What is the shortest method to get most gold pieces (gp)

All from Shieldhaven Prison

Office 97 MS likes to change formats just to make older versions obsolete, even when doing the same thing.

The last sheet didn't work under OpenOffice3, but did under v4. OpenOffice is free, and I'm sure it will work fine once I get used to it. I didn't try Libre Calc, but it probably would too.
This is a definite improvement over the previous logging methods. The only drawback compared to mining-log is not being able to run different adventures and segregate the data afterwards. The big advantages are that it's quicker and less work.

 

Here's Veza's data from the level 5 adventure "Rescue From Shieldhaven Prison":

Code:

All							
Lv	C	U	R	E	L	Tot	%
2	18	0	0	0	0	18	3,1 %
3	85	16	4	1	0	106	18,4 %
4	54	19	4	0	0	77	13,4 %
5	101	40	1	0	0	142	24,7 %
6	163	35	9	3	1	211	36,6 %
7	14	4	3	0	0	21	3,6 %
8	1	0	0	0	0	1	0,2 %
Tot	436	114	21	4	1	576	
%	75,7 %	19,8 %	3,6 %	0,7 %	0,2 %		
							
Treasure							
Lv	C	U	R	E	L	Tot	
2	6	0	0	0	0	6	
3	0	0	0	0	0	0	
4	0	0	0	0	0	0	
5	48	16	1	0	0	65	
6	0	0	0	0	0	0	
7	0	0	0	0	0	0	
8	0	0	0	0	0	0	
Tot	54	16	1	0	0	71	12,3 %

 

Sorry, it seems I do happen to use one feature which is not available in Office 97 (referring to other worksheets). It's pretty essential for the tool so I'm regretfully not able to save it as that Excel version.

 

No problem, Jarmo. I'm happy to see it already works with free software too.

 

Here's the level 5 numbers combined with the earlier short run from here:

Code:

All							
Lv	C	U	R	E	L	Tot	
2	25	0	0	0	0	25	3,2 %
3	122	17	6	1	0	146	18,8 %
4	68	26	4	0	0	98	12,6 %
5	134	51	2	0	0	187	24,1 %
6	223	48	12	5	1	289	37,2 %
7	18	8	3	0	0	29	3,7 %
8	2	0	0	0	0	2	0,3 %
Tot	592	150	27	6	1	776	
	76,3 %	19,3 %	3,5 %	0,8 %	0,1 %		
							
Treasure							
Lv	C	U	R	E	L	Tot	
2	8	0	0	0	0	8	
3	4	0	0	0	0	4	
4	0	0	0	0	0	0	
5	65	20	2	0	0	87	
6	0	0	0	0	0	0	
7	0	0	0	0	0	0	
8	0	0	0	0	0	0	
Tot	77	20	2	0	0	99	12,8 %

 

1044 drops from L10 - Garnet Demon Portal

 

Thanks once again, Veza! Here's the data in the format we have grown to know and love (yeah, right):

Code:

Level 10 adventure "Garnet Demon Portal"							
All							
Lv	C	U	R	E	L	Tot	%
7	10	3	0	0	0	13	1,2 %
8	29	4	2	0	0	35	3,4 %
9	515	145	34	6	2	702	67,2 %
10	105	19	3	0	0	127	12,2 %
11	59	27	6	0	0	92	8,8 %
12	38	12	5	0	0	55	5,3 %
13	16	4	0	0	0	20	1,9 %
Tot	772	214	50	6	2	1044	
%	73,9 %	20,5 %	4,8 %	0,6 %	0,2 %		
							
Treasure							
Lv	C	U	R	E	L	Tot	
7	0	0	0	0	0	0	
8	0	0	0	0	0	0	
9	105	25	7	0	2	139	
10	0	0	0	0	0	0	
11	0	0	0	0	0	0	
12	0	0	0	0	0	0	
13	5	3	0	0	0	8	
Tot	110	28	7	0	2	147	14,1 %

 

While the main theme of this thread has been on calculating the drop percentage for specific rarities - great work by the way - as I kept looking at the numbers I couldn't figure out how the level drop was being calculated. I took some time and ran some numbers and these are my results. It was a brief run through and I only hope that my math was right but I think it is. A couple things first. I had to make some assumptions:

1) Rarity is determined before anything else - I believe this has been acknowledged. This means my calculations had to be based off all the drops within a single rarity, naturally I took "common" because it would give me the most data points.
2) This means assumption #2 is that the determination of the item to drop is the same process and level weighting at every rarity, otherwise my calculations would only be applicable for commons, ugh.
3) Based off the way that drops in all the data sets always seem to be skewed toward the levels that have the most items available (6, 9, 12, 15, 18), but also concentrated near the level being played, I assumed that once a rarity was determined that all items within that rarity at the level being played and +/-3 levels were thrown in a single pool, but that each item was given a weighted chance of being drawn based on its level in relation to the level being played. My goal was to calculate that weighting. - This is the biggest assumption and if anyone can confirm or deny this being the process it would greatly increase the ability to calculate these numbers.
4) Weighting would have to be shifted in some way for adventures <Lvl 4 or >Lvl 15 because +/-3 starts to put you outside the existence of items. - I avoided these data sets.

