The second iteration of 2v8 is now LIVE - find out more information here: https://forums.bhvr.com/dead-by-daylight/kb/articles/480-2v8-developer-update

Tableau Dashboards - 2v8 Survivor Statistics (100 Games)

Beaburd
Beaburd Member Posts: 998
edited August 12 in General Discussions

Hey folks! I posted this on Reddit, but I figured I'd include it here since I frequent this forum a lot as well.

I have been learning some different data visualization software lately, primarily Tableau Desktop, and opted to make a personal pet project to put those skills into use and better ingrain what I've learned so far.

To that end, I played 100 games of the new 2v8 mode during and around the free weekend, collected the data from each game, then made and connected a few interactive Tableau dashboards to better make sense of the data. I also included some basic insights of my own to add some flavor to the project as well as to give me an excuse to test out some new features I hadn't used before (show/hide buttons).

The project is hosted on Tableau's public site, linked below:
-

PS: Hover over text and charts for key takeaways points, and use the show/hide buttons to control the display of filters and insights.

10 Fun Facts from the Data:

  1. Games are very snowbally, with almost everyone either dying or escaping with very little results inbetween
  2. Escape and win rates are almost exactly 50% for both sides
  3. If the lowest scoring teammate had at least 13,000 points, the survivors always won in the 100 recorded games
  4. Escapist has the highest pick rate among all 4 classes
  5. Guide has the highest win rate among all 4 classes
  6. Medic has the lowest pick rate among all 4 classes
  7. Medic has the lowest win rate among all 4 classes
  8. Team win rates are at their lowest if the team either does not have any medics, or has more than 2 (so 1-2 medics per team is ideal)
  9. The killer duo survivors won most against was Wraith + Nurse; whereas they had the lowest win rate vs Wraith + Huntress
  10. The map with the highest survivor win rate was The Thompson House; whereas the map with the lowest win rate for survivors was Suffocation Pit* followed by Disturbed Ward

*Suffocation Pit was bugged where some survivors could not be rescued, and I only got this map 3 games out of 100. That's why I felt the need to include Disturbed Ward as the runner up, since there's more data on that map.

If anyone has feedback or questions on the data, my insights, or the dashboards in general, feel free to leave a comment! Always happy to try help and/or learn more myself. Otherwise, hope you have fun browsing the data and dashboards!

Thanks for your time!

Post edited by Beaburd on

Comments

  • ChaosWam
    ChaosWam Member Posts: 1,845

    This is some real dedicated data collection, for a solo project. Kudos on the research!

  • Beaburd
    Beaburd Member Posts: 998
    edited August 3

    Thanks!

    I went for an old book format to help simulate the feeling that these stats came from a book taken from the game's own archives - which I think would be super cool to have one day.

    That said, super cool to have a dev review it! Definitely no rush - it's the weekend, so I hope you enjoy it first and foremost!
    I appreciate that!

    @ChaosWam

    This is some real dedicated data collection, for a solo project. Kudos on the research!

    Thanks you!

    Definitely played waaaay too much during the free weekend to get this data (45 games in a day I think), but hey, at least the new mode was fun enough that I didn't mind lol benefits of doing a project on something you enjoy.

  • Dreamnomad
    Dreamnomad Member Posts: 3,951
    edited August 3

    It's interesting to me that guide has highest win rate. Kind of makes sense though. Finishing gens is how to win.

  • crogers271
    crogers271 Member Posts: 1,831

    That's interesting data.

    I haven't played as much (probably 50 to 75 games), and mostly post free weekend, but this tends to track. I play medic usually because I try to play what the other survivors are not and I think its a little stronger than people think, but I also enter the game with a strategy for the role.

    On the feast or famine - I feel like I've had a lot more games with a middle result than you did, but more importantly I think is the average number of gens completed. It's not unusual to have a 4k in the 1v4 mode with multiple gens remaining. I find that to be extremely rare in 2v8 and the gen count pretty much always gets knocked out.

