Why I try not to be a ‘Model Believer’

Early in my career, I was working on an infrastructure deal when my boss leaned over and said, half-jokingly about someone else in the room: ‘Careful - he’s a model believer.’

The phrase stuck with me because it captured something subtle and important: it wasn’t about technical skill; it was about posture and the risk of mistaking the precision of a model for the reality of a deal.

Excel just turned 40, and despite repeated predictions of its demise, it remains the most important piece of software in corporate decision-making. As Bloomberg recently put it, Excel is not just surviving the rise of AI - it is absorbing it. The 'green-and-white' grid endures because it forces assumptions to be explicit and trade-offs to be visible. That power, however, is exactly what makes it dangerous when mistaken for reality.

(For those interested, the Bloomberg piece is here: https://www.bloomberg.com/features/2025-microsoft-excel-ai-software/)

I’ve been building models in Excel for about 20 years now - first in the US, and later from Paris on deals across Europe and beyond. I enjoy designing models that are clean, transparent, and fit for purpose. And maybe because of that, I’ve learned not to believe in them.

1. The role of financial models as useful fiction

Financial models sit at the center of almost every investment decision in renewable energy:

•we use them in development: to choose the right location, select technology, optimize layouts, confirm financial viability, and bid or negotiate offtake prices

•we rely on them for the construction phase: to approve capex, size debt, structure tax equity, and secure financing

•we also rely on them throughout a project’s life, as the models underpin valuation and capital allocation decisions

But models play very different roles at different moments. A development model, a financing model, and a portfolio model should not - and cannot - be the same thing. Trying to use a financing model for development is like trying to use a microscope to navigate a forest.

2. A financial model is:

•a way to estimate value and to understand what that value is actually sensitive to

•a framework to test risk allocation and scenarios

•a decision-support tool

•a way to communicate a project’s story, in the common language of numbers

3. A financial model is not:

•a decision-maker

•a neutral collection of formulas - structure, layout, and presentation matter as much as math

•a black box that produces answers

•a representation of static or absolute truth

In most cases, a model exists to confirm what experienced people already sense through intuition and back-of-the-envelope math, typically in a more precise form. And occasionally, it surfaces something non-obvious - like a tax efficiency or a debt-sizing constraint - but it should never replace judgment.

I can’t count the number of times a seasoned professional has caught modeling errors just by glancing at the outputs without diving into the model itself. They see that the shape of the cash flow is wrong, much like a pilot knows the engine sound is off before looking at the dials. This shows how often models simply formalize and communicate what experience already suggests, putting it in black and white for broader buy-in.

4. What I mean by a ‘model believer’

A model believer isn’t someone who’s bad at modeling. Quite the opposite - they’re often very good technically.

A model believer is someone who:

•treats the model output as an answer, rather than as evidence

•responds to uncertainty by adding complexity

•believes that if the spreadsheet is detailed enough, the decision must be sound

But models don’t reduce uncertainty; they just organize it. Precision is not accuracy, and a perfectly-balanced waterfall does not mean you’re asking the right question.

This matters particularly in capital-intensive sectors like renewables and infrastructure, where small modeling assumptions can appear to justify very large bets.

In most deals, a small handful of inputs drive the vast majority of the outcome: project capacity, net capacity factor, key dates (start of construction, commercial operation date), total installed cost, power price shape, and the cost of capital. Other elements tend to be second-order noise.

Non–model believers don’t ignore the rest of the spreadsheet, but they know where to focus their efforts. They spend time designing the deal, the structure, or the risk allocation around the variables that truly move value, rather than polishing decimals that don’t move the outcome.

5. Models as decision engines, not valuation machines

One of the biggest mistakes I see is trying to make a single model do everything.

Deal pricing… portfolio strategy… development decision-making… liquidity planning… board reporting… financing negotiations… Those are fundamentally different decisions - yet they’re often forced through the same 100 MB macro-heavy monolith.

In practice, better decisions don’t come from a better model. They come from a clearer, or often simpler, understanding of what decision the model is meant to support.

6. A concrete example

We saw this clearly when working with a renewable energy developer who was transitioning from a pure development model to a combined developer–independent power producer (IPP) platform.

Their existing financial tooling had been built around a legacy M&A-focused engine - technically impressive, but optimized for producing transaction valuations rather than supporting ongoing operational and strategic decisions. It was built to sell, not to steer.

Instead of building a more sophisticated model, we stepped back and designed a suite of models, each with a clear role:

•a light project model for fast, transparent development decisions

•a corporate model for liquidity planning, capital allocation, and portfolio strategy

•a separate lender-grade project finance model, used only when that level of cash-flow granularity actually mattered

7. Why senior people are often right to distrust models

There’s a reason experienced investors, executives, and board members often appear anti-model. They’re not rejecting analysis. They’re rejecting tools that don’t capture current reality and which:

•take minutes to recalculate after a simple change

•hide key mechanics behind layers of complexity

•answer questions no one is actually asking

When models are built as monuments to technical skill rather than as tools for judgment, skepticism is not only understandable - it’s rational.

8. So why I’m not a model believer

Models are indispensable, but belief is the wrong posture. Curiosity beats conviction, judgment beats precision, and clarity beats complexity almost every time.

The best models don’t just provide answers, they help you ask better questions and allow sound, rapid-fire analysis when it matters.

That’s the difference between building a model and building a decision engine. In a sector where policy, power prices, technology, and capital markets all move faster than any spreadsheet can keep up with, that distinction matters.