How do you guys normalize this data? I've never had the energy to do it, but if I was going to I would have tried something like:
1. Download ppg, apg & gpg for every skater that has done close to a full season. Say 50+ games.
2. Check if same type of distribution can be confirmed for each of the seasons. The real top players should be possible to remove here if it is needed to confirm distribution, since the assumption is that are outliers. If the same type of distribution cannot be confirmed, the approach has failed. If we can confirm the distribution, we can continue. If it is normal, what I describe below should work. If not, we might need to think a bit more.
3. Calculate the average & std for the ppg, apg and gpg per season.
4. Calculate how many std the top player per season is above the average is these specific stats.
This std will then be the answer to who had the best season. If you want to translate it back into points you just translate these numbers to whatever is your reference season.