Read on Twitter
1/ Finau's win probability starting the day was 30%. Taking into account his score (68, 10th lowest round of the day) – but remaining ignorant of the rest of the field's scores – that probability improved slightly to 32%. In terms of SG, Finau beat his expectation by ~1 shot.
2/ Finau has, on average, over-performed his strokes-gained expectation when in the top 10 entering Sunday (and even more so when leading) throughout his career. Unfortunately, "solid" Sunday performances don't often result in wins.
https://datagolf.com/pressure-tool?dg_id=11676
https://datagolf.com/pressure-tool?dg_id=11676
3/ Funnily enough Finau is actually 10th all-time since 2004 in terms of SG relative to expectation when leading (only 6 data points, however). And in terms of expected wins, he still doesn't make the all-time top 20 worst performers (which includes Rory and Spieth).
4/ Finau's problem has been that he's had too many good-but-not-great Sundays. Restricting to situations with at least a 5% win equity to start the day, Finau gained more than 2 shots over expectation just twice. Many rounds fell in the +0.5-2 range: good, but not good enough.
5/ Also, while your attention is here, the table below has the "luckiest" final round performers since 2004. In the example above, Finau's luck was -0.32 wins. (When we only knew Finau's final round score, and not the rest of the field, we thought he'd win 32% of the time.)
6/ The amazing number in this table is Tiger's "luck" number: 9.8 wins! If you remove this luck component, the wins gap between Tiger and Phil shrinks from 19 to 7.9! Essentially what this means is that players in contention with Tiger played much worse than our model expected.
7/ There are a couple obvious caveats to this. Here is the link to these tables: https://datagolf.com/performance-and-pressure#table-link, with a discussion about the Tiger number after the final table.
8/ To wrap up the thread I'd encourage anyone interested in this stuff to slog through the full blog: https://datagolf.com/performance-and-pressure. These expected win metrics are far from perfect, and aren't particularly useful, but they are super interesting.
