You wonder how the top 100 softcore solo would look without the advantage from paragon?
You have made a great GR time for your paragon level and would like to see how it compares to the others?
You are demotivated to push your maximum limits, because your best time won't rank high enough due to your low paragon?
Well, I have a surprise for you: The unofficial top 100 global softcore solo <<< no paragon >>> leaderboards. Enjoy!
Thanks all for the feedback! I also added an option to manually trigger rankings refresh (note that if the data from the blizzard api has not been updated, you might still not see the new leaderboards positions, which are shown ingame).
@Mad_Tom: It should have been the GR time, yes.
@slaugther: Solo-only leaderboard you have in mind? I like the idea, but even if doable at some point in the future, it would be full with low times, because atm nobody who plays competitively is doing it solo-only.
@Desolacer: I might add HC for future seasons, if the paragon problem stays unsolved and the xp/hour reach this of SC atm.
on the wizard solo leaderboard there is in the top 10 a para 800ish guy cleared 69 in 1:28. This is impossible. Seems bugged. Just for the record
edit: it also seems to use the current paragon lvl as indicator. and not the paragon lvl with which the grift was completed (in some cases this is kinda not "fair"). in the game the paragon lvl of the clear is snapshotted
@horrax1: It was a bug, his time was 887, only 8 applied and the data from api was cut after. The paragon level when the GR clear is made is always taken, don't worry about that.
Overall pretty good stuff, I like it a lot and hope it will get more refined in the future. Most of the rankings correspond very well to my personal expectations. To add a little to the discussion:
I think the calculations slightly favor extremely high as well as extremely low paragon, and also extremely low tiers. Looking at the DH leaderboards, I'd say that the current ranks 2-4 show this quite nicely. I think R4 (can't use the Chinese character names on Dfans) with GR 73 / 14:27 / P1254 should be higher ranked than Fadestep with GR 76 / 14:54 / P2467 or Bernkastel with GR 75 / 14:48 / P2053, and Bernkastel should be slightly higher than Fadestep. For DHs I usually assume ~200-350 (obviously more the higher you go) more paragons will make you clear the next tier with the exact same gameplay, the toughness issues should be getting resolved by the additional armor alone more or less.
An example for the other extreme is the rank 21 (Chinese) with GR67 / 10:21 / P796. It's a sick time and low paragon but I think it's still slightly overvalued. Monsters have so little hp at some point that you can blow up just about any monster type and density so quickly that you just need a somewhat decent rift and nothing else, e.g. I tried one rift on ns just for fun and got this Silver Spire with like 8 min time advantage on 68: http://puu.sh/m5wwV/6237c27bb6.jpg
On season with Stricken this could have been an easy sub 10 min clear, which would rank me pretty high already for very little effort (my items are somewhat bad on ns so you could say I was running maybe the equivalent of P1000). Also if you check Gaby's video of his P650 tier 77 run you can see that extremely low paragons can push slightly higher than what the calculations assume right (even on paragon dependant classes like Barb), mostly because you don't need that much more damage to beat higher tiers since you can benefit more from huge AoE situations and Stricken on RG (obviously not counting Gaby's ridiculously high gems, but he made a right point with this video).
Another example are ranks 13-15. GR 72 / 13:22 / P1811 should be ranked slightly higher than GR 74 / 14:41 / P2373. The time difference alone would more or less make up for 1 tier difference, so we could assume the GR 72 could have been a 73 with the same time (14:41). This would only leave the paragon difference of 562, which should be around 1.5 tiers at this point. The same goes for GR 70 / 13:20 / P993 which should be slightly above the other two, not below.
I think it might have something to do with the coefficient you use for the different classes, you have this part were it calculates 0.8 - 5 * (Max P - Min P - 1000) / 12000 and if I interpret this right you chose the 12000 as a somewhat of a "baseline" mainstat, even though this might not be the same for every class (some use mainstat on jewelry or mainstat gems instead of diamonds etc.), e.g. my solo character has only 10800 mainstat at P1360. I don't know how to properly translate this into the calculations, but it's something you could begin looking into to optimize it further.
I hope these insights help a little, keep up the good work
@wudijo: Thanks for the feedback man! Great write up! I knew something was wrong with certain extremes and after reading your post, I checked the math again to discover something overlooked. I have applied this COEF to the main stat ratio, not the paragon itself. Now this is changed and some of your predictions are illustrated with the Chinese guy #5929 going from #4 -> #2 for example.
Thanks again for helping creating meaningful leaderboards! And Merry Christmas everyone!
Just looked into the updated leaderboards and it seems much better now, also checked the other examples that I listed and those also seem much more accurate after the change. I guess eventually you want to min/max it a little more for every class depending on what items are most commonly used (e.g. 2h vs. dual wield, mainstat on jewelry, mainstat vs. allres gems etc.) in those specific builds. I don't know if stuff like gems could also be included eventually, but if you can draw the data for that it should be quite easy (just need to look into Stricken a little more closely and estimate a mean gain per rank). A column like "avg gem level" would be neat as well (obviously those capping out at 50 can be left out of the equation).
