itmeJP Community

Statistics of CoS


(Brugia) #1

As CoS went on and Adam had seemingly a lot of good rolls. I wanted to see if this is actually true or just the common perception bias.
For now I only have recorded data from week 11 to 13, but I might add the earlier weeks, when I have time.

So just one warning I had to use the total of the rolls (dice roll + modifier), since Adams and part of JPs rolls are not visible in the vods.

For this post I will focus purely on combat.
Two notes:
1. Ramus' rolls are not Dans, they are Adams saving throws (+one), since Ramus only attacked (in dnd terms) 1 time (successfully) in the 3 weeks.
2. There might be small discrepancies causing numbers not to add up completely. They might be caused by weirdness in gameplay (Adam attacking NPCs, Ramus attacking the party) or errors in my data

Name	Average	Success  1/20	Number of  Damage   Targeted	Attacked        Damage 
	Roll			Attacks	   Dealt	  	Successfully    Taken
Berg	15.23	69.23%	 1/1	26	   148      21		47,62%	        60
Janus	14.07	64.29%	 0/0	14	   110	    14	        21,43%  	30
Gideon	14.92	57.69%	 1/3	26	   165	    45	        57,78%		198
Adam	13.47	46.67%	 3/4	90	   315					646
Ramus	10.97	46.00%	 0/0	37	   228	     8		12,50%		21

Average Roll - All rolls summed up and divided by the total number of rolls
Success - Percentage of attacks the character did successfully
1/20 - Number of 1s and 20s rolled on a d20
Number of Attacks - Total number of attack rolls
Damage Dealt - Total damage done by the character
Targeted - Times the character was target of an attack
Attacked Successfully - Percentage of attacks against the character that were successful
Damage Taken - The total damage the character took

I think for the most part the numbers speak for themselves. No real Adam bias here. Although I'm interested if it was different in the earlier episodes especially the first party.

Edit: Okay this forum doesn't like preformatted text, let me fix that

(Broxalar) #2

Absolutely fantastic to see that raw data. It'd be kinda interesting to match Adam rolls vs specific players. As we can see Gideon is surely on a hit list but due to the character he's taken the beating well. Thanks for your work on this, it's great to see.

(Unfortunatename) #3

Dice roll + modifier is not very helpful. If someone has a higher modifier then their average will be higher, if they have a lower modifier then the average will be lower etc.

Advantage and disadvantage is another factor that can make it hard to analyze the data since it changes the number but unlike a flat modifier its hard to compensate for.

For instance, Janus has a low number of rolls a high modifier, a lot of rolls with advantage, and took a -5 on top of his bonus's a number of times. How can you make a proper judgment?

If you can factor out the modifiers (and handle advantage separately) then your data could allow you to make a better judgment as to whether people are being luck/unlucky

(Brugia) #4

I agree with you that dice roll + modifier is problematic. The Problem here is that from the Vods I can only get partial data. Only the player macros (but not jps animal forms) have the information I need to get the actual roll. And even there only the d20 rolls have it, damage rolls again are sadly hidden behind their macros.
Because it is problematic I explicitly said it that it was what I was using.

Concerning Advantage disadvantage, I'd thought about how to handle it, could not come up with a decent idea how to handle it and for now I'm essentially ignoring it. My reasoning for that is, that only very few rolls are actually affected by it and I'd assume that for the most part advantage/disadvantage balances itself out. I'd love to hear ideas about how to handle it.

Regarding Janus, I could filter my data handling those shots differently.

Just keep in mind that these stats are not supposed to show who got lucky/unlucky in their rolls. That would be easy to do, if I had the actual dice rolls, but currently I'm unable to do so. It is more of a performance comparison between all players.

