Hello,
Since the start of October I decided to wrote lots of things about each of my games in an excel document:
https://docs.google.com/spreadsheets/d/ ... 2014683610
My initial goal was to see how good was the Hawk which looked good, but got no ability to avoid the guards. And at the same time, I noted lots of others factors.
And I just did 100 games, so now that I got data, I will analyse that.
All games noted were 2 players with 3 floors of 4x4. I joined some games, maybe 30, and all others were from my settings: I let the other player choose the difficulty (normal or hard), random wall, choice of characters.
Let's start:
Column A: who added the line of that document, it was exclusively me since I didn't shared the document much. We got 100 games.
Column B: who was my teammate. I got a total of 59 unique names, which seem pretty impressive. The number of players playing that game looks pretty high.
Column C: elo of my teammate. I got: 2 beginner, 33 apprentice, 17 average, 10 good, 24 strong, 3 expert, 11 master. Not much to add, that look pretty much what I expected.
Column D: how good the plays of boths players were. I think I'm really good, so it mainly depend of my teammate. I think in maybe 3 games I did a major mistakes, and so I removed one or two ranks from it, but all others were the level of my teammate. I try to only remove one rank if I disagree with the meta strategy (like not swapping floors, or using an early thermal bomb or so on).
S (5): Extremely good plays, no or few minor missplays/miss optimisations.
A (26): Very good plays, no unecessary risks taken. Few mistakes done. (sometimes no mistake done, but nothing extraordinary on the game)
B (33): Good plays with unecessary risk taken OR one major mistake OR multiple minor mistakes
C (14): Try to be safe in end of turn, but don't know how to do it well.
D (16): Multiple high risks taken. Don't think too much in a turn.
E (6): Risks taken all over the place, don't know how to play well.
I wanted to see if the level of the person was related to how good they played. I got with ranking S=6, A=5, B=4, C=3, D=2, E=1 that we got in average: Beginner=2,5 / Apprentice=3,15 / Average=3,82 / Good=4,00 / Strong=4,25 / Expert=4,33 / Master=3,91.
So outside of masters, it goes in ascending order which mean... That players improve with more games, pretty nice. We can see that you gain 0,67 level after 100 games. And after it improve slowly, with about 0,2 every 100 games.
Column E: My choice of characters, not the much interresting for that post. Note that this date is biaised into Normal character since some tables don't allow advanced. I like complicated characters which are weird are give interesting decisions. N=Normal. A=Advanced. And so I got 15 Raven A and 15 Juicer A. 8 Spotter A, 8 Hacker N (that I tried more since I dislike him). And 1 Rigger N, 2 Rook A, 2 Hacker A.
Column F: My teammate choice of characters, much more interresting. Here is the order:
Acrobat N (15) > Hawk N (12) & Juicer N (12) > Raven N (11) > Hacker N (10) > Hawk A (8) > Rook A (5) > Acrobat A (5) > Peterman N (4) > Spotter N (4) > Rigger N (4) > Hacker A (4) > Rook N (2) > Raven A (2) > Juicer A (1) > Rigger A (1) > Peterman A (0) > Spotter A (0).
Again the data is biaised into Normal characters. Really surprise by the Rook advanced peeked five times which could be variance since I don't feel it's much interesting? If you like to play it at 2 players, I would love to see the reasons for it. Hawk A is prefered which make sense, look strong and is easy to play. But few love for weird characters, if you like the game, why don't try to improve/change your playstyle with them? Peterman A is bugged so that make sense. But Spotter A, Juicer A, and Raven A are all interesting characters which require some change of gameplay.
Column G: Difficulty. 61 Normal & 39 Hard.
Column H: Winrate:
I got a total in normal of 49 win and 12 lose, which give a winrate of 80,3%.
And a total in hard of 18 win for 21 lose which give a winrate of 46,2%.
Now the winrate of characters. Harder to analyse since some of them don't have enough data, and I need to separate normal and hard. Furthermore, easy to play characters should be picked by worst player, which make the data biaised down in them.