@wudijo: Yep, it could be eventually done, but as Drahque pointed out, gem levels and gear are left out of the equation atm. But also there is more gain from pylons at the lower tiers and less lag, so these somewhat even out. Don't forget that we can't take something so flawed to its core (ranking system based on paragons) and create something perfect out of it. You have to be a magician to do that lol.
@Kakaru: Hi, Kakaru and welcome to the forums! You spotted well there - scores go well above 7k. But the #100 is always 1. The reason is that I simply display the difference. No need to display 7001 for #100 for example -> it could be just 1 and the scores of the others gets reduced by 7k. Its more reading friendly.
@Justice_One: As far as I know they can buy xp buff equal to that from pools, but it can't stack with pools itself (also if it there was big advantage a Chinese would have been the most efficient paragon grinder and that is not so). So they are legit for the standard leaderboards based on this info. And for the score leaderboards - they are paragon free, remember? Bots/xp buffs give close to zero advantage there.
I guess eventually you want to min/max it a little more for every class depending on what items are most commonly used
I don't think min/max means what you think it means. It is only ever meaningful in a game theoretic setting, where it is a criterion for decision which maximizes the minimum payoff you'll get, whatever your opponent's strategy is. It is a very conservative criterion, too, in that you may miss big benefits to avoid an extermely unlikely outcome.
This is clearly not the case here. Min/max cannot be used here. What you meant was "optimize", or "refine".
I know it is off-topic but this abuse of language needs to stop, I've seen it everywhere in the diablo community, but it has absolutely nothing to do there. Help spread correct terminology!
[On-topic]
Great post OP. I'd like to see your exact formula in the main page in a transparent way so we can contribute to improve upon it and min-max it a bit. (haha, kidding with that last one)
You wonder how the top 100 softcore solo would look without the advantage from paragon?
You have made a great GR time for your paragon level and would like to see how it compares to the others?
You are demotivated to push your maximum limits, because your best time won't rank high enough due to your low paragon?
Well, I have a surprise for you: The unofficial top 100 global softcore solo <<< no paragon >>> leaderboards. Enjoy!
neat
Great job, paragon 1000+ is still High to me :')
I cleared a GR70 in S4 with my WD at around P880 and I am not on the list (there are P964 on the list that cleared GR70).
When was the last refresh? I cleared this 4 days ago.
EDIT: Aha, my clear time might have been to bad... 13m:44s
My D3 armory
My Youtube Channel
Great work! Greetings to the author
Thanks all for the feedback! I also added an option to manually trigger rankings refresh (note that if the data from the blizzard api has not been updated, you might still not see the new leaderboards positions, which are shown ingame).
@Mad_Tom: It should have been the GR time, yes.
@slaugther: Solo-only leaderboard you have in mind? I like the idea, but even if doable at some point in the future, it would be full with low times, because atm nobody who plays competitively is doing it solo-only.
@Desolacer: I might add HC for future seasons, if the paragon problem stays unsolved and the xp/hour reach this of SC atm.
@giovax: Yep, so it is, based on 17% mob health increase and pylon effects.
on the wizard solo leaderboard there is in the top 10 a para 800ish guy cleared 69 in 1:28. This is impossible. Seems bugged. Just for the record
edit: it also seems to use the current paragon lvl as indicator. and not the paragon lvl with which the grift was completed (in some cases this is kinda not "fair"). in the game the paragon lvl of the clear is snapshotted
@horrax1: It was a bug, his time was 887, only 8 applied and the data from api was cut after. The paragon level when the GR clear is made is always taken, don't worry about that.
Is this both EU & US?
Season 4: US, EU, KR/TW, China
Era 4: US, EU, KR/TW. No China atm, because their api is bugged for era 4.
Thanks OP, this is pretty cool.
Overall pretty good stuff, I like it a lot and hope it will get more refined in the future. Most of the rankings correspond very well to my personal expectations. To add a little to the discussion:
I think the calculations slightly favor extremely high as well as extremely low paragon, and also extremely low tiers. Looking at the DH leaderboards, I'd say that the current ranks 2-4 show this quite nicely. I think R4 (can't use the Chinese character names on Dfans) with GR 73 / 14:27 / P1254 should be higher ranked than Fadestep with GR 76 / 14:54 / P2467 or Bernkastel with GR 75 / 14:48 / P2053, and Bernkastel should be slightly higher than Fadestep. For DHs I usually assume ~200-350 (obviously more the higher you go) more paragons will make you clear the next tier with the exact same gameplay, the toughness issues should be getting resolved by the additional armor alone more or less.
