Leandro

Raid is starting in 45 minutes, if you want to help, come to the Hive. Leechers need not apply! We are looking for a bunch of empaths, some tanks, some kins, some melee. Because it's Superbowl night there aren't that many people online, so we're probably going to take a little longer than usual, but we expect to be done by 10:30PM EST the latest.

Updated the 30-34 table; we now have 1235 rolls. LOTG 7.5 seems quite "common" at either a 4x or 4.5 weight (with a 1:41 chance based on those rolls), Miracle: Recov is sitting in the 2.5x weight (1:72 chance) and Numina Regen/Recov looks like a 2x weight (1:95 chance). High margin of error, data is likely to change, etc etc.

Updated the table with those rolls.

Thanks Pum! We filled this range very nicely. I think it's time to move on to 30-34.

Let's continue this range a bit more until Unbounded Leap: +Stealth and Theft of Essence: Acc/End/Rech have at least 10 rolls each. Then move to another range until the same condition is met for each range: 10 rolls in the unluckiest recipe.

We're now at 3356 rolls in the 25-29 range, and we're still showing a 10x trend in here. Very gradual weight increment, too. I still find the data valuable because I think most people want to know the actual chance for a recipe to drop during a roll (LOTG 7.5 is about 1:50 in this range), but it's kinda killing the idea to reverse-get the actual weights.

We now have 2737 rolls on the 25-29 table. Still suffering on the bottom end, those recipes just don't want to drop.

We're sticking to 25-29 until the bottom recipes (the ones that have been rolled the least) have been rolled at least 10 times each. That should increase the confidence in those numbers a lot.

Thanks to Minotaur and Sardan we have 2566 rolls in the 25-29 range. Still a lot of uncertainty at the bottom of the table, but the top shows some clear "winners" in that table.

724 rolls from PumBumbler now in the 46-50 range. Interesting bit: "Expedient Reinforcement: Resist Bonus Aura for Pets" doesn't have the "(Recipe)" text in the recipe.

I added another 150 rolls to the level 10-14 section. I think we have identified two recipes that have a 3.5x weight for sure: Impeded Swiftness: Chance for Smashing Damage and Eradication: Chance for energy damage. There's a recipe weighted 3.9x and then those two at 3.5x and 3.4x, then a gap until 3.1x. Really looks to me like the more recipes we add, the more that gap widens, which seems to indicate those recipes are going to sit at 3.5x forever.

Theft of Essence: Acc/End/Rech dumped itself back in the 0.5x weight, which really annoys me. That recipe is seriously undeperforming if it's supposed to be a 1x weight.

If you're asking this question, you probably are a squishy (blaster, troller or defender) and you're looking for some way to avoid being held so much. My advice is: slot IO sets with Defense bonuses. They can't hold you if they can't hit you. I have 17.7% ranged defense on my main controller, and the difference is very noticeable. It's saved my neck plenty oftimes.

Level 25-29 is up to 1944 rolls. Thanks, Hell_Jumper!

The more rolls are added, the more it looks like the Positron's Blast's proc is the only 5x of that pool.

The Character Ccopy tool finally let a copy of my main through, so I added another 150 rolls to Level 10-14 to try to push the numbers closer to integers. There are 9 recipes that very stubbornly refuse to get in line for an integer, and I'm crossing my fingers that a truckload more data will show them belonging to one grouping or another. While I work on that, everybody else please follow TopDoc's advice and work on 25-29.

I did some weight comparisons based on the level ranges we already have, and I'm tossing the data here for discussion. Excuse the commas as decimal separators, it's what Latin America uses and I don't feel like fixing it manually right now.

2,6x Basilisk's Gaze: Acc/End/Rech/Hold (10-14)
2,2x Basilisk's Gaze: Acc/End/Rech/Hold (15-19)
3,7x Basilisk's Gaze: Acc/End/Rech/Hold (20-24)
1,5x Basilisk's Gaze: Acc/End/Rech/Hold (25-29)

Three out of four ranges would confirm TopDoc's idea, but the 20-24 range is so close to 4x that the recipe is being either very misrepresented, or we need a lot more data to bring it down to 2x in line with the others.