Edit: Just to show the problem of the hidden rolls. Of my recorded 330 d20 rolls I only have the unmodified roll of 159. Biggest offender is of course Adam with only 22 of his 166 rolls known.
Just if you were wondering the stats for the actual rolls:

Name       Number of Rolls  Average Roll
Berg	        52	       10.71
DM	        22	        9.45
Gideon	        27	       10.77
Janus	        38	       10.55
Ramus   	20	        9.05
Grand Total    159	       10.30

(Brugia) #5

If I understood you correctly you wanted Adams rolls vs specific players:

	  Average Average				
Name      Attack  Damage	
Berg      13.29	   5.00			
Gideon    14.16	   6.48
Janus     12.50	   7.50
Ramus     10.63	  10.50

Clearly Gideon is winning in getting Adams love.

(Broxalar) #6

Even though Dan gets hit less it seems like he's getting hit harder than most. Thanks for that though! It's fun to see the numbers laid out! Appreciate the work you've put into this!

Although overall I meant a more Adam number of attempted hits and amount missed or succeeded to see who had drawn the most ire.

(BorisIgnatievich) #7

So Adam is just swingy as fuck (lots of 1s and lots of 20s) rather than lucky...

I can totally see why it feels like he's constantly winning though, against a DC of 17 or whatever there is fuck all difference between a 1 and a 12, but the difference between a 19 and 20 is big :adamwizard:

(Unfortunatename) #8

You may be able to guess adam's plus to hit, the average guard has a (+3 to hit monster manual p347) I completely get not wanting to do this but on the other hand the monster manual is pretty.

When it comes to damage are you recording the dicerolled or just the results?
It may be interesting to see damage rolled vs expected damage,

Look a graph:


Look an analysis of Advantage as a mechanic:


The effect of advantage varies based on the target number, the lower a target number is the more advantage increases your chance of sucess (eg DC 10 advantage is "worth" a flat+4 or +5 but by Dc 20 its more like a flat+1).

It will also take a long long time to balance out and certain characters can work to gain advantage more often. If you consider inspiration, players will be rolling with advantage much more often than they will with disadvantage. The rate at which they do varies by player habit and character abilities eg Janus vs Ramus. This means that advantage and disadvantage will not necessarily cancel each other out (its very unlikely).

It's complicated, and will mess up ya numbers!

Edit: This post looks perfect in the edit but the "hide deatails" stuff isn't hidden in the posts, wtf

(Brugia) #9

Yes for some rolls I would be able (always when he rolls a one or a 20) to read of the modifier he uses and I would be able to derive the other rolls from that. Or get the monster manual out, but there is no guarantee that he doesn't modify the monster. Might keep in mind tho, to at least get a higher number of his actual rolls, also i could get JPs rolls that way, since I'm already recording when he shapeshifts, so at least i could do a proper player comparison.

Advantage/Disadvantage, yeah the further they get, the more of a problem it will become (looking at the babarian). Currently out of the 330 rolls only 19 were of either, so a slight problem.
I'll think about it some more probably.

For Damage whenever I can I record the actual dice results, but for damage thats almost never the case (hidden inside the macros). So just results.

Edith: control + enter posts the post, good to know

(Brugia) #10

Well how swingy he is I should be able to tell, even though it will be somewhat hidden behind modifiers. Will look into making that a stat.
And for 1s and 20s, he is behind the average, at least for these 3 episodes. I think he critted more in the earlier ones.
I think it just feels like he is winning, because he usually has more attacks than the players and usually hits someone per round. Or he just randomly (almost) one hits people sometimes.

(Unfortunatename) #11

The raw dice for player damage macros should be easy to get, pg149 PHB has all the weapon damage, we know the player modifiers to damage,

NPC's (guards) are mostly 1d8+1 (longsowrd, crossbow), or 1d6+1 (spear, scimtar).

Of course IF the players get to a higher level this gets harder to do but its a reasonably big IF.

(Unfortunatename) #12

When you look at the stats it looks like Adams success chance is quite low. This is in part down to the player AC and to hit being higher than monster AC and to hit.

