Why this project
The 'BTC, XAU & SPY Performance Analysis' is my latest Excel mini project. I wanted to see whether I could pull real market data, build it into a spreadsheet, and compare the performance of three very different assets over the same five-year window.
I ran into basic fundamental issues and, laughably, ended up arguing with GPT about the results. The data showed some genuinely interesting findings, and GPT correctly advised caution about making assumptions — part of its restrictions, I suppose.
Data collection
The data came from investing.com, covering BTC/USD, XAU/USD and SPY/USD from January 2021 to January 2026.
Getting the data into Excel was straightforward. I keep thinking my projects need more heavy cleaning, but so far the data has been pretty simple — converting to tables and making pivots where necessary. In fact, most of this table did not require pivots...
The indexation problem
The most interesting technical barrier was figuring out how to compare these assets fairly.
Bitcoin trades in the tens of thousands. Gold is around two thousand. SPY is in the hundreds. A raw price chart tells you nothing useful — you have to index each asset so they start from a common baseline.
I had no idea how indexing worked or why it was used. I just saw it as a representation of value on stock exchanges.
But smack everything with 100 as a baseline — to get it through my head, you REPLACE the first value or price with 100 — then:
big lightbulb moment for me.
What the data actually says
Once I had the indexed comparison working, a few things stood out:
Gold returned approximately +180% over the five years, the highest of the three. The S&P 500 had the lowest return by index change. Even GPT disagreed with me, but I checked the stock exchange — it was accurate.
Thing is, I live in the UK. From my own trading efforts, I made money trading against the dollar in 2021–2022, let alone gold. I wince at the lost potential opportunity (not trading advice!).
Bitcoin was far more volatile. It saw its biggest and longest drawdown between May 2022 and March 2024, followed by a sharp, shorter increase from November 2024 to January 2025. Gold climbed steadily throughout, while SPY and BTC ended at a similar level — with SPY slightly ahead.
The volatility comparison was striking. Bitcoin's biggest dips were in May 2021 and January 2022. Gold and the S&P 500 saw similar downturns during the same periods, but with far less extreme moves.
What I learned
The biggest takeaway from this project was understanding how easy it is to misrepresent financial data without meaning to. Or better yet, being gaslit into thinking you are wrong by overcautious LLMs. There is a balance to be had.
As I try to become an analyst — maybe engineer one day — it hurts to think I do not understand basic things. I had to research standard deviation too, something I have used before but did not know it was called that...
This is information from my dataset, extremely contextual, and should not be used in any financial decisions. But it is exciting to think that as I get good at this and reference other data points, I could use analysis to help make some financial decisions in the future.
Better yet, some LLM use to help better predict future movements. But I am not there yet, and I need to be patient and keep studying.