**Short answer:** the strongest relationship in the data is between a startup's funding stage and its founding year. Funding amount is related to both, but more weakly. The data has only a few usable columns, so these are the only pairs I could compute. **What I measured** (pooled across the five monthly tables, with stage coded as an ordered scale from Pre-seed to Series F; step-2 query) | Pair | Correlation | Rows used | |---|---|---| | Founded year vs. stage | **-0.54** | 256 | | Stage vs. ln(Amount) | **+0.40** | 256 | | Founded year vs. ln(Amount) | **-0.25** | 385 for amount | - **Founded vs. stage (-0.54):** older companies are at later stages. The stage table (step-1) shows average founding year falling steadily as stage advances. Pre-seed companies average about 2021 and Seed about 2019. Series A and Pre-series A average about 2018, Series B about 2016, and Series C through F about 2012–2014. - **Stage vs. amount (+0.40):** the median round size rises almost monotonically with stage (step-1): - Pre-seed: $0.5M - Seed: $1.6M - Pre-series A: $3M - Series A: $12M - Series B: $22M - Series C: $60M - Series D: $71M - Series E: $364M - Series F: $316.5M The only inversion is Series F coming in slightly below Series E. The correlation is moderate rather than strong because round sizes vary widely within each stage. - **Founded vs. amount (-0.25):** older companies tend to raise more, but this is mostly an indirect effect of stage. Stage is the stronger link. **Caveats** - **Amount is stored as text.** Values like "$1,000,000" were converted to numbers, and I used the natural log because the amounts are heavily skewed. "Undisclosed" and blank values (35 and 25 rows in step-0) are excluded. - **The "undisclosed" stage is excluded from the correlations.** It has 37 rows and an odd profile: a median amount of $23.5M and an average founding year of 2003. It looks like a mix of older, established companies. - **I did not test the text columns** (Sector, Location, Investor, Founders, What it does), so I can't say how strongly they relate to amount or stage. - **Only about 256 of the 426 rows have both an ordered stage and a founding year**, which limits the stage-based correlations. - **May 2022 is the least reliable table.** It stores Founded as text and has misaligned or blank rows. - **The tables share no join key,** so this is a pooled view of the monthly files rather than a joined analysis.