**The strongest relationship in this dataset is between a startup's founding year and its funding stage (r ≈ −0.54). Stage and deal size are next (r ≈ +0.40), then founding year and deal size (r ≈ −0.25).** These came from pooling the five monthly tables (426 rows) and computing correlations on a numeric version of each column. | Pair | Correlation | What it means | |---|---|---| | Founded vs. Stage (ordinal) | **−0.54** | Older companies sit at later stages. Average founding year is about 2021 for Pre-seed, 2019 for Seed, 2018 for Series A, 2014 for Series C and E, and 2012 for Series D. | | Stage vs. ln(Amount) | **+0.40** | Later stages raise more. The median round rises from about $0.5M for Pre-seed and $1.6M for Seed to $12M for Series A and $22M for Series B. It reaches $60M for Series C, $71M for Series D and $316M–$364M for Series F and E. | | Founded vs. ln(Amount) | **−0.25** | Older companies raise larger rounds, but this is weaker. It is largely explained by stage. | The correlations use 385 rows with a usable amount and 256 rows with a usable stage. The relationship between stage and amount is clearly monotonic in the stage medians, apart from the "undisclosed" stage group. **Caveats** - **Amount is stored as text.** Values like `$1,000,000` and `Undisclosed` had to be parsed, and I treated "Undisclosed" and blanks as missing. There are also malformed entries such as `$11.300,000`, `2,000,000` without a `$`, and `6400000`. - **Stage is messy.** About 127 rows have a blank stage. Some have variants (`Preseed`, `Undiclosed`, `Series-D`, `$Series A`), and a few have investor names or amounts landing in the Stage column. The Stage correlations are therefore based on only the clean, recognisable stages (Pre-seed through Series F) and understate the true relationship. - **May 2022 has misaligned rows.** Its `Founded` column is text, so some of its rows are misaligned or blank. - **Other columns weren't tested.** I did not compute relationships for Location, Sector, Investor, Founders or What it does. These are free-text or categorical columns, and no numeric correlation applies to them. The tables share no join keys, so these relationships are within-table patterns pooled across the months, not links between tables.