 from the wind resource only. It excludes long-term resource, modelling, availability and loss uncertainty, and it is computed at 100% of each farm, not at the buyer's share. A full P90/P50 is lower. To read the numbers as your own budget P90, scale them by your share of each farm and then by your availability and losses.
Does adding megawatts in the same wind system raise the P90?
Not much. In 2018 Tauron bought 180 MW across five In.ventus farms in Poland, where its other farms are too; the deal changed the P90 by −0.08 GWh. By contrast, TRIG's 213 MW Jädraås farm in a second wind system (central Sweden) added 26 GWh of P90 a year.
Method and data
Each buyer's European wind farms as held at the end of 2025, from Global Energy Monitor holdings, owner reports and deal records.4 A market check against company reports made 1345 changes; for 29 of the 38 buyers with a reported capacity, our gross MW are within 10% of it. A spot check of 20 changes found 2 capacity errors, which we corrected. The acquired farms are the farms of the deal that the buyer still holds. All MW are gross. Two Greencoat vehicles account for 35 of the 136 deals; the 2025 check below holds without them.
Gross yield from turbine-level wind resource modelling, ERA5 with the Global Wind Atlas adjustment, 1996 to 2025.2,3 The model follows year-to-year variation better than the level of yield, so values per TWh and in % are more reliable than absolute GWh. In a test against metered Danish farms, the yearly anomalies correlate at 0.72.
Diversification benefit = 1 − σportfolio / Σ σfarm; a point is one percentage point. Value of a point = 1.28 × the stand-alone variation of the farms, per TWh, pooled over all deals in a hierarchical Bayesian model (90% interval 0.93 to 1.00). The value follows from that identity; the finding is how narrow its range is across deals. "Raised the benefit" means a Bayesian probability of at least 90%, from each deal's block-bootstrap uncertainty; a stricter 95% bootstrap test gives 40 deals. The P90 gain counts the diversification part only; the stand-alone variation of the acquired farms changes the P90/P50 ratio of a deal too, and often more. A method and validation note is available on request.
about 1 GWh of P90 per point per TWh
2025 yield of the acquired farms minus that of the buyer's other farms, each against its own 1996 to 2024 median. Fisher test p = 0.01; rank correlation 0.05, and 0.35 with the benefit computed on 1996 to 2024 only. One year cannot prove the effect, and the strongest offsets come from Swedish farms. The ranking holds in the cuts a reviewer asks for: without the Greencoat vehicles, 8 of 10 top-group deals offset the year against 18 of 42 where the benefit fell; without the deals that added mostly Swedish farms, 7 of 11 against 20 of 64. The groups are small, so we read the direction, not the exact shares.
The P90 is 1-year and wind resource only; it excludes long-term resource, modelling, availability and losses. Results are at 100% of each farm, not at the buyer's share. To read these numbers as your own budget P90, scale them by your share of each farm and then by your availability and losses.
References
- Markowitz, H. (1952). Portfolio selection. Journal of Finance, 7(1), 77–91.
- Hersbach, H., et al. (2020). The ERA5 global reanalysis. Quarterly Journal of the Royal Meteorological Society, 146, 1999–2049.
- Davis, N. N., et al. (2023). The Global Wind Atlas. Bulletin of the American Meteorological Society, 104(8).
- Global Energy Monitor. Global Wind Power Tracker. Farm lists and owners.
- IPP Journal (2019). PNE sale of Malarberget to Enlight.
- Greencoat Renewables (2022). Announcement, 24 June 2022: forward purchase of the Ersträsk North wind farm from Enercon.
- Tauron (2018). Press release: purchase of five wind farms from in.ventus.
- TRIG (2026). Annual Report 2025: production against budget by country and wind resource against P50.
- Energías Renovables (2023). Naturgy and the ASR Wind portfolio, 16 May 2023.
- Choueifaty, Y., and Coignard, Y. (2008). Toward maximum diversification. Journal of Portfolio Management, 35(1), 40–51.
- Repath trace of TRIG's European wind portfolio at the end of 2025: 5.151 MW gross, 23.4 TWh of modelled gross yield a year. Farm lists from Global Energy Monitor, TRIG's reports and a market check in Q4 2025. On TRIG's own reporting basis, the list matches the 1.775 MW that TRIG reports (gap 0%), so we take it as close to complete. The GWh are gross yield from turbine-level wind resource modelling.
Repath prepared this article for information only. All yields, diversification benefits and P90 gains are modelled estimates from turbine-level wind resource modelling (ERA5 with the Global Wind Atlas adjustment, years 1996 to 2025). They are not statements about the performance or decisions of any named company. Deal and portfolio facts come from public sources. This article is not investment advice.
Data: Contains modified Copernicus Climate Change Service information (ERA5), 1996 to 2025; neither the European Commission nor ECMWF is responsible for any use of this information. Global Wind Atlas 3 (DTU Wind Energy and the World Bank Group), CC BY 4.0. Farm lists adapted from the Global Wind Power Tracker, Global Energy Monitor, CC BY 4.0, with changes by Repath. Country outlines: Natural Earth (public domain).
About the author. Dr. Joaquin V. Ferrer is CEO of Repath. He holds a PhD from the University of Potsdam, with research in climate risk modelling at the Potsdam Institute for Climate Impact Research (PIK), and is a licensed civil engineer.