Guides
Community created guides, helpful strategies, and more.
TOPIC | [guide] Coli skills + enemy stats
@Sylvandyr
Have you checked the [url=http://www1.flightrising.com/forums/gde/937136/96#post_17887038]Ghostlight Ruins trainer[/url] in Coliseum 2.0?
[quote]exactly 76 QCK for your lv25 Haste to not waste the first turn against all the 48 QCK monsters[/quote]
My 70 QCK dragon gets the Haste turn against Aer Phantom/Lux Spectre 3-packs (50 QCK), so maybe the new threshold is QCK/2.
@Sylvandyr
Have you checked the Ghostlight Ruins trainer in Coliseum 2.0?
My 70 QCK dragon gets the Haste turn against Aer Phantom/Lux Spectre 3-packs (50 QCK), so maybe the new threshold is QCK/2.
Have you checked the Ghostlight Ruins trainer in Coliseum 2.0?
Quote:
exactly 76 QCK for your lv25 Haste to not waste the first turn against all the 48 QCK monsters
My 70 QCK dragon gets the Haste turn against Aer Phantom/Lux Spectre 3-packs (50 QCK), so maybe the new threshold is QCK/2.
@Maki Oh wow, that's really interesting. I noticed the relative turn orders shifted a bit for certain enemies after the first turn bug was fixed, but I haven't taken a good look yet. This edge case data will be really important for figuring out how the turn order algorithm really works (instead of just figuring out something "close").
I haven't checked the Ghostlight Ruins trainers in detail for 2.0, but I need to. A lot of my GLR posts are outdated. I think I'll bump this to the top of my priority list when I get home from vacation.
Thanks so much!
[EDIT]: Started keeping track on a spreadsheet for detailed comparison:
Sheet link
Gonna try to tincture and record lower QCK (maybe starting with 40 QCK).
I haven't checked the Ghostlight Ruins trainers in detail for 2.0, but I need to. A lot of my GLR posts are outdated. I think I'll bump this to the top of my priority list when I get home from vacation.
Thanks so much!
[EDIT]: Started keeping track on a spreadsheet for detailed comparison:
Sheet link
Gonna try to tincture and record lower QCK (maybe starting with 40 QCK).
@Maki Oh wow, that's really interesting. I noticed the relative turn orders shifted a bit for certain enemies after the first turn bug was fixed, but I haven't taken a good look yet. This edge case data will be really important for figuring out how the turn order algorithm really works (instead of just figuring out something "close").
I haven't checked the Ghostlight Ruins trainers in detail for 2.0, but I need to. A lot of my GLR posts are outdated. I think I'll bump this to the top of my priority list when I get home from vacation.
Thanks so much!
[EDIT]: Started keeping track on a spreadsheet for detailed comparison:
Sheet link
Gonna try to tincture and record lower QCK (maybe starting with 40 QCK).
I haven't checked the Ghostlight Ruins trainers in detail for 2.0, but I need to. A lot of my GLR posts are outdated. I think I'll bump this to the top of my priority list when I get home from vacation.
Thanks so much!
[EDIT]: Started keeping track on a spreadsheet for detailed comparison:
Sheet link
Gonna try to tincture and record lower QCK (maybe starting with 40 QCK).
@Sylvandyr
I did some more testing and I noticed this:
Round 1:
if QCK dragon/monster >= 1 -> 1 turn
if QCK dragon/monster >= 1.95 -> 2 turns [tested 75 QCK against Blue Moon Aviar 38 QCK]
Round 2:
if QCK dragon/monster >= 1.45 -> 2 turns [tested 58 QCK against Hippocampus 40 QCK (2 turns) and 75 QCK against Blue Tang Hippogriff 52 QCK (1 turn)]
For Ghostlight Ruins:
You need at least 50x1.95-30=67.5~68 QCK to get the Haste turn against the fastest monsters (Aer Phantom / Lux Spectre).
You need 50x1.45=72.5~73 QCK in order to get 2 turns in the second round, but this is not very important, since you can Rally Eliminate two monsters in the first round and then you need only one turn for the third monster.
Edit: There is a Skydancer (9 QCK!) among my fodder Imperials, so I have to retest this.
78 QCK (+ 5 QCK x2) is enough to get two turns in the second round against Blue Tang Hippogriff (52 QCK), so it's 78/52=1.5.
I did some more testing and I noticed this:
if QCK dragon/monster >= 1 -> 1 turn
if QCK dragon/monster >= 1.95 -> 2 turns [tested 75 QCK against Blue Moon Aviar 38 QCK]
Round 2:
if QCK dragon/monster >= 1.45 -> 2 turns [tested 58 QCK against Hippocampus 40 QCK (2 turns) and 75 QCK against Blue Tang Hippogriff 52 QCK (1 turn)]
For Ghostlight Ruins:
You need at least 50x1.95-30=67.5~68 QCK to get the Haste turn against the fastest monsters (Aer Phantom / Lux Spectre).