    You're counting hatch escapes the same as door escapes, correct? If they are hatch escapes without the gens getting completed they should be a different stat.

    Nurse stats - this really varies based off the level of nurses you hit. I matched against some who barely needed their partner, and some who looked to be learning the role and were basically an auto loss.

  • Beaburd
    Beaburd Member Posts: 998

    Yeah if your experiences are after the free weekend, I wouldn't be surprised if you noticed more mid-level survivor escapes and less extremes than I recorded. I think the fact I was dealing more with new killers and new survivors could have played a large part in the extreme end game results I experienced. Could be interesting to try this again one day without free weekend and seeing how it differs (although I don't know if I have it in me to do another 100 runs right now lol).

    In regards to the hatch question, I differentiate hatch vs gate escapes for my personal stats at all times, but not for team stats. For team stats, I just assume if 3 or more people escape, they all got gate; otherwise, if 2 or less escaped, they just got hatch. I thought that was sufficient given hatch spawn conditions. Is your concern that hatch escapes when less than 8 generators are repaired is may skew the data for average generators completed? If so, no worries, I recorded the # of generators done before hatches spawned or were closed.

    Okay my treatment of Nurse stats was not great, thank you for bringing this up in particular.

    I should have clarified that although survivors have the highest win rate against Nurse + Wraith, I do NOT think that means she's weak. In fact, the second worst win rate survivors have is against Nurse + Huntress with a 38% win rate for survivors there, and I remember being DOMINATED in those games. It really should be highlighted, and is kind of reflected by the stats, that Nurse can either be extremely weak (implied with Nurse + Wraith) or extremely strong (implied with Nurse + Huntress) depending on who wields her.

    @Dreamnomad

    It's interesting to me that guide has highest win rate. Kind of makes sense though. Finishing gens is how to win.

    Yeah, Guide stats shocked me a bit, but I agree do make sense.

    What's kind of interesting is that all games that had 0 Guides (though it was only 3 games) were losses. Out of the 24 potential people that could have escaped in those 3 games, only 3 people escaped as well. That means all compositions without a Guide in this data sample had a 0% win rate and 12.5% escape rate.

    Now the sample size is low to the point I would take these findings with (even more than) just a grain of salt, but I still think it's an decent reminder on how important it is to have people doing generators, and to have a class dedicated to both doing them and reminding others to do them.

  • dwight444
    dwight444 Member Posts: 438

    this is cool op, fellow Tableau user here

    I remember being so put off by it initially because it seemed so complicated but once I started learning it I realized how intuitive and easy it was

  • Beaburd
    Beaburd Member Posts: 998

    Hey, that's awesome - Tableau users unite!

    Yeah, exactly the same for me.
    It actually took me a year to finally start learning this thing after putting it on the back burner thinking it would be overly complicated. But like you said, once you get started, it's actually pretty intuitive.

    I say that but I still struggle with comprehending how to do some calculated fields to get the results I want.
    An example is the Team Stats page where I just typed out static print with the best and worst killer duos and maps, when in reality I wanted to make it a dynamic field that changed when you selected certain maps and/or killers. But I was having trouble figuring out how to implement win and escape rates to that field, and somehow combine my killer 1 and killer 2 columns. I can do it in Excel or SQL, but I figured I'd just take a short cut, type it in, and make it a challenge to figure out how to do later.

    Anyway I'm rambling - happy you liked it! Thanks for checking it out <3

  • dwight444
    dwight444 Member Posts: 438

    try using parameters; they're meant to represent a data field in place of a constant value

  • danielmaster87
    danielmaster87 Member Posts: 9,440

    I think in theory the mode is still highly survivor sided, and would be in reality if MMR was a thing. 8 gens is simply too few to do, especially with how many filler pallets back to back there are to waste killers' time. There was 1 match I had which turned from a 6-8 man out to a 2-man out, because there were 5-6 people at the final gen that was seconds from completion, and every single person but me ran off and stopped it from getting done. No body blocks, no commiting, and they didn't even finish a different gen to make up for it. They just gave up. So of course killers are gonna kill a lot when that's what they're up against.

