Questions based off the (Statistics) Data Sheets

knell · May 2019

Since I won't be able to participate in the Discord Q&A tomorrow, I thought maybe I can get lucky and have them answered here - if not, I wonder if some of the community members are kind enough to shoot some of these questions tomorrow during the Q&A - I'd appreciate it!

@Peanits

1. Thanks for collecting the data from right before rank resets; that was a great idea! But is it possible to combine data from previous months also so we could have a larger sample size? (e.g. - Jan. 6-12 + Feb 4-10 + March 4-10 + Apr 1-7 etc.)

2. There were several requests for the data to not include trials that had DCs in it. Were you able to acquiesce to these requests (especially for 'Party Size Survival Rate' and 'Highest and Lowest Kill Rates')? If DC-ed matches had to be included in these samples for this time around, were the DCs counted as Kills/Escapes or just taken out (as neither)?

3. For the 'Party Size Survival Rate', did you combine both "2 Solos and 1 Duo Team" and "2 Duo Teams" together under "2 Players"? Is it possible to separate them?

4. Similarly, did you combine 4-Solo players, 2-Solo + 1-Duo, 1 Solo + 1 Trio all together as being "SOLO" in 'Party Size Survival Rate'? Is it possible to separate them?

5. For the Average Game Length, what is causing that dip in all platforms at Rank 2?

6. Is there a reason why 'Highest and Lowest Kill Rates' and 'Party Size Survival Rate' were not divided by Ranks also? It would have been especially interesting if we could have seen the same statistics for 'Party Size Survival Rate' for a.) Highest Rank of SWF and b.) Average Rank of SWF.

7. I find it interesting that in the 'Highest and Lowest Kill Rates', Legion is depicted as having one of the lowest kill rates on PC. I assume this is still with the Moonwalking exploit in effect. I wonder with the recent changes, less player will be using them. Can you show us which killers are most often used in percentages?

knell · May 2019

So how was the Q&A? Does anyone have a link to the video of the Q&A session? It doesn't seem to be on youtube yet.

Also, did any of my questions above get answered? I personally would especially like to have had my concern about DC-ed matches being included/excluded as part of the samples in 'Party size Survival Rate' and 'Highest and Lowest Kill Rates' answered. If I remember correctly, there was a time when the Data Team not only included matches with DCs as part of their samples, but had even counted Survivor DCs as 'kills,' which destroyed my confidence in the information provided back then. Does anyone know if that has been addressed and corrected this time around?

Peanits · May 2019

I'll do short answers to spare the word wall:

1: It would be possible, but it would take a lot longer to do since you would have to gather data for each date range and then average them together, and it would also be less accurate as the meta changes and updates shift things around.

2: This isn't something we did this time around but it's something we could potentially look into in the future. This gets muddy though since someone could disconnect when they would be dead anyway. People will also disconnect when things aren't going their way, so if you exclude all disconnects, you'd likely end up with a "perfect scenario" set of data.

3: That data is based on the amount of players in a party of that size alone, not the odds of any two group sizes appearing together. Basically of all matches played and between all the survivors in those matches, about 28.62% (on PC) were playing a two man party, for example.

4: See the above. Basically it's all individual, it's not per match.

5: This data was from the period of time where it was very difficult to rank up as a survivor. On top of that, these stats take your rank after a match, not before. So in short, usually if a survivor died at this time, they would derank, and as a result it would count as a death towards rank 2, not rank 1. This made the survival rate higher for rank 1 and lower for rank 2. It's a bit of an oversight, basically.

6: It would basically just be a ton of data and we had a very limited amount of time to get it all set up. We could maybe do that next time, but we'd need to plan it ahead a little more.

7: I don't have the exact stats on hand at the moment, but we could look into that next time.

RE: The Q&A: It was just a Discord Q&A, it was not a stream.

knell · May 2019

@Peanits

Thanks for the quick response!

For Question #2, I'd think that removing matches where people DCs in a situation where they are about to die anyway from the sample size shouldn't affect the statistical results much if there are enough trials. People disconnecting when things are not going their way, however, WOULD affect the results, since it would have made these matches much harder for all the other survivors (an automatic 3 vs 1 situation), skewing the stats in favor of killers, making it seems as if the killers are more powerful than they actually are.

The idea I'm getting at is that you wouldn't want to balance a game based on a data where people are disconnecting, but rather, should be balanced on what you describe as being "a perfect scenario." Isn't that why you are removing 3 vs 1 matches in the latest patch - because the game isn't balanced around those scenarios?

So if the matches that include DCs were part of the sample... I'm afraid to ask, what were survivor DCs counted as? (Please don't say 'kills.' Please don't say 'kills.' PLEEEEEEASE don't say 'kills.')

PS - Yeah, I don't know what Discord is. Wish I could have seen it though! Is there anywhere I can get the info that was shared during that time, or is it one of those, 'you had to be there' type of situations?

ba_tetsuo · May 2019

I'd personally like to see time it takes to finish the 5th gen vs full game time, as well as swf survival rates per killer

Questions based off the (Statistics) Data Sheets

Comments

Official Dead by Daylight Channels