I have also added a third analysis sheet just for mirror matches. This looks at ALL mirror matches (even ones with characters that are fatigued or not fully equipped). With only 26 matches to look at, so far, it'll take a while before these results are anywhere close to meaningful.
It's not many but we gathered data from some mirror matches a while back, would it be helpful to add? It was from before starbases so the stats were the same. Collected a total of 211, showing 116 wins and 95 losses. Including Mirror-Mirror Matches. ^^
I have also added a third analysis sheet just for mirror matches. This looks at ALL mirror matches (even ones with characters that are fatigued or not fully equipped). With only 26 matches to look at, so far, it'll take a while before these results are anywhere close to meaningful.
It's not many but we gathered data from some mirror matches a while back, would it be helpful to add? It was from before starbases so the stats were the same. Collected a total of 211, showing 116 wins and 95 losses. Including Mirror-Mirror Matches. ^^
Yea, that's more or less what made me want to,undertake this project. Unfortunately, I'm tracking a lot more data points, so I can't use your old results.
If you have read the enough forum threads and played this game/gauntlet long enough, you don't need any more explanation or collecting any data.
1. When gauntlet starts, it is easy to win even without refresh.
But once you read top 50, your chance of losing is much much much bigger. You can lose 5/5 in 30s and be done for 4hr. Starts very easy and then suddenly it is like super hardcore hell level.
2. The crits/proficiency rolls are total BS. Same as #1.
I always choose the match where I have advantage: higher/better stats, higher crits, but there are always unbelievable loses that can only means it is rigged. feels like getting sucker punch and you just want to throw away the phone.
3. The round selection are designed to give you disadvantage: Skills that don't match your crews, Skills for the crew that you just lost, etc etc. I could refresh merits with 10 times and I still don't get the match that I want
4. Every 3rd match for the crate is always harder to win
5. Talking about merits, I have one thread asking about longest streak without any refresh (when in top 50), no one actually responded with video it can be done even though they did it.
6. The wall of same characters all the time esp combined with #3 above
7. others??
Gauntlet is not pure RNG, the algorithm there is messed up big time.
In fact when people blamed RNG, they are wrong, and they should blame the algorithm -- which means the developer that implemented that --- the designer that design it --- and the company that couldn't figure this out themselves and asking their players to prove it
To me the behold problem is a big problem for veteran players but players have options not to buy the pack since they require dil purchase that requires $$$.
OTOH Gauntlet is free and can get you legendary crew and free parts. Not playing is a loss, but playing is prob an even bigger "loss". FOMO forces ppl to play gauntlet whether they want it or not.
You can read how ppl respond when they play gauntlet:
1. try to make fun of it
2. just get the free stuff
3. turn the screen away and just click continue
4. etc etc
All in all, if DB gets rid gauntlet, it would have made my experience better.
It’s easy at first and then gets harder as you move up the ranks? Wow, who knew? /sarc
There are currently some good questions to be asked with regards to actual crit rate vs. stated crit rate, skill selection for matches immediately after a crew loses (or if someone is very weak in a particular skill across the five crew), and even which opponents are offered. A lot of data collection is happening right now and early conclusions are that it isn’t broken RNG or Evil DB pulling the strings to make us lose...just a lot of selection bias and other fun quirks of human psychology that make it seem worse than it is. More data needs to be collected, of course.
And if it is such a horrible, game-ruining experience for you...why participate? No daily missions require you to go through the meat grinder and the achievements are pretty out of whack with the effort required to get them, so why put yourself though such a terrible experience? Let people enjoy things.
You should just say it without the /sarc /sarc
I have been on and off the game myself the past few months... sometimes I could miss gauntlet for a day or more, so I don't care much about it either now. I already had my locutus immortalized also, so I don't care about the 20k achievement
I enjoyed the game for a year and half myself.
Yes the terrible experience kills but I never said to anyone don't enjoy the game.
Even though I am on and off, I still play events that interest me and I still love grinding certain events for crew that I am interested in.
If you like to collect data, feel free to do that, I never said no.
IMO, it is the developer job to do that. Many successful companies do that so they can react and plan for better product before things turn south
But for me the point to play game is to play game, I have enough stuff with real life and work, and game is not for me to do another work. In other threads I admired ppl doing that and I thanked them for sharing info that help me play this game better.
And, yet, I have collected a fair bit of data here which you have yet to comment on.
I've been trying to add in some amount of formal statistical analysis to my results here. A binomial test for Crit% seemed very straightforward and ended up being pretty easy to implement. There are a few areas where it makes the numbers look a little more suspect, but nothing is really popping out yet.
I have tried to apply the same analysis to the proportions for each decile of the roll magnitude. This relies on the assumption that if it is uniformly distributed, each decile should see approximately 10% of the rolls allocated to it. However, I'm seeing significant over-performance for in the 0-10% decile across all rolls coupled with a significant under-performance in the 10-20% decile.
My only thought on the cause for this is rolls with smaller ranges might be skewing results. Something like Mirror Picard's SCI roll has a skill 22 to 88 (spread of 67). In the Starbase level 0 scenario, he ends up having a 10.4% chance of rolling in the 0-10% range (22-28) and a 9% chance of rolling in the 10-20% range (29-34). This would obviously be different depending on starbase level, but I think the 0-10 range is always going to be a "big" range and 10-20 is always going to be a "small" range, which is why we see that discrepancy.
Unless anyone has better ideas, I am going to update the metrics to filter out any rolls with a range of less than 100. I suspect the Mirror Picard SCI roll is a big contributor to this observation, but low-leveled characters would see the same sort of impact.
Comments
So you start a cmd eng gauntlet and the results are sci med? Bug or do you just combine screenshots that are non related?
It's not many but we gathered data from some mirror matches a while back, would it be helpful to add? It was from before starbases so the stats were the same. Collected a total of 211, showing 116 wins and 95 losses. Including Mirror-Mirror Matches. ^^
this https://forums.disruptorbeam.com/stt/viewthread/62394
and this one https://forums.disruptorbeam.com/stt/viewthread/64071
and this one https://forums.disruptorbeam.com/stt/viewthread/66693/
/old forums
~· Fly with the Subspace Eddies! ·~
Yea, that's more or less what made me want to,undertake this project. Unfortunately, I'm tracking a lot more data points, so I can't use your old results.
And, yet, I have collected a fair bit of data here which you have yet to comment on.
I have tried to apply the same analysis to the proportions for each decile of the roll magnitude. This relies on the assumption that if it is uniformly distributed, each decile should see approximately 10% of the rolls allocated to it. However, I'm seeing significant over-performance for in the 0-10% decile across all rolls coupled with a significant under-performance in the 10-20% decile.
My only thought on the cause for this is rolls with smaller ranges might be skewing results. Something like Mirror Picard's SCI roll has a skill 22 to 88 (spread of 67). In the Starbase level 0 scenario, he ends up having a 10.4% chance of rolling in the 0-10% range (22-28) and a 9% chance of rolling in the 10-20% range (29-34). This would obviously be different depending on starbase level, but I think the 0-10 range is always going to be a "big" range and 10-20 is always going to be a "small" range, which is why we see that discrepancy.
Unless anyone has better ideas, I am going to update the metrics to filter out any rolls with a range of less than 100. I suspect the Mirror Picard SCI roll is a big contributor to this observation, but low-leveled characters would see the same sort of impact.
Regarding the binomial dist, I assume you are going to look at each roll and create 8 p-scores:
5%, 25%, 45%, 65% for player and computer?
12 actually. I combine them for an additional score.
EDIT:
25% scores looking a little suspect, but it's not a HUGE discrepancy yet.