Using that 59-60 season, look at the actual stats (
here) - you have 10 goalies of significance, with Riggin Rollins and perhaps Paille being too small a sample. Is that Sawchuk season anything to write home about? He's basically league average that year. He's still 4th in save percentage and 3rd in GAA, because again, 6 teams. He received a bunch of 1st half votes for the all-star team because he had an great October, but it's 10 games, and he has 5 competitors. The same goes for Lumley and Worsley, neither of whom had good years. They both still finished in the top 10 in save percentage though.
Lumley and Worsley are guys I'm really skeptical of - they both had 2 good years, and not much else. I see them somewhere on the spectrum between Varlamov and Quick, to use modern equivalents. Nobody gives those two guys much credit for being the 15th best goaltender in the league in some of their more average years. Sawchuk less so, because of his early peak, but during the save percentage era, he basically spent a decade being league average. If Terry Sawchuk retired after 54-55, and Serry Tawchuk played in the NHL from 55-56 through 69-70, Serry's not making it in the Hall of Fame based on his performance during those years. In fact, Serry Tawchuk's performance in those ~600 games, depending on what years you cut off at the end due to partial playing time, is basically equivalent to Meloche or Peeters in their careers, roughly league average by GA%-, over about the same number of games.
Your premise almost becomes a tautology - the goaltending is great because all these goaltenders are Hall of Famers - Hall of Famers are obviously great goaltenders.
It just ignores the fact that a) making the Hall of Fame isn't reflective of your performance every year, and b) Hall of Famers can have bad years too. (Also c) the Hall of Fame's track record with player election isn't the greatest.)
With regards to percentiles, you'd expect in 2 separate samples of similar sizes, that seasons in each percentile would be roughly equivalent to each other. So if you have say 200 goaltender seasons, and divide them in half randomly, you'd expect that the top 10 seasons in each sample would be the same in both samples, seasons 11-20 would be the same, and so on. There's a fudge factor at the top and bottom of each percentile, but you're not going to get a result where the 50th best season in one sample is better than the 20th best season in the other.
That's basically what you're trying to say, that because all these Hall of Famers make up the majority of this 108 player-season sample, that the 50th best season is better than the 20th best season of a different 100 player-season sample, whether that be in the 80s, 90s, or today. The other aspect of that, is because we have 30+ teams now, the sample fills up much quicker. Here's a comparison just by GAA, of goaltenders playing 50+ games, the first sample being 1917-18 through 66-67, the second sample being 2005-06 through 23-24.
GAA Under | 66-67 | 23-24 |
2 | 9 | 4 |
2.1 | 12 | 13 |
2.2 | 20 | 26 |
2.3 | 24 | 61 |
2.4 | 30 | 108 |
2.5 | 42 | 153 |
2.6 | 50 | 191 |
2.7 | 61 | 225 |
2.8 | 65 | 255 |
2.9 | 71 | 283 |
3 | 79 | 308 |
3.1 | 84 | 324 |
3.2 | 90 | 332 |
3.3 | 95 | 338 |
3.4 | 99 | 344 |
3.5 | 100 | 345 |
Total | 108 | 349 |
There are 108 total seasons in the first sample, there are 349 total seasons in the second sample. The top end runs close together, there's 41 player-seasons 2.2 and under pre-expansion, 43 post-lockout, so you're going to end up with a near 50/50 balance if you were to rank the best 100 or so seasons, but look at how post-lockout fills in. You add another 35 player-seasons between 2.2 and 2.3, while pre-expansion needs to expand out to 2.7 before it picks up another 35 player-seasons, and then post-lockout adds another 47 seasons to match the 108 of the pre-expansion total. You rank those 216 seasons, and the last 50-70 seasons are all going to be pre-expansion. Cut it down to the top 125, and you're going to have 30-40 pre-expansion, and 85-95 post-lockout.
Also, this shows the power of controlling the sample size. Setting it at 50 excludes everything prior to WW2, including all the low scoring 20s and 30s goalies, who played 44 or 48 games. You have 108 player-seasons of 50 games, 162 of 48 games, 214 of 44 games, and 235 of 40 games. The equivalent for post-lockout is 349 player-seasons of 50, 381 of 48, 444 of 44, and 533 of 40. The GAA sample changes from 9-4 under 2, to 49-11 under 2. The low-scoring era means that of the 235 player-seasons in the pre-expansion sample, 107 were under 2.4, akin to the 108 the post-lockout sample had. However, reducing the games played to 40 there boosts it to 149 player-seasons, with 211 being under 2.5, and 270 under 2.6, while a similar 92-87 breakdown exists for player-seasons under 2.3. So without normalizing for scoring, the top 200 or so seasons breakdown 50/50ish, but go out to 500, and again the last 90 or so are all going to be pre-expansion, because all 270 of the post-lockout sample is under 2.6, while the last 90 of the pre-expansion sample are above 2.6.
No matter what you do, you're going to run out of great seasons pre-expansion before you run out of great seasons post-lockout.
[I also have some breakdowns on that initial sample of 108 post-lockout player-seasons - they cover every non-shortened year (no 12-13, 19-20, or 20-21), involve 28 of the 32 teams, and 50 different goaltenders.]