I’ve reached that age where I’ve started obsessively wondering if everyone else has more money than me. Instead of going to therapy, I decided to spend way too many hours digging into the ACS (American Community Survey) and SCF (Survey of Consumer Finances) to build a definitive way to compare myself to my peers.
The result is, first; not good for my selfworth, second; Broke or Rich.
The "Trust Me" Section:
Privacy: It doesn’t store your data. It doesn't share your data. There are no ads, and it doesn't even use cookies. It’s a static site on Render; your financial insecurities stay in your own browser. I only track the amount of clicks, no IP-adresses or anything.
The Over-Engineered Methodology:
There is no perfect dataset to compare yourself on every metric. So to get around the "missing data" problem, I used:
IPFP (Iterative Proportional Fitting): I used this to synthesize joint distributions from marginal trait data.
Lognormal + Pareto Tails: The bulk of the population fits a lognormal curve, but for the "Rich" part of the site, I had to fit Pareto tails to properly account for the power-law distribution of the top 1%.
Outlier-Weighted Percentiles: My model uses a weighted approach where outliers carry more mass. If you have $5M in the bank but you’re currently "funemployed" with $0 income, you aren't the 50th percentile, you’re still rich, and the algorithm treats you as such.
I’d love to hear what you think of the UI, the data synthesis, or my choice of Pareto index for the right tail.
Mawenzi•1h ago
The result is, first; not good for my selfworth, second; Broke or Rich.
The "Trust Me" Section:
Privacy: It doesn’t store your data. It doesn't share your data. There are no ads, and it doesn't even use cookies. It’s a static site on Render; your financial insecurities stay in your own browser. I only track the amount of clicks, no IP-adresses or anything.
The Over-Engineered Methodology: There is no perfect dataset to compare yourself on every metric. So to get around the "missing data" problem, I used:
IPFP (Iterative Proportional Fitting): I used this to synthesize joint distributions from marginal trait data.
Lognormal + Pareto Tails: The bulk of the population fits a lognormal curve, but for the "Rich" part of the site, I had to fit Pareto tails to properly account for the power-law distribution of the top 1%.
Outlier-Weighted Percentiles: My model uses a weighted approach where outliers carry more mass. If you have $5M in the bank but you’re currently "funemployed" with $0 income, you aren't the 50th percentile, you’re still rich, and the algorithm treats you as such.
I’d love to hear what you think of the UI, the data synthesis, or my choice of Pareto index for the right tail.