2,7x Basilisk's Gaze: Chance for -Rech (10-14)
3,1x Basilisk's Gaze: Chance for -Rech (15-19)
2,2x Basilisk's Gaze: Chance for -Rech (20-24)
2,7x Basilisk's Gaze: Chance for -Rech (25-29)

This recipe weights at 3x in three out of four ranges, but again, the 20-24 range has it weighting closer to 2x.

2,7x Blessing of the Zephyr: Knockback Reduction (4 points) (10 14)
1,9x Blessing of the Zephyr: Knockback Reduction (4 points) (15 19)
1,5x Blessing of the Zephyr: Knockback Reduction (4 points) (20 24)
1,7x Blessing of the Zephyr: Knockback Reduction (4 points) (25 29)

I think it's telling that this recipe weights closer to 3x in the 10-14 range (the one we have the more data on) but 2x in later ranges. Most people who know what they're doing will slot the lower level recipe in order to keep the bonus while exemplared down, making the low range recipe weighted higher than others.

2,0x Celerity: +Stealth (15-19)
0,7x Celerity: +Stealth (20-24)
0,5x Celerity: +Stealth (25-29)
0,9x Freebird: +Stealth (15-19)
0,5x Freebird: +Stealth (20-24)
1,2x Freebird: +Stealth (25-29)
0,8x Time & Space Manipulation: +Stealth (15-19)
0,7x Time & Space Manipulation: +Stealth (20-24)
1,5x Time & Space Manipulation: +Stealth (25-29)
1,9x Unbounded Leap: +Stealth (15-19)
1,9x Unbounded Leap: +Stealth (20-24)
0,2x Unbounded Leap: +Stealth (25-29)

Here we see that the low level Celerity and Unbounded Leap weights 2x while the other ranges are most likely 1x (the 25-29 Leap probably is suffering from lack of data misrepresenting it). Since people who PVP are more likely to slot lower level recipes, this would seem to confirm that the recipes are weighted individually in each level range. The Fly and TP stealth procs, less common in PVP, appear to remain 1x all around.

3,7x Salvo: Acc/Dam/End/Range (10-14)
3,6x Salvo: Acc/Dam/End/Range (15-19)
2,5x Salvo: Acc/Dam/End/Range (20-24)
1,8x Salvo: Acc/Dam/End/Range (25-29)

This recipe is interesting because it starts at almost 4x and goes down to 2x gradually, suggesting a 4x-4x-2x-2x progression. And it's not even a proc that becomes more valuable at lower levels.

Based on what I've seen so far, I don't think the weight for all recipes is the same across all level ranges. Certainly there's a lot that keep their weight on all level ranges, but I think that assuming all of them do is a bad idea.

And once again, the lowest-ranked recipes keep trending towars 1.0x. The bottom one jumped from 0.3x to 0.6x, the three above it went to 0.8x, and then two more reached for 0.9x. I think it's safe to say that the 10x range we were getting before was simply due to lack of data.

I think we can clearly call Impervious Skin: End/Rech/Res as the only 5x weight in the 10-14 range, as the next recipe in the table has a 4.1x weight. The rest of the breaks aren't so clear, but you can "kinda" see them. Great job there, Pum.

Pum, I agree with TopDoc that working on the 10-14 range will be the easiest. I would, however, also like to keep gathering data on the other level ranges, as I have the feeling that the number of recipes we'd need to roll for our data to match the internal tables is ridiculously high. I think the "rough" data we have been collecting is already valuable as a "practical guide", and having complete "rough" tables for every level range will be valuable by itself, even if they're not perfect.

So what about this: one of us keeps on rolling 10-14 until the data fits what Synapse told us or we are completely sure that something is out of whack, while the rest keeps on filling the rest of the levels with "practical" data. Sounds good?