So I calculated the winrate, and if it was lower than number of games * 15, I reduced that winrate to that value. Then I took the average between normal and hard. Here is the order:
- Hawk A finished first (4/1n 5/2h (4victories and 1 defeat in normal, 5victories and 2 defeats in hard)). I don't think strong players pick her, it just look a very strong ability.
- Acrobat N got 10/3n 4/2h. Pretty surprised by him. It should be biaised into more losing since that's easy to play. Look like being able to bypass the guard is extremely powerful in corridors.
- Spotter N got 7/0n 1/2h.
- Juicer N got 7/1n 3/4h. I have found that many people don't optimise her, which make her winrate way lower than it should be. Which make sense since optimising her is boring and unecessary long.
- Peterman N got 5/0n 2/1h. Maybe because I picked him to try all of them, which increase it's winrate.
- Hawk N got 6/3n 4/5h.
- Spotter A got 4/0n 2/2h. Again, nobody pick it, so my games increase it's winrate.
- Rook A 2/0n 4/1h. It's my surprise.
- Juicer A 3/3n 5/5h. That one is surprising to me. I have found many downside of Juicer A, the bigger one is that it's hard to help a bad player. You are safer for yourself, but can't really help the other player. I think it's variance too and she should be higher. But I could be wrong, and maybe even with the many things she does, she isn't as strong as I though.
- Raven A 5/4n 3/5h. That date could be biaised up because I'm picking her. But if that's the case, Juicer A should be even lower which is weird. I don't believe she is stronger than others character though. You gain maybe 3 free turns per games, but is gaining 12 actions enough? That crack you a safe you don't choose for free.
- Raven N 13/1n 0/1h. Strangly that data have one game with her in hard. So that make her winrate in hard at 0. She is higher than that.
- Hacker N got 9/3n 1/5h. Could be because I play it badly in hard. But I think he is really weak. I don't think that ability is negative though.
- Hacker A got 4/0n 1/1h. Not enough plays in hard, he is stronger.
- Rook N got 5/2n 1/0h. Not enough games in hard. He is stronger.
- Acrobat A is 5/1n 0/4h. 0/4 in hard is extremely low. I really like that character but his ability look too annoying to use? Or is it variance?
- Rigger N got 4/0n 0/1h. Not enough data.
It's really hard to compare characters. I will come back to it later but I think that the luck of the game and the floor difficulty is more important than the characters. Having a deadbolt in an important place is way more detrimental than a bad character.
If you have any idea to compare those characters better with my data, feel free to do it yourself, or ask, I could do it.
Column I: Turns at the end.
Couting everything, I have an average of 275 turns.
Counting only wins, I got an average of 302 turns. Counting only loses, I got 220.
Counting only normal games win, I got an average of 298 turns. Counting only hards games win, I got 313. Surprise that hard games has the same duration than normal games. The 15 turns difference can easily be explained by the fact that you can use your 2 extra stealth as a ressources to escape faster in floor 3.
11 games normal got more than 350 turns for 5 games hard. Remember we got less hard games.
3 games normal got above 400 turns (409, 421, 432). 2 games hard got above 400 turns (448, 511). I though when looking at data that there was rare hard game with a lot more turns. But that doesn't look true. Look like just variance that I got no normal game with very high number of turns.
I looked at the average number of turns depending of characters. Hawk which should be lower are ranked 5th and 10th. Rigger N and Hacker N are first and second. Slowest is Raven N.
I then tried to only count wins in that date, and hawk are 8th and 13th.
Again, nothing seem to matter in the turns at the end, so the choice of characters doesn't look to matter. So my argument Hawk is good because games are shorter is not true, or is too much minor.
Column J: Lowest stealth left ~2/3 turns before escaping (add 10 for hard mode).
I will not analyse that, doesn't look interesting. But feel free to check that data if you want.
Only 15 normal games out of 49 wins (30.6%) got 2 or more stealths at the end. Compare to 46.2% winrate in hard.