    Some of the class results surprise me and some don't. Obviously Escapist is the best class, because it has Windows, Sprint Burst, and gives a mini Sprint Burst to allies. Medic should to be the next best pick, so people not picking it and losing with it is kinda weird. Fast healing and seeing injured teammates is never a bad thing, and considering people want to heal seemingly constantly instead of doing the gens, I don't get why so few people pick it. The only issue is when Medics stay on gens and non-Medics always go for uncage and heal, because those defeat the purpose of the class. You could deal with having 1, maybe 2, Guides but that's about it. It's not that useful a class. And Technician I'm iffy on because you really don't need gen stuff to make gens go fast, but at the same time it can make up for your teammates' slack.

  • Beaburd
    Beaburd Member Posts: 998
    edited August 3

    I would say the mode, given equal skill levels, is survivor sided myself, and think the data reflects that to a degree.

    Although the win rate of my team over 100 games was exactly 49 wins, 2 draws, and 49 losses (what a perfect split), it is important to note that this was during free weekend.

    And although that means there can be both new killers and survivors alike, due to the number of people in each role, it was far more likely to get a new survivor on your team than a new killer. Pair that thought with the scatterplot that showed even having one person with low points (i.e. low contribution) on your team can negatively affect survivor win rates, and I do think the flood of new players would affect survivor win rates in a more negative fashion than it would for killer. That said, the same scatterplot's r-squared value shows one individuals under-performance is far from the only factor that would contribute to a game loss. I feel it is important to mention that so those players don't get singled out - while it's significant, there's a lot factors at play that lower win rates.

    Not to mention I myself, as a more experienced player relative to the free weekend average, have a 60% escape rate and way higher average point generation (and even minimum points) than the average low performer. I think if the average player were more experienced, had higher average points (i.e. contributed to objectives more), and survived longer, we would see escape and win rates raise more in the survivors favor.

    That said, these are ideal high-level skill scenarios. In the real world scenario with inefficient players and a wide variety of skill disparities, especially during free weekend, I think they nailed the balance pretty well for this event as a whole.

    Side note:
    It would be kind of neat to try this again someday when there isn't a free weekend and/or players are more experienced to out the above theories though.

    Concerning classes, I actually think Medic is understandably at the bottom and it was actually one of the theories I was most looking forward to testing. The ability for auras and healing speed sounds decent, but they are such conditional benefits to Guide.

    1. Medic gets team aura-reading while healing, which is harder to achieve than Guide just touching a random generator.
    2. Medic's aura reading is mainly beneficial to injured folks looking for heals; whereas Guide is beneficial to those same people + anyone free to do generators.
    3. Guide's gen aura also benefits people who aren't around by reminding them to do the objective (has for me anyway); meanwhile, seeing a far off healing aura that you can't interact with isn't that beneficial for others
    4. Medic reduces 4.15 seconds off a solo heal, or 1.85 seconds off a duo heal; which is 8.3 seconds and 5.55 seconds of total survivor time saved if you give those savings to each person involved.
      1. Meanwhile Guide saves a flat 2.7 seconds off generator repairs for every Great skill check; albeit if you get one, and if you hit the Great.
    5. The time saved from Medic depends entirely on what your team does with that time; whereas the time saved by Guide is directly invested in the objective
    6. Medic's ideal scenario and time to shine is also during uncages with mass resets. But what if everyone else, especially healthy non-medics, are doing uncages and helping with resets instead? Your value as a class diminishes because of other peoples actions.