You need 50x1.45=72.5~73 QCK in order to get 2 turns in the second round, but this is not very important, since you can Rally Eliminate two monsters in the first round and then you need only one turn for the third monster.
Edit: There is a Skydancer (9 QCK!) among my fodder Imperials, so I have to retest this.
78 QCK (+ 5 QCK x2) is enough to get two turns in the second round against Blue Tang Hippogriff (52 QCK), so it's 78/52=1.5.
@Sylvandyr
I did some more testing and I noticed this:
Round 1:
if QCK dragon/monster >= 1 -> 1 turn
if QCK dragon/monster >= 1.95 -> 2 turns [tested 75 QCK against Blue Moon Aviar 38 QCK]
Round 2:
if QCK dragon/monster >= 1.45 -> 2 turns [tested 58 QCK against Hippocampus 40 QCK (2 turns) and 75 QCK against Blue Tang Hippogriff 52 QCK (1 turn)]
For Ghostlight Ruins:
You need at least 50x1.95-30=67.5~68 QCK to get the Haste turn against the fastest monsters (Aer Phantom / Lux Spectre).
You need 50x1.45=72.5~73 QCK in order to get 2 turns in the second round, but this is not very important, since you can Rally Eliminate two monsters in the first round and then you need only one turn for the third monster.
Edit: There is a Skydancer (9 QCK!) among my fodder Imperials, so I have to retest this.
78 QCK (+ 5 QCK x2) is enough to get two turns in the second round against Blue Tang Hippogriff (52 QCK), so it's 78/52=1.5.
I did some more testing and I noticed this:
if QCK dragon/monster >= 1 -> 1 turn
if QCK dragon/monster >= 1.95 -> 2 turns [tested 75 QCK against Blue Moon Aviar 38 QCK]
Round 2:
if QCK dragon/monster >= 1.45 -> 2 turns [tested 58 QCK against Hippocampus 40 QCK (2 turns) and 75 QCK against Blue Tang Hippogriff 52 QCK (1 turn)]
For Ghostlight Ruins:
You need at least 50x1.95-30=67.5~68 QCK to get the Haste turn against the fastest monsters (Aer Phantom / Lux Spectre).
You need 50x1.45=72.5~73 QCK in order to get 2 turns in the second round, but this is not very important, since you can Rally Eliminate two monsters in the first round and then you need only one turn for the third monster.
Edit: There is a Skydancer (9 QCK!) among my fodder Imperials, so I have to retest this.
78 QCK (+ 5 QCK x2) is enough to get two turns in the second round against Blue Tang Hippogriff (52 QCK), so it's 78/52=1.5.
@Sylvandyr @Maki
I also looked at the turn order. I was looking for a holistic, predictive formula, and so was doing much of my testing in the training fields, where I could pass ~70 turns without any fear of death.
I found that the turn order pattern seemed to loop once every unit had taken a number of turns equal it its speed. For example, if the battle is one dragon with 38 speed and two enemies with 5 and 6 speed, after the dragon takes 38 turns and the enemies take 5 and 6 turns, the same order pattern repeats.
I did not find any nice patterns with the order within each cycle though. I suspect it is some sort of thing that builds towards a threshold, but I did not have any success finding that threshold.
In looking at my logs, I also saw a dragon with 9 speed get two turns before a dragon with 5 speed (speed x1.8), but an enemy with 8 speed did not get the double turn. I know data collection is a hard part, so let me know if you would like me to PM the logs. (No sense clogging this up with hundreds of lines battle cycles.)
I also looked at the turn order. I was looking for a holistic, predictive formula, and so was doing much of my testing in the training fields, where I could pass ~70 turns without any fear of death.
I found that the turn order pattern seemed to loop once every unit had taken a number of turns equal it its speed. For example, if the battle is one dragon with 38 speed and two enemies with 5 and 6 speed, after the dragon takes 38 turns and the enemies take 5 and 6 turns, the same order pattern repeats.
I did not find any nice patterns with the order within each cycle though. I suspect it is some sort of thing that builds towards a threshold, but I did not have any success finding that threshold.
In looking at my logs, I also saw a dragon with 9 speed get two turns before a dragon with 5 speed (speed x1.8), but an enemy with 8 speed did not get the double turn. I know data collection is a hard part, so let me know if you would like me to PM the logs. (No sense clogging this up with hundreds of lines battle cycles.)
@Sylvandyr @Maki
I also looked at the turn order. I was looking for a holistic, predictive formula, and so was doing much of my testing in the training fields, where I could pass ~70 turns without any fear of death.