Name  Average  Success   1/20  Number of 
      Roll     Rate            Attacks
 Adam  13.47    46.67%	 3/4	90
AC    Example character   Hit on a +3    Hit on a +4
8     Zombie                  80%            85%
12    Tiger Gideon            60%            65%
14    Gideon                  50%            55%
17    Berg                    35%            40%
18    Ramus                   30%            35%

Thinking of last episode, If we take the +3 column and split attacks evenly between Gideon Berg and Ramus we would get (115/300)=38.3% hit rate, if we use tiger Gideon which has a lower AC this rises to 41.7%. However If we target tiger Gideon 50% of the time and split the other 50% evenly between Berg and Ramus we reach 46.3%

But if we look at the actual rates that people were hit (again taking a +3)

Name     Actual      Predicted  Targeted
Berg     47,62%	     35%        21
Ramus    12,50%      30%        8
Gideon   57,78%	     60%        45

We see that with a 1 AC difference Berg has gotten a shit sandwich, Ramus is untouchable and Gideon is a pincushion with a few less pins than he deserves.

We also see that Adam has been going after Gideon an awful lot (nearly half of the time) and his hit rate 46.7% is nearly exactly as expected, 46.3%. Adam's hit rate "should" be worse but by sensible targeting he has bolstered his hit rate quit a bit.

(Brugia) #13

Just a small update with a few numbers of the 'uneventful' session of yesterday. I could work out most of the unmodified rolls this time and could fill in quite a bit of the missing ones in my data from the weeks 11-13 by looking at 1s and 20s, looking up JPs transformed stats and damage rolls of the party. Although in the mixed battles with multiple enemies/attack types I will have to go through the vods again since I did not recorded the type of attack until now. Before I do that I will most likely go to previous weeks so I have a kinda complete data set on the current characters.

Well lets get to the numbers:
Only 78 entries this time instead of the 251 entries of week 13. Well to be expected if only one fights happens, 2 of the hours were mostly roleplay and one hour of character creation.
This time you get all the d20 rolls(Initiative, Skills, Attacks, Saves), since the amount of rolls of just combat was very low.

Week 14
Name	Average Successrate Average 	Number 	  Damage Damage
	Roll		    mod. Roll	of Rolls  Done   Taken
Adam	9.54	58.33%	    13.13	15	  51   
Berg	14.50	60.00%	    16.38	 8	  18      6
Gideon	10.30	75.00%	    13.30	10	  39     21
Ramus 	4.60	0.00%	     8.00	 5	  21     14
Zawaz	10.40	50.00%	    13.20	 5	   8     18

I added the average roll, which is the raw d20 result in and kept the average modified roll from before. So you can compare those two.
Also keep in mind that not all d20 rolls can be quantified as a success or fail, Initiative for example. I did throw them out for calculating the successrate.

Something notable is that Ramus seems to have been kinda unsuccessful this session, even if you look at the 3 saving throws he caused, only one of them was unsuccessful. But still he managed to get 2 of the 6 Killing Blows and deal more damage then Berg and Zawaz.
Noteable as well is Bergs very high average roll and Gideon successrate, which would have been even higher if I added all those extra claw attacks he did on all the dead Kobolds.
And at last Berg is leading the friendly fire category now.

(Xaphere) #14

Week 14 spoiler:
Can we get stats for Snidges? I feel like there was some RNG magic there.

(TheDesec) #15

I think this was week 15. either this, or last weeks' q&a (14) was named wrong.

do you want mr. torbidgers' attribute stats? because there's not really and action stats to be had, at zero rolls! I think it was
15 STR, 16 DEX, 15 CON, 10 INT, 13 WIS, 15 CHA, 12 HP, 16 AC
but I am absolutely not sure!

(Brugia) #16

week 14 spoiler below

Well not sure if a spoiler is needed since, my stats on week 14 already spoil a bit, but anyways.
I don't have the math on hand anymore to give you an exact number and I'd have to read into it to do it again. For 4d6k3 the average score you get is ca. 12.25, and if you just compare Berg and Snidges. Berg wins with being less likely to exist by far.