    There's just a lot of conditions and nuance to the Medic class, and it is so reliant on other people understanding your role and letting you thrive and succeed. That kind of team-dependency just does not exist for any other role. Even Guide, the other selfless class, at least gives the user the ability to control when to use its abilities and directly invests its benefits into core objectives. That's the biggest difference that makes Medic fall behind imo.

  • Xxjwaynexx
    Xxjwaynexx Member Posts: 334

    Awesome work bro!!! Interesting to see, I play primarily killer and with that huntress I dont know how many games I've played but with wraith we cook for sure 9/10 times. Nurse has been my worst mate but that because you can only make a select few mistakes with nurse in this mode. The snowball effect is super strong in this mode. These are the stats we need kudos to you

  • Predated
    Predated Member Posts: 2,976

    out of the data and personal analysis, I can 100% say the following on survivor:

    1. the 4th perk doesnt need to be locked untill the final healthstate, in fact, it would benefit killers more in general than it would survivors. Healing yourself is time consuming and in that same amount of time, a teammate could have healed you up and you would have progressed a gen much faster together than alone. That, and self-recovery should be at least 30% increased. Killers can already down, cage and keep chasing in a fast enough time to get pressure. People slugging definitely wont help.
    2. Medic is useless in mass, 2 is the most optimal amount. Scout is arguably the strongest one. While Guide generally gives quite an obvious reason for bigger chances of winning due to generator speed, 4 scouts can basically give permanent info on killer locations if they are positioned correctly. The only thing keeping Scout from truly popping off, is that the "killer is within line of sight within 64 meters" seem to be broken. I could see a killer, and their aura would not be revealed. Because while I could see them, my character couldnt make eye contact. And thats the only issue I think scout has to truly show how powerful it was. I think if they fix that and reduce the distance to 48 meters, it would be fine (64 meters is huge, thats a 128 meter diameter, meaning you can stand in any main building and reveal the killer anywhere as long as your character doesnt have something covering their eyes)
    3. There should be less aura reveals on survivors, or aura reveals that a scout sees when a killer kicks a gen or pallet should be mapwide too (a huntress hitting a surv with a hatchet reveals it to all, so I dont see why this would be different).
    4. I do think hatch should spawn once 2 survivors havent been downed/caged. That would fully prevent slugging for an 8k(its dumb that this even is a thing), and forces survivors to rescue their teammates or head out, IF they can find hatch before.
    5. Now ofcourse for killers, Nurse would be fine with this even if she gets a little nerf. Huntress wouldnt need a nerf if they apply all those changes to survivor. Billy would need a little tweaking, Wraith and Trapper can get a significant buff if they do(like wraith getting All-Seeing Blood in his kit, and Trapper getting iri stone or smth).

  • sinkra
    sinkra Member Posts: 426

    The situation is waaaay different now than during the free weekend. Back then my escape rate was about 50% in 2v8, now it's more like 25%.

  • crogers271
    crogers271 Member Posts: 1,831

    Is your concern that hatch escapes when less than 8 generators are repaired is may skew the data for average generators completed?

    It's more of a conceptual concern in that for some players, me included, a hatch escape is different from a normal escape when tracking escapes. To me a hatch escape is still a 4k/8k for the killer(s), the survivors just won a mini-game that exists in that scenario.

    Okay my treatment of Nurse stats was not great, thank you for bringing this up in particular.

    It's not the stats so much what the Nurse is. I'm having trouble delineating exact numbers from the percentages, but it looks like you faced between 15 and 20 Nurses. A beginner Nurse is radically different than a skilled Nurse. So imagine one of the skilled Nurses you faced was switched to a beginner Nurse, we're looking at one of the 8ks potentially switching to a 0k off that one change.

    100 games is good baseline to draw some general views, but I think the Nurse data set in particular has a higher chance of not being reflective given the variability.

  • 100PercentBPMain
    100PercentBPMain Member Posts: 1,072

    Thanks for the data. I felt like Thompson House was the best level for survivors by far, it's nice to see your data reflect that