The data:

I did two sets of calculations based off data supplied in this thread. When I have time I might do more.

Level 7 Run, 676 data points:
-3: .014
-2: .258
-1: .2729
0: .2414
+1: .1746
+2: .024
+3: .014

Level 10 Run, 772 data points:
-3: .021
-2: .1216
-1: .3285
0: .2371
+1: .2164
+2: .053
+3: .0224

If my process is correct, the most interesting aspect of this is that it appears we have better odds to get a drop from one level down from what is being played than that level itself.

 

Whoa! Did anyone else notice this is +3 but -4 for the level drops?

 

Ok, ran 4 more data sets. I'm less convinced that my method mimics the actual item drop computation process... but since I did it here's the weighted average for all 6 data sets (just over 5000 data points).


4.9%
9.8%
31.4%
29.9%
19.2%
3.6%
1.1%

 

It's +-4 for treasure only by my reckoning. Normal items is always +-3 levels.

 

I'm not so sure about that. Both of your example levels (7 and 10) have a black hole of a level with a huge number of items below them (6 and 9). This might skew the numbers towards the low side.

If you look at level 8 numbers here you can see the raw drops skew overwhelmingly to the high side as that's where the black hole level 9 is. What weightings does your calculation show for the level 8 numbers?

 

The whole point of my process is to take into account that certain levels have more items associated with them (see my "assumption #3) and figure out the weighting apart from that aspect. That said, as I mentioned in my last post I'm not satisfied with how the numbers are working.

Here's Lvl 8:


15.1%
8.8%
24.2%
25.9%
22.1%
2.5%
1.4%

As you can see the problem is NOT at Lvl 6 (8.8%) or Lvl 9 (22.1%) which have weighting that seems plausible, it's actually at Lvl 5 (15.1%) which to me seems to be an issue.

 

If treasure drops under different rules, then would it seem there has to be an added step in the calculation to determine what drops, a step which says either treasure or non-treasure item? If that's the case it could vary previous conclusions on the best places to farm gold.

 

The L6 kobold mines have a treasure drop rate of about 43%. All other adventures have a treasure drop rate of about 16%, so I believe the gold advantage still goes to the mines.
That being said, I'm still very curious about type and level determination. It might be Rarity, Treasure, then Item - but I don't now how to prove it. (Aside from having Jon or Farbs tell us, which would completely derail this thread.)

 

Yes the mines probably are best, I remember there was talk earlier in the thread that because 50% of the legendaries L5 and below were treasure, that a L2 or L3 farm might produce more gold long-term based on legendaries alone. However if there is a separate step determining treasure or not, that theory goes out the window.

 

1161 drops from L5 - Shieldhaven Prison. My RNG seemed more generous than usual last night.

 

I added to the ~9200 drops above my own ~2500 drops and got the following:
The probabilites of getting an item with the following rarity are, with a 95% probability, within those borders:

Common: 0.7433 0.759
Uncommon: 0.1896 0.204
Rare: 0.03884 0.04616
Epic: 0.00665 0.00994
Legendary: 0.00063 0.00193

You (Jarmo) asked in CH lobby "when could we stop logging our drops?" and mentioned that drop rates could be at most 0.15% off. I'd say that happens (likely) when these 2 numbers for every rarity differ less than 0.0015. This is true only for legendaries atm.

Used data: 8784 (75.1%) commons, 2301 (19.7%) uncommons, 497 (4.25%) rares, 97 (0.83%) epics, 15 (0.13%) legendaries.



Edit: Ah, you meant sth else with that '0.15%' in chat. Well, you can use some other number instead, then
Probably would be better to use a % from the drop % itself for each rarity individually as the highest acceptable error. 20%, for example.
That would mean we are done with commons, uncommons and rares since (maxdrop.common-mindrop.commons) / avedrop.commons = about 2%, it is 7% for uncommons and 17% for rares.
Unfortunately epics and legendaries are way more off (40% and 100%), since we have the least information about them. So we are definetely not ready yet

 

Thanks for the additional info, Feleran!

 

Thanks for the data, Veza! You are very generous to share your efforts so freely with the rest of us. Here's some make-up on your data:

Code:

Level 5 adventure "Rescue From Shieldhaven Prison"							
All							
Lv	C	U	R	E	L	Tot	%
1	0	2	1	1	0	4	0,3 %
2	31	1	0	0	0	32	2,8 %
3	141	19	9	0	0	169	14,6 %
4	125	33	7	2	1	168	14,5 %
5	234	70	9	3	1	317	27,3 %
6	346	69	11	12	1	439	37,8 %
7	18	5	1	1	1	26	2,2 %
8	5	1	0	0	0	6	0,5 %
Tot	900	200	38	19	4	1161	
%	77,5 %	17,2 %	3,3 %	1,6 %	0,3 %		
							
Treasure							
Lv	C	U	R	E	L	Tot	
1	0	2	1	1	0	4	
2	15	0	0	0	0	15	
3	0	0	0	0	0	0	
4	0	0	0	0	0	0	
5	105	30	7	2	0	144	
6	0	0	0	0	0	0	
7	0	0	0	0	0	0	
8	0	0	0	0	0	0	
Tot	120	32	8	3	0	163	14,0 %