(Raorou) #17

After seeing the stats of JPs character, at first I was like "Adam really hates JP" but then remembered Gideon is a Druid of the Moon and it became "Holy shit, JP is a really good tank." Turns out, giant animals are real good at getting peoples' attention. :laughing:

Good work on the data collection and number crunching. Looking forward to more. :+1:

(TwoToneTerran) #18

Casters tend to be the best classes in the game when you have poor stats, Druids especially.

(Brugia) #19

So it's time again. I have finally finished all weeks of season 2 and today you will get a basic overview of most of the numbers. When I have time I will compartmentalize the stats so there is a more detailed look at the different kind of rolls e.g. all the different d20s, bless, Advantage/Disadvantage, damage vs expected damage and take a look at how the stats differed in the weeks to find more outliers like Adam in week 9, because having an average of above 13(edit: 14 was the wrong number, sorry) in your attack rolls is not normal.

Just a small note on the data: I was able to derive most of the actual rolls through JPs hovering over rolls or rolls being 1, or 20, or 2 and 19. If neither of those two worked I looked up the stats in hopes Adam has not changed them, so it should be correct for the most part.
Out of 890 entries 33 are partial, like only having the total end result of the roll (multi dice roll or monsters I don't have access to), missing die size for spears (could be 6 or 8) or missing classification of "random" rolls from Adam. As well as 24 skill rolls for which I could not determine success or fail, partially because not every skill roll is either or.

Now the overview with a legend below:

Week 9-14
Name	Average Std.   Average  Success	1/20    Roll	Damage 	Damage  Killing 
        Roll    Dev.   Modifier                 count   Done    Taken   Blows
Adam	10.41	5.67	2.45	47.01%	17/10	279/13	519	1037	13
Berg	11.22	5.70	3.22	68.75%	 1/2	 60/2	168	  62	 6
Gideon	10.63	5.68	2.94	58.33% 	 5/9	 87/10	207	 248	10
Janus	11.10	5.88	4.90	75.00%	 5/5	 86/4	279	  86	16
Ramus	 9.55	5.18	2.61	58.54%	 5/5	 44/56	308	  46	17
Wester	10.91	5.92	3.15	41.38%	 1/3	 34/0	 80	  61	 2
Zawaz	10.40	8.59	2.80	50.00%	 1/1	  5/0	  8	  18	 0

Average Roll - Average of all d20 rolls
Std. Dev. - Standard Deviation of the d20 rolls. A d20 has a SD of 5.77
Average Modifier - Average of the modifier added to all the d20 rolls
Success - Rate of success of all attack, spells, saves and part of the skill rolls
1/20 - How many 1s and 20s on a d20 were rolled
Rollcount - How many rolls were done by the character/how many rolls the character directly caused e.g. Save against Sacred Flame
Damage Done - Sum of the damage done to other characters
Damage Taken - Sum of the damage taken
Killing Blows - How many NPCs were killed or for Adam how often he knocked an character unconscious or killed them

Now a few notable things from the stats:
* Berg is the best roller of them all and Ramus is his counterweight on the other end, good thing he lets Adam do most of his rolling.
* Zawaz with his few rolls is almost as swingy as you can get.
* Janus' good attributes and resulting modifiers to his good rolls probably helped him quite a bit at giving him an edge over the rest.
* Gideon has almost as many crits as Adam with a third of the rolls
* Ramus could only overtake Janus in damage done after the combat heavy week 13 and will be leading the category most likely until he dies.
* The cleric of the group has more kills then the rest of the current group summed up.

(BorisIgnatievich) #20

Outstanding work, must have been one hell of an effort :itmejp10:

Adam rolling less 20s and more 1s than you'd expect (14 of each for 280 rolls) is counter to the impression everyone seems to have. which is, of course, why you look at the numbers

Saying that though, Gideon does seem to either miss or crit, which might be why the crits and killing blows are almost identical :itmejplol: