You Own Fewer Bets Than You Think

On 13 Aug 2026, the Climate Prediction Center put the odds of a very strong El Niño this winter above 90%, and assigned a 69% chance to a historic event during the October to December season, one that would exceed every El Niño in the record going back to 1950, measured on a three-month relative Niño index of +2.5°C or more.[1] In June, when this series began, CPC put the odds of a very strong event at 63%, measured against the conventional +2.0°C threshold for November to January.[2] The two forecasts use different index thresholds and seasons, so I would not compare 63% and 90% directly. What has changed is the forecast weight: within two months, the probability attached to an extreme El Niño has increased materially. That is faster than most underwriting and portfolio review cycles.

Five of the eight strong El Niño events since 1950 were followed by La Niña, and the transition tended to happen quickly, sometimes with only a single neutral quarter in between.[3] In an IC, I would therefore run the sensitivity across at least two years rather than stop at the El Niño year. A dry phase can be followed quickly by a wet one, leaving the same portfolio exposed to different resource and market stresses across consecutive years.

This series has covered how weather reaches the revenue line, why the standard downside case can understate it, and what happens when the input crosses a border. Next, I would look at this at the portfolio level. That is where several individually manageable resource risks can become one concentration.

Diversified by name, concentrated by factor

A book of four assets in three countries across two technologies will usually look diversified in an IC paper. Recut the same portfolio by ENSO sensitivity, and it may look concentrated.

A run-of-river plant in Laos, a wind portfolio in Tamil Nadu, and a solar platform in Java have different offtakers, regulators, construction risks, and currencies. That separation matters for project risk. ENSO correlation cuts across those categories: the same Pacific temperature anomaly can weaken the monsoon in one market and suppress the wind season in another. Most infrastructure portfolio construction is organised around geography, technology, and counterparty. Climate sequence risk does not map neatly onto any of the three.

The position count can therefore overstate the fund's number of independent bets. Correlated shortfalls ultimately show up in the same place: lower EBITDA, weaker cash yield, or missed distributable cash. Vietnam in 2023 showed the national-scale version, where a drought became a USD 1.4 bn economic event within weeks.[4] At portfolio level, several modest misses are easier to dismiss as unrelated. I would treat them as repeated exposure to the same factor.

Distance is a weak proxy for diversification here. What matters is how each asset responds to the same climate phase. Assets can be separated by 2,000 kilometres and still move in the same direction under El Niño. A position with a genuinely different sensitivity can do more for portfolio construction than another asset that simply adds a new country to the map.

The recovery year is the second risk

In most base cases, the wet year after a drought looks helpful because it pulls the multi-year average back towards plan. I would be careful with that assumption.

Catchments drawn down through a drought refill, but once storage and turbine constraints bind, incremental inflow has limited value and may create spill or flood-management constraints. Within a year or two, reservoirs can move from shortage to inflows that exceed what can be stored or turbined. Grids that relied on thermal backup through the dry phase can also face curtailment or negative pricing when renewable output recovers.

For equity, the problem can occur twice: lower generation during the drought, followed by spill, curtailment, or weaker prices when output returns. The same long-run averaging that can understate drought exposure can also understate the consequences of the wet phase. A warmer atmosphere holds more moisture, which steepens the wet extreme even where event frequency is unchanged. I would not model the La Niña year simply as mean reversion back to plan. It deserves its own operating and price sensitivity in the downside case.

You cannot hedge the Pacific

A natural portfolio response is to hedge the common factor. I am sceptical that a clean instrument exists for a mid-sized infrastructure portfolio.

Weather derivatives and parametric structures exist, and they work well where the trigger sits close to the asset, such as rainfall at a specific gauge or temperature at a specific station. They are harder to use at the scale of ENSO. An instrument that settles on an ocean index tells you the Pacific warmed. It does not tell you whether a particular catchment filled, whether a wind season arrived, or whether haze and heat cancelled an irradiance gain. That gap between the index and the asset is basis risk, and at basin scale it can be large enough that the premium buys imperfect correlation rather than reliable protection.

This is an analytical judgement, and it is contestable at the margin. A sufficiently large portfolio may average out some of the basis risk. For a mid-sized book with a handful of concentrated positions, that is difficult to rely on. The hedge can still settle very differently from the EBITDA sensitivity it is meant to offset.