Column K: Does someone stayed in a low floor to only help others?
I like personnally to play fully support, hide the whole game, crack floor 1, sometimes crack floor 2, and give an upside to my teammate. Those rare games are counted there.
3 Spotter A, 3 Rook N, and 1 Spotter N are present there. All played by myself. So no other player decided to do that. I find the same things in battlegrounds duos where I like to play support and adapt to my teammate. But it look like it's not what most players do.
Column L: Who added / rolled in safe 1?
Hard to analyse. Hawk A got a nice 0%. Hawk N got 61%.
Juicer N, Rook A, Peterman N, Spotter N, Hacker A, Rook N got above 50%. Doesn't look related to their character whatsoever. But some games I purposely let an ability goes up. But that look rare, and not enough to see it in data.
Hawk A is the exception. Hawk N though is nice too as a second peeker since you peek at 70% of the floor, and that ability let her get the important rooms left. You don't have to let her goes up. Or at least it's how I play, and that's reflected in the data.
Column M: Who displayed the most tiles in floors 2&3?
Read right below.
Column N: Who added / rolled in safe 3?
The first thing I checked was if the person who peeked at floors 2&3 was the same than the one who opened safe in floor 3. I find it was the case in 58% of cases (58/71). In 29% of cases there was a tie somewhere, or the game didn’t last enough to open safe in floor 3.
Hawk A still got a staging 83% for displaying tiles in floors 2&3, and 75% to crack floor 3.
Don’t really know what I can do with that data so I will stop here.
Columns O to T.
Those columns were different, I give a rating on how difficult the game was without clear directions. I tried to be neutral, but I’m biaised. If the game finish by a lose, I tried to be neutral, but I would rate them lower because I’m in that mood per example.
S mean the game was easy.
E mean the game was hard.
Column O: Lose stealth from start of game to 1/2 turns after someone goes to floor 2? (A:no / B: yes).
That column was the only one with clear direction:
S (62): no direct risk of stealth loss
A (11): risk of stealth loss
B (13): one stealth lost
C (8): one stealth lost+two others risks of stealth lost
D (5): two stealth lost
E (1): worst than that
We can see that 73% of games goes with a floor 1 without stealth lost. I repeat that it is 2 players.
21% of games got one stealth loss.
And 66% got two.
Column P: Was there problems (ie risk of stealth loss) for peeking tiles?
53S, 16A, 19B, 2C, 6D, 1E.
This is the best ratings across all remaining columns outside of escaping (in%, not shown here). Which mean it’s the easiest task to do.
Column Q: Was there problems for adding & rolling dices / reaching the safe safely?
46S, 16A, 16B, 11C, 5D, 5E.
This is the hardest task by far (~30% more problematic in %, 15% more flat).
Column R: Was there problems (risk of stealth loss) to escape?
42S, 10A, 4B, 6C, 4D, 4E.
In %: 60S, 14A, 6B, 9C, 6D, 6E.
Extremely surprised by that. 28% of games got a C or lower. You can regularly find someone in floor 1 or 2, waiting for the other player to escape, or find a gold bar in floor 3. While some items can resolve that part easily, some others are extremely more annoying. I got one game (with comments) that I took event because going back to floor 2, swapping position, make my teammate goes to the safe in floor 3, make my teammate going back to the stairs looked far worst than taking events and hoping to find a right one.
Column S: Luck. In important situations (like stealth loss), did you get lucky? (default B)
5S, 19A, 38B, 22C, 15D, 1E.
Coming soon.
Column T: Floor difficulty. Was the floors layout friendly (no deadbolt in center per ex)(mainly floors 2&3)?
14S, 29A, 17B, 25C, 11D, 4E.
Coming soon.
Column U: Comments.
Others columns are formula.
Thanks for reading, I will add more stuff over time / with your potential ideas of analysis. If you have any questions I could check with that date feel free to ask. If you want to use my data feel free to do it.