An example for the other extreme is the rank 21 (Chinese) with GR67 / 10:21 / P796. It's a sick time and low paragon but I think it's still slightly overvalued. Monsters have so little hp at some point that you can blow up just about any monster type and density so quickly that you just need a somewhat decent rift and nothing else, e.g. I tried one rift on ns just for fun and got this Silver Spire with like 8 min time advantage on 68: http://puu.sh/m5wwV/6237c27bb6.jpg
On season with Stricken this could have been an easy sub 10 min clear, which would rank me pretty high already for very little effort (my items are somewhat bad on ns so you could say I was running maybe the equivalent of P1000). Also if you check Gaby's video of his P650 tier 77 run you can see that extremely low paragons can push slightly higher than what the calculations assume right (even on paragon dependant classes like Barb), mostly because you don't need that much more damage to beat higher tiers since you can benefit more from huge AoE situations and Stricken on RG (obviously not counting Gaby's ridiculously high gems, but he made a right point with this video).
Another example are ranks 13-15. GR 72 / 13:22 / P1811 should be ranked slightly higher than GR 74 / 14:41 / P2373. The time difference alone would more or less make up for 1 tier difference, so we could assume the GR 72 could have been a 73 with the same time (14:41). This would only leave the paragon difference of 562, which should be around 1.5 tiers at this point. The same goes for GR 70 / 13:20 / P993 which should be slightly above the other two, not below.
I think it might have something to do with the coefficient you use for the different classes, you have this part were it calculates 0.8 - 5 * (Max P - Min P - 1000) / 12000 and if I interpret this right you chose the 12000 as a somewhat of a "baseline" mainstat, even though this might not be the same for every class (some use mainstat on jewelry or mainstat gems instead of diamonds etc.), e.g. my solo character has only 10800 mainstat at P1360. I don't know how to properly translate this into the calculations, but it's something you could begin looking into to optimize it further.
I hope these insights help a little, keep up the good work
Profile | Twitch | Youtube | Twitter | Recruitment: Zero Empathy are looking for the best players in Europe
@wudijo: Thanks for the feedback man! Great write up! I knew something was wrong with certain extremes and after reading your post, I checked the math again to discover something overlooked. I have applied this COEF to the main stat ratio, not the paragon itself. Now this is changed and some of your predictions are illustrated with the Chinese guy #5929 going from #4 -> #2 for example.
Thanks again for helping creating meaningful leaderboards! And Merry Christmas everyone!
Just looked into the updated leaderboards and it seems much better now, also checked the other examples that I listed and those also seem much more accurate after the change. I guess eventually you want to min/max it a little more for every class depending on what items are most commonly used (e.g. 2h vs. dual wield, mainstat on jewelry, mainstat vs. allres gems etc.) in those specific builds. I don't know if stuff like gems could also be included eventually, but if you can draw the data for that it should be quite easy (just need to look into Stricken a little more closely and estimate a mean gain per rank). A column like "avg gem level" would be neat as well (obviously those capping out at 50 can be left out of the equation).
Profile | Twitch | Youtube | Twitter | Recruitment: Zero Empathy are looking for the best players in Europe
@wudijo: Yep, it could be eventually done, but as Drahque pointed out, gem levels and gear are left out of the equation atm. But also there is more gain from pylons at the lower tiers and less lag, so these somewhat even out. Don't forget that we can't take something so flawed to its core (ranking system based on paragons) and create something perfect out of it. You have to be a magician to do that lol.
@Kakaru: Hi, Kakaru and welcome to the forums! You spotted well there - scores go well above 7k. But the #100 is always 1. The reason is that I simply display the difference. No need to display 7001 for #100 for example -> it could be just 1 and the scores of the others gets reduced by 7k. Its more reading friendly.
Shouldn't the chinese be excluded? They have xp boosts, which is why their paragon is so high.
@Justice_One: As far as I know they can buy xp buff equal to that from pools, but it can't stack with pools itself (also if it there was big advantage a Chinese would have been the most efficient paragon grinder and that is not so). So they are legit for the standard leaderboards based on this info. And for the score leaderboards - they are paragon free, remember? Bots/xp buffs give close to zero advantage there.
Do they lose the $$ potion if they die like we do?
I don't think min/max means what you think it means. It is only ever meaningful in a game theoretic setting, where it is a criterion for decision which maximizes the minimum payoff you'll get, whatever your opponent's strategy is. It is a very conservative criterion, too, in that you may miss big benefits to avoid an extermely unlikely outcome.
This is clearly not the case here. Min/max cannot be used here. What you meant was "optimize", or "refine".
I know it is off-topic but this abuse of language needs to stop, I've seen it everywhere in the diablo community, but it has absolutely nothing to do there. Help spread correct terminology!
[On-topic]
Great post OP. I'd like to see your exact formula in the main page in a transparent way so we can contribute to improve upon it and min-max it a bit. (haha, kidding with that last one)