I used a 5-year period for each guy. For Grove, it was 1928-32. For Koufax, it was 1962-66. The table below has some comparisons:
RSAA means "runs saved above average." It comes from Lee Sinins' Complete Baseball Encyclopedia. The numbers are park adjusted. So Grove has a big lead here, both in total and per 9 IP. I will come back to these numbers later when I plug them into the Pythagorean formula.
Grove was 60% better than average at preventing HRs (that is what the 160 means). He gave up 49 HRs while the average pitcher would have allowed 78 (100*(78/49) is about 160). This gives him a pretty big edge over Koufax. But they are not park adjusted. If they were, Grove would have an even bigger edge. Here are the HR park factors for the Philadelphia A's from 1928-32 from the STATS, Inc. All-Time Baseball Sourcebook: 126, 165, 153, 104, 199 (the 126 means that Shibe gave up 26% more HRs than the average park). Now Shibe Park may have had some asymmetries, so that lefties hit alot more HRs. With Grove more likely to face righties (being a lefty himself), it is possible the park did not hurt him as much as these factors suggest. But A's righties Foxx, Miller and Dykes generally had much higher slugging percentages at home than on the road (from Retrosheet). So my guess is that Grove certainly was not aided by his park in preventing HRs.
Koufax allowed 89 HRs while the league average was 124 and had the following HR park factors in his years: 50, 63, 62, 49, 70 (meaning Dodger Stadium allowed fewer HRs than average). So he was helped quite a bit yet Grove still has the big edge here. He allowed 89 HRs while the league average was 124.
Relative SO/BB is each pitcher's strikeout-to-walk ratio divided by the league average. Grove had a 2.67 strikeout-to-walk ratio while the league average was 0.95. The 2.67/0.95 is multiplied by 100 to get 281. That beats the 225 of Koufax or 100*(4.57/2.03).
The ERA+ comes from Baseball Reference. It is ERA relative to the league average but also adjusted for park effects. Grove only has a slight edge here.
WAR comes from Baseball Reference (and they get it from Sean Smith at Baseball Projections). It is "Wins Above Replacement for Pitchers. A single number that presents the number of wins the player added to the team above what a replacement player (think AAA or AAAA) would add. This value includes defensive support and includes additional value for high leverage situations."
It is not clear to me how Koufax beats Grove here. Grove has alot more RAR or "runs above replacement." It might have something to do with the leverage adjustments. None are made for Grove since the play-by-play data has not been posted at Retrosheet. The WAR and RAR numbers imply that for Grove's years, it took 11 extra runs to win a game (441/40.1 = 11) and only 8.26 for Koufax (347/42).
Baseball Projections says that it normally takes about 10 extra runs to get a win. I wonder if they are using the formula which says it takes 10 times the square root of the number of runs scored per inning by both teams. For Grove's years I calculated that to be 10.7 and for Koufax got 9.54. That would give Grove a WAR of 41.21 (441/10.7) and Koufax 36.37 (347/9.54).
Pitching Runs is "Adjusted Pitching Runs." It comes from Baseball Reference. It is "A set of formulas developed by Gary Gillette, Pete Palmer and others that estimates a pitcher’s total contributions to a team’s runs total via linear weights." Lee Sinins told me it might also be based on decisions, but I am not really sure. Anyway, Grove has a big lead here, too.
Now to come back to RSAA and try to calculate the Pythagorean pct for each guy using RSAA per 9 IP. The AL of 1928-32 averaged 5.12 runs per game (yearly averages weighted by Grove's IP) and 5.12 - 1.98 = 3.14. So if Grove allows 3.14 while his team scored 5.12, he would have a winning pct of .727. Koufax would allow 2.78 while his team would score 4.05 runs per game. That gives him a pct of .679.
One thing I have not mentioned yet or tried to take into account is integration. Last January, I compared Grove's career to Randy Johnson's. See How Might Integration Have Affected The Lefty Grove/Randy Johnson Debate? I tried to estimate how much better the hitters would have been during Grove's time if the percentage of players who were non-white was about the same as during Johnson's. I also tried to adjust for the number of non-white pitchers and non-white fielders. I came up with Grove's ERA going up about 10%. What if I did that here?
Then Grove would allow 3.45 runs per game and his pct would fall to .688. That is still higher than Koufax.
But if we use the adjusted pitching runs, Grove allows 3.32 runs per game (5.12 - 1.8). He would have a pct of .704. Koufax would allow 2.63 runs per game (4.05 - 1.42). He would have a pct of .703. That would make the two about even. Grove would get the edge due to more IP.
But if we raise Grove's runs per game by 10%, to 3.65, his pct would be only .663. That would put Koufax ahead.
Finally, if we knock down Grove's ERA+ from Baseball Reference of 172 by 10%, he would be at 155, below Koufax's 167. The 10% adjustment for integration is just an estimate. It is the same one I used when comparing Grove to Johnson. The % of players and pitchers who were non-whites during Koufax's time was probably lower than during Johnson's time. So adding 10% to Grove's ERA is probably too much. I don't think I know the right adjustment to make. But this gives us some idea of what the effect of integration might be.
If I lowered Grove's strikeouts per 9 IP by 10% from 5.91 to 5.32 and raised his walks per 9 IP from 2.21 to 2.43, his new strikeout-to-walk ratio would be 2.19. That divided by 0.95 would be 2.30. So his relative SO/BB would be 230, still higher than Koufax's 225.
If I raised Grove's HRs by 10%, he would have allowed 54 HRs. Then 78/54 = 1.45. That times 100 is 145. That is still higher than Koufax's relative HR rate of 139.