Thanks
Since the start of October I decided to wrote lots of things about each of my games in an excel document:
https://docs.google.com/spreadsheets/d/ ... 2014683610
My initial goal was to see how good was the Hawk which looked good, but got no ability to avoid the guards. And at the same time, I noted lots of others factors.
And I just did 100 games, so now that I got data, I will analyse that.
All games noted were 2 players with 3 floors of 4x4. I joined some games, maybe 30, and all others were from my settings: I let the other player choose the difficulty (normal or hard), random wall, choice of characters.
Let's start:
Column A: who added the line of that document, it was exclusively me since I didn't shared the document much. We got 100 games.
Column B: who was my teammate. I got a total of 59 unique names, which seem pretty impressive. The number of players playing that game looks pretty high.
Column C: elo of my teammate. I got: 2 beginner, 33 apprentice, 17 average, 10 good, 24 strong, 3 expert, 11 master. Not much to add, that look pretty much what I expected.
Column D: how good the plays of boths players were. I think I'm really good, so it mainly depend of my teammate. I think in maybe 3 games I did a major mistakes, and so I removed one or two ranks from it, but all others were the level of my teammate. I try to only remove one rank if I disagree with the meta strategy (like not swapping floors, or using an early thermal bomb or so on).
S (5): Extremely good plays, no or few minor missplays/miss optimisations.
A (26): Very good plays, no unecessary risks taken. Few mistakes done. (sometimes no mistake done, but nothing extraordinary on the game)
B (33): Good plays with unecessary risk taken OR one major mistake OR multiple minor mistakes
C (14): Try to be safe in end of turn, but don't know how to do it well.
D (16): Multiple high risks taken. Don't think too much in a turn.
E (6): Risks taken all over the place, don't know how to play well.
I wanted to see if the level of the person was related to how good they played. I got with ranking S=6, A=5, B=4, C=3, D=2, E=1 that we got in average: Beginner=2,5 / Apprentice=3,15 / Average=3,82 / Good=4,00 / Strong=4,25 / Expert=4,33 / Master=3,91.
So outside of masters, it goes in ascending order which mean... That players improve with more games, pretty nice. We can see that you gain 0,67 level after 100 games. And after it improve slowly, with about 0,2 every 100 games.
Column E: My choice of characters, not the much interresting for that post. Note that this date is biaised into Normal character since some tables don't allow advanced. I like complicated characters which are weird are give interesting decisions. N=Normal. A=Advanced. And so I got 15 Raven A and 15 Juicer A. 8 Spotter A, 8 Hacker N (that I tried more since I dislike him). And 1 Rigger N, 2 Rook A, 2 Hacker A.
Column F: My teammate choice of characters, much more interresting. Here is the order:
Acrobat N (15) > Hawk N (12) & Juicer N (12) > Raven N (11) > Hacker N (10) > Hawk A (8) > Rook A (5) > Acrobat A (5) > Peterman N (4) > Spotter N (4) > Rigger N (4) > Hacker A (4) > Rook N (2) > Raven A (2) > Juicer A (1) > Rigger A (1) > Peterman A (0) > Spotter A (0).
Again the data is biaised into Normal characters. Really surprise by the Rook advanced peeked five times which could be variance since I don't feel it's much interesting? If you like to play it at 2 players, I would love to see the reasons for it. Hawk A is prefered which make sense, look strong and is easy to play. But few love for weird characters, if you like the game, why don't try to improve/change your playstyle with them? Peterman A is bugged so that make sense. But Spotter A, Juicer A, and Raven A are all interesting characters which require some change of gameplay.
Column G: Difficulty. 61 Normal & 39 Hard.
Column H: Winrate:
I got a total in normal of 49 win and 12 lose, which give a winrate of 80,3%.
And a total in hard of 18 win for 21 lose which give a winrate of 46,2%.
Now the winrate of characters. Harder to analyse since some of them don't have enough data, and I need to separate normal and hard. Furthermore, easy to play characters should be picked by worst player, which make the data biaised down in them.
