Columbia University’s new research method vs. Dashflow’s user-friendly solution
Let’s cut through the jargon for a moment. If you’re a UK Chartered Surveyor or American Capital Markets broker who’s more at home interacting with institutional clients than wrestling with complex financial spreadsheets, you’re not alone.
The rabbit hole of financial modelling and crunching IRRs isn’t why you got into Commercial Real Estate—it’s about the properties, the people, and the transaction life of the market. Yet, the industry keeps pushing more tech and more intricate software your way, expecting you to become a part-time data analyst – even suggesting we retrain as expert AI prompting engineers.
But what if there’s a way to handle the number-crunching without getting bogged down in algorithms, formulae and AI prompts? Imagine focusing on what you do best—building relationships and sense-checking great investment opportunities—while knowing the auditable and transparent financial modelling is backing you all the way. That’s where the best developments today come into play: bridging the gap between sophisticated cashflow modelling and the practical needs of professionals working in the urban environment.
Columbia University’s latest research into AI and cashflow modelling
Recently, researchers at Columbia University posted a novel approach that emphasises the importance of a double-audit system for both modelling and verifying IRRs. Using advanced ChatGPT-type prompting techniques, they aim to cross-validate IRR calculations performed in Python with those in Excel. While this method represents a significant advancement, it also brings to light several challenges that hinder its practical adoption in the industry.
The Double-Audit approach: an innovative yet complex solution
Columbia University latest exploratory research involves a double-audit mechanism involving two separate computational methods aided by the likes of ChatGPT to produce:
- Python Calculations: to calculate a deal IRR using Python scripts, which can offer precision and the ability to handle complex datasets.
- Excel Verification: to generate Excel-friendly outputs through AI prompts which allow the use of Excel IRR functions to cross-validate the Python results.
The idea is that if both IRRs match (i.e. Python IRR = Excel IRR), it provides assurance that the modelling is accurate, enabling investment professionals to focus on validating input assumptions—the domain where their human expertise remains indispensable (i.e. yield entry and exit pricing, rents psf, market rents psf, &tc).
The Challenges: technical complexity, AI risks, and unrealistic expectations
While innovative, this AI-Python-Excel approach assumes a level of expertise and conditions that do not align with the current and medium term realities of the CRE industry:
- Advanced new technical skills required: Investment surveyors and brokers would be expected to be proficient not only in Python but also in drafting precise AI prompts to generate accurate Excel cashflow representations. This level of technical expertise is uncommon among professionals whose primary focus is on real estate, asset management, market analysis and creative problem-solving in bricks ‘n mortar scenarios.
- Fluency in Discounted Cash Flow (DCR) theory: The approach assumes that users have a deep understanding of DCF modelling theory and the Time Value of Money. However, in practice, many CRE professionals rely on software solutions that abstract these complex calculations, allowing them to concentrate on strategic decision-making rather than the mathematical underpinnings.
- Expert-level English proficiency for AI prompts: Effective AI prompting demands not only technical knowledge but also the ability to articulate complex financial concepts clearly and precisely in English – first time and every time. Think about the amusing but stickler-for–English-precision Stephen Fry or Simon Heffer’s ‘Strictly English’ grammar skillset. Every speedily written sentence amid office pressures is liable to a misunderstanding or at least an ambiguous use of language. Each of these instances is a huge deal for complex prompts. Each can lead to potential major AI misinterpretations, resulting in errors that are not only grotesque but difficult to detect within a convoluted multi-human-assumption workflow.
- Risk of AI hallucination: AI models like ChatGPT can sometimes produce outputs that are plausible but incorrect—a phenomenon known as hallucination. This risk introduces uncertainty into the verification process, potentially compromising the integrity of the financial models.
- Time-consuming and laborious: The back-and-forth process between Python scripting, AI prompting, and Excel verification is cumbersome. It detracts from the core responsibilities of investment professionals, who are better utilized in analyzing deals and crafting investment strategies.
In brief, all the above makes user-testing –let alone software-company testing– extraordinarily difficult if not impossible. How does one test for consistent and accurate IRR calculations? How does one solve a deterministic problem with a none-deterministic tool like ChatGPT or its ever-evolving Large Language Models (LLMs) upgrades?
Dashflow: simplifying IRR verification without the technical hurdles
Enter ‘Modelling-As-A-Service’ (MaaS) —an award-winning methodology that has been transforming IRR verification and automating Excel modelling since 2017. The Team at Dashflow recognised early on that CRE professionals need tools that enhance efficiency without adding more complexity to their busy day. Here’s how Dashflow addresses the challenges posed by the Columbia research:
- Immediate Double-Audit without technical barriers: MaaS already provides two separate calculation modes—e.g. an in-app IRR and an automatically generated Excel cashflow model with live spreadsheet formulas. Users receive instant verification without needing to write Python code or craft and adjust AI prompts.
- No requirement for deep DCF theory knowledge: Dashflow’s user-friendly interface is an example of how to abstract away the complexities of DCF modelling. Users input deal parameters through intuitive screens, and MaaS handles the intricate calculations behind the scenes.
- Elimination of AI misinterpretation risk: By automating the entire process within a controlled environment, Dashflow removes the uncertainties associated with LLM outputs. The Excel models produced via AI logic and reasoning are transparent and easily interpretable.
- Efficiency and time savings: Dashflow streamlines the modelling process, allowing professionals to focus on validating their assumptions and making informed decisions rather than getting bogged down in technical verification steps (i.e. getting mired in butchered spreadsheets and exponential manual labour).
Why Dashflow’s Modelling-As-A-Service approach is a superior method for CRE Pros
- Designed for the industry: Dashflow understands that CRE professionals excel in real estate expertise, asset management, and market analysis—not necessarily in programming or advanced financial modelling.
- Transparent and accurate: The generated Excel models (without additional prompts) provide full transparency, showing exactly how the IRRs are calculated. This level of consistency in output and clarity is essential for presenting in a timely manner to investment committees, lenders, colleagues and JV partners.
- Proven performance: Dashflow’s IRR calculations have been demonstrated to be within 5 basis points of traditional, expensive software solutions like Argus, but with the added benefit of greater transparency and ease of use.
- User-Friendly Experience: With an interface that requires no specialized technical skills, Dashflow empowers professionals to do what they do best while leveraging powerful automation to enhance accuracy and efficiency.
Conclusion: Embracing practical innovation in CRE modelling
Columbia University’s initial findings shine a light on the critical need for robust IRR verification methods in the CRE industry. However, their proposed solution, while intellectually stimulating, may not be practical for widespread adoption. It relies too much on advanced technical skills and assumptions about user expertise.
Dashflow’s Modelling-As-A-Service (‘MaaS’) approach offers a pragmatic alternative that delivers the benefits of an instant double-audit system without the mountain of complexities and delays in crafting prompts. By automating cashflow modelling and providing immediate, transparent verification, Dashflow aligns with the real-world needs of CRE professionals dealing with a range of higher priorities in the workplace (origination, due diligence, strategy, completion, market updates, client events, etc).
Experience the future of CRE financial modelling
In a field where time is money, and accuracy is paramount, professionals need tools that bridge the gap between sophisticated financial modelling required for institutional clients and practical usability between calls, meetings and relentless updates as views change, are tweaked and adjusted per imagined scenario. The right tool needs to remove unnecessary hurdles, reduce risks, and allows professionals to focus on strategic decision-making.
Discover how to transform your investment modelling process
Ready to enhance your IRR verification without the technical headaches?
Get in touch to learn more about Modelling-as-a-Service (MaaS) and embrace the best, efficient, cost-effective and reliable way to build your CRE financial models.