In practice, I would spend more time on position sizing, contract structure, and what the next asset does to the existing book. Those are imperfect mitigants, but they are within the investor's control.

What actually works is bought at entry

Storage is one of the clearest structural offsets because its economics can improve when price volatility rises. Australia's National Electricity Market provides the cleanest evidence available: the average price spread captured by batteries reached AUD 243 per MWh in the fourth quarter of 2024, up from AUD 129 per MWh a year earlier, and quarterly net revenue for the grid-scale fleet more than doubled over the same period.[5]

The current Australian market also shows why I would be careful calling storage a hedge. Through the same months, the El Niño probability moved from 63% to above 90%, and NEM battery revenues fell to record lows. May 2026 averaged AUD 29,000 per MW per year, the weakest month since the index began in July 2022, with NEM-wide two-hour spreads down 65% year on year, from AUD 280 to AUD 99 per MWh.[6] August 2026 held flat at AUD 44,000 per MW per year, with no price anywhere in the market above AUD 500 per MWh.[7] The factor was building, and the offset was not paying.

There is an important caveat. The revenue compression is primarily a supply story. A grid-scale fleet 2.6 times its year-ago size is competing for the same arbitrage pool, and household batteries are flattening the evening peak that arbitrage depends on.[6] The 2026 numbers do not show storage failing as an ENSO offset. They show Australian storage economics running through a build cycle. For equity, that distinction does not remove the underwriting problem: a position sized on drought-year spreads still has to survive the years when the spread is absent. I would underwrite storage as a standalone merchant or contracted investment first, then give credit for the portfolio benefit.

Hybridisation is more targeted. Solar and hydro are seasonally asynchronous: solar resource is strongest in the dry season, precisely when reservoir inflow is weakest, and hydro can firm solar through the wet season when irradiance falls.[8] Floating solar on an existing reservoir captures that complementarity while sharing the grid connection already built for the dam, and reduces evaporation from the reservoir surface at the same time.[9] For an owner already exposed to hydro, the investment can improve the cash-flow shape of the existing position rather than simply adding another generation source.

The cheapest mitigation is usually in the contract, provided it is negotiated before signing. Deemed generation and resource availability provisions determine how much contracted revenue survives a year when the resource does not arrive. In an IC paper, I would put these clauses alongside tariff, tenor, and counterparty credit because they determine how much of the resource shock reaches EBITDA and cash yield. They often receive less diligence than the headline tariff even though they can materially change the downside case.

What this means for investors

I would add one table to the portfolio review: every material renewable position down the rows, ENSO sensitivity across the columns, with EBITDA and cash-yield impact under a strong event.

The first test is straightforward: under a strong El Niño scenario, how many material positions miss at the same time, and by how much? Translate the output into portfolio metrics such as EBITDA sensitivity, cash-yield impact, and exposure as a percentage of NAV or committed capital. If most of the book moves in the same direction, that belongs in concentration limits and portfolio construction policy. It should not sit only as a risk paragraph inside individual IC papers.

The same framework should inform new investment decisions. An asset with genuinely offsetting climate sensitivity may justify a different incremental return assessment because of its impact on existing exposures. Another attractive hydro, wind, or solar position may need a higher entry hurdle if it increases a factor concentration already in the portfolio. Standalone entry yields miss that effect.

Portfolio shape, contract terms, hybridisation, and storage sizing are largely determined at commitment. If the exposure only becomes visible after the dry season starts, it is too late to fix through portfolio construction. It needed to appear in the underwriting case or concentration limit before the capital was committed.

Closing view

The bigger problem is organisational. The deal team diligence resources at the asset, while macro and strategy cover rates, currency, and country. A Pacific climate signal that raises the probability of correlated shortfalls across several assets can sit between those mandates. The risk gets discussed, but no one is clearly required to quantify its impact on the portfolio or the next investment.

I do not think the answer is a better ENSO forecast. CPC already provides one. I would rather know who owns the exposure inside the investment process, how it appears in the portfolio review, and when it changes the entry hurdle for the next deal.

If those questions have no clear owner, the portfolio probably has more climate concentration than the position count suggests.