So I calculated the winrate, and if it was lower than number of games * 15, I reduced that winrate to that value. Then I took the average between normal and hard. Here is the order:
- Hawk A finished first (4/1n 5/2h (4victories and 1 defeat in normal, 5victories and 2 defeats in hard)). I don't think strong players pick her, it just look a very strong ability.
- Acrobat N got 10/3n 4/2h. Pretty surprised by him. It should be biaised into more losing since that's easy to play. Look like being able to bypass the guard is extremely powerful in corridors.
- Spotter N got 7/0n 1/2h.
- Juicer N got 7/1n 3/4h. I have found that many people don't optimise her, which make her winrate way lower than it should be. Which make sense since optimising her is boring and unecessary long.
- Peterman N got 5/0n 2/1h. Maybe because I picked him to try all of them, which increase it's winrate.
- Hawk N got 6/3n 4/5h.
- Spotter A got 4/0n 2/2h. Again, nobody pick it, so my games increase it's winrate.
- Rook A 2/0n 4/1h. It's my surprise.
- Juicer A 3/3n 5/5h. That one is surprising to me. I have found many downside of Juicer A, the bigger one is that it's hard to help a bad player. You are safer for yourself, but can't really help the other player. I think it's variance too and she should be higher. But I could be wrong, and maybe even with the many things she does, she isn't as strong as I though.
- Raven A 5/4n 3/5h. That date could be biaised up because I'm picking her. But if that's the case, Juicer A should be even lower which is weird. I don't believe she is stronger than others character though. You gain maybe 3 free turns per games, but is gaining 12 actions enough? That crack you a safe you don't choose for free.
- Raven N 13/1n 0/1h. Strangly that data have one game with her in hard. So that make her winrate in hard at 0. She is higher than that.
- Hacker N got 9/3n 1/5h. Could be because I play it badly in hard. But I think he is really weak. I don't think that ability is negative though.
- Hacker A got 4/0n 1/1h. Not enough plays in hard, he is stronger.
- Rook N got 5/2n 1/0h. Not enough games in hard. He is stronger.
- Acrobat A is 5/1n 0/4h. 0/4 in hard is extremely low. I really like that character but his ability look too annoying to use? Or is it variance?
- Rigger N got 4/0n 0/1h. Not enough data.
It's really hard to compare characters. I will come back to it later but I think that the luck of the game and the floor difficulty is more important than the characters. Having a deadbolt in an important place is way more detrimental than a bad character.
If you have any idea to compare those characters better with my data, feel free to do it yourself, or ask, I could do it.
Column I: Turns at the end.
Couting everything, I have an average of 275 turns.
Counting only wins, I got an average of 302 turns. Counting only loses, I got 220.
Counting only normal games win, I got an average of 298 turns. Counting only hards games win, I got 313. Surprise that hard games has the same duration than normal games. The 15 turns difference can easily be explained by the fact that you can use your 2 extra stealth as a ressources to escape faster in floor 3.
11 games normal got more than 350 turns for 5 games hard. Remember we got less hard games.
3 games normal got above 400 turns (409, 421, 432). 2 games hard got above 400 turns (448, 511). I though when looking at data that there was rare hard game with a lot more turns. But that doesn't look true. Look like just variance that I got no normal game with very high number of turns.
I looked at the average number of turns depending of characters. Hawk which should be lower are ranked 5th and 10th. Rigger N and Hacker N are first and second. Slowest is Raven N.
I then tried to only count wins in that date, and hawk are 8th and 13th.
Again, nothing seem to matter in the turns at the end, so the choice of characters doesn't look to matter. So my argument Hawk is good because games are shorter is not true, or is too much minor.
Column J: Lowest stealth left ~2/3 turns before escaping (add 10 for hard mode).
I will not analyse that, doesn't look interesting. But feel free to check that data if you want.
Only 15 normal games out of 49 wins (30.6%) got 2 or more stealths at the end. Compare to 46.2% winrate in hard.
Column K: Does someone stayed in a low floor to only help others?
I like personnally to play fully support, hide the whole game, crack floor 1, sometimes crack floor 2, and give an upside to my teammate. Those rare games are counted there.
3 Spotter A, 3 Rook N, and 1 Spotter N are present there. All played by myself. So no other player decided to do that. I find the same things in battlegrounds duos where I like to play support and adapt to my teammate. But it look like it's not what most players do.
Column L: Who added / rolled in safe 1?
Hard to analyse. Hawk A got a nice 0%. Hawk N got 61%.
Juicer N, Rook A, Peterman N, Spotter N, Hacker A, Rook N got above 50%. Doesn't look related to their character whatsoever. But some games I purposely let an ability goes up. But that look rare, and not enough to see it in data.
Hawk A is the exception. Hawk N though is nice too as a second peeker since you peek at 70% of the floor, and that ability let her get the important rooms left. You don't have to let her goes up. Or at least it's how I play, and that's reflected in the data.
Column M: Who displayed the most tiles in floors 2&3?
Read right below.
Column N: Who added / rolled in safe 3?
The first thing I checked was if the person who peeked at floors 2&3 was the same than the one who opened safe in floor 3. I find it was the case in 58% of cases (58/71). In 29% of cases there was a tie somewhere, or the game didn’t last enough to open safe in floor 3.
Hawk A still got a staging 83% for displaying tiles in floors 2&3, and 75% to crack floor 3.
Don’t really know what I can do with that data so I will stop here.
Columns O to T.
Those columns were different, I give a rating on how difficult the game was without clear directions. I tried to be neutral, but I’m biaised. If the game finish by a lose, I tried to be neutral, but I would rate them lower because I’m in that mood per example.
S mean the game was easy.
E mean the game was hard.
Column O: Lose stealth from start of game to 1/2 turns after someone goes to floor 2? (A:no / B: yes).
That column was the only one with clear direction:
S (62): no direct risk of stealth loss
A (11): risk of stealth loss
B (13): one stealth lost
C (8): one stealth lost+two others risks of stealth lost
D (5): two stealth lost
E (1): worst than that
We can see that 73% of games goes with a floor 1 without stealth lost. I repeat that it is 2 players.
21% of games got one stealth loss.
And 66% got two.
Column P: Was there problems (ie risk of stealth loss) for peeking tiles?
53S, 16A, 19B, 2C, 6D, 1E.
This is the best ratings across all remaining columns outside of escaping (in%, not shown here). Which mean it’s the easiest task to do.
Column Q: Was there problems for adding & rolling dices / reaching the safe safely?
46S, 16A, 16B, 11C, 5D, 5E.
This is the hardest task by far (~30% more problematic in %, 15% more flat).
Column R: Was there problems (risk of stealth loss) to escape?
42S, 10A, 4B, 6C, 4D, 4E.
In %: 60S, 14A, 6B, 9C, 6D, 6E.
Extremely surprised by that. 28% of games got a C or lower. You can regularly find someone in floor 1 or 2, waiting for the other player to escape, or find a gold bar in floor 3. While some items can resolve that part easily, some others are extremely more annoying. I got one game (with comments) that I took event because going back to floor 2, swapping position, make my teammate goes to the safe in floor 3, make my teammate going back to the stairs looked far worst than taking events and hoping to find a right one.
Column S: Luck. In important situations (like stealth loss), did you get lucky? (default B)
5S, 19A, 38B, 22C, 15D, 1E.
Coming soon.
Column T: Floor difficulty. Was the floors layout friendly (no deadbolt in center per ex)(mainly floors 2&3)?
14S, 29A, 17B, 25C, 11D, 4E.
Coming soon.
Column U: Comments.
Others columns are formula.
Thanks for reading, I will add more stuff over time / with your potential ideas of analysis. If you have any questions I could check with that date feel free to ask. If you want to use my data feel free to do it.
Thanks