I found that the turn order pattern seemed to loop once every unit had taken a number of turns equal it its speed. For example, if the battle is one dragon with 38 speed and two enemies with 5 and 6 speed, after the dragon takes 38 turns and the enemies take 5 and 6 turns, the same order pattern repeats.
I did not find any nice patterns with the order within each cycle though. I suspect it is some sort of thing that builds towards a threshold, but I did not have any success finding that threshold.
In looking at my logs, I also saw a dragon with 9 speed get two turns before a dragon with 5 speed (speed x1.8), but an enemy with 8 speed did not get the double turn. I know data collection is a hard part, so let me know if you would like me to PM the logs. (No sense clogging this up with hundreds of lines battle cycles.)
I also looked at the turn order. I was looking for a holistic, predictive formula, and so was doing much of my testing in the training fields, where I could pass ~70 turns without any fear of death.
I found that the turn order pattern seemed to loop once every unit had taken a number of turns equal it its speed. For example, if the battle is one dragon with 38 speed and two enemies with 5 and 6 speed, after the dragon takes 38 turns and the enemies take 5 and 6 turns, the same order pattern repeats.
I did not find any nice patterns with the order within each cycle though. I suspect it is some sort of thing that builds towards a threshold, but I did not have any success finding that threshold.
In looking at my logs, I also saw a dragon with 9 speed get two turns before a dragon with 5 speed (speed x1.8), but an enemy with 8 speed did not get the double turn. I know data collection is a hard part, so let me know if you would like me to PM the logs. (No sense clogging this up with hundreds of lines battle cycles.)
Thank you both for the help! (Sorry for the slow response; still overseas.) Since 2.0 is here now and we have new data, I think it'd be awesome if we can crack this algorithm once and for all, though it will definitely be tough.
@Maki Thank you for retesting with new QCK fodder :) The exactness of 1.5x looks promising, I wonder if it's that way for all enemy QCK levels or if it happened to hit that exact value.
For the Blue Tang Hippogriff, do you remember if it was Blue x2 + Cardinal x1, or the other way around? It's seemed to me in the past that each actor contributes to the turn order and there can be slight differences depending on the number of each QCK.
@RageAsylum I would love it if you could PM the logs please. :) Thank you!
---
I think there could also be a slight difference between having the QCK to start with vs. Haste to reach a certain QCK level. (We'd have to test it with low dragon QCK levels to be able to match the Hasted value.) I have no written proof, but I seem to remember this based on observation. (There are cases where I'll Eliminate the same-QCK enemy that hadn't moved yet vs. one that just moved, and the thing that just moved didn't get another turn where the one that hadn't moved did.) Might be worth revisiting.
I'm thinking of writing a tool where we can test our various theories for algorithms and just plug in data to verify whether we can find out anything from that particular algorithm. (If there are any edge case mismatches, we'll have to redo it.) I need to think some more on how to simulate KOs to remove the actor from the list of QCK actors. (I want the data entry to be relatively painless.) I'll put some more thought into that before publishing a tool. There would be some sort of simple text format with key words for submission and something that allows you to easily load the actors. (Probably the known QCK enemies would be saved and you'd just need load a slight thing from a text file for the dragons on your side.)
Once this thing is written, it shouldn't be hard to add more theories or tweak current ones to search for a match.
@Maki Thank you for retesting with new QCK fodder :) The exactness of 1.5x looks promising, I wonder if it's that way for all enemy QCK levels or if it happened to hit that exact value.
For the Blue Tang Hippogriff, do you remember if it was Blue x2 + Cardinal x1, or the other way around? It's seemed to me in the past that each actor contributes to the turn order and there can be slight differences depending on the number of each QCK.
@RageAsylum I would love it if you could PM the logs please. :) Thank you!
---
I think there could also be a slight difference between having the QCK to start with vs. Haste to reach a certain QCK level. (We'd have to test it with low dragon QCK levels to be able to match the Hasted value.) I have no written proof, but I seem to remember this based on observation. (There are cases where I'll Eliminate the same-QCK enemy that hadn't moved yet vs. one that just moved, and the thing that just moved didn't get another turn where the one that hadn't moved did.) Might be worth revisiting.
I'm thinking of writing a tool where we can test our various theories for algorithms and just plug in data to verify whether we can find out anything from that particular algorithm. (If there are any edge case mismatches, we'll have to redo it.) I need to think some more on how to simulate KOs to remove the actor from the list of QCK actors. (I want the data entry to be relatively painless.) I'll put some more thought into that before publishing a tool. There would be some sort of simple text format with key words for submission and something that allows you to easily load the actors. (Probably the known QCK enemies would be saved and you'd just need load a slight thing from a text file for the dragons on your side.)
Once this thing is written, it shouldn't be hard to add more theories or tweak current ones to search for a match.
Thank you both for the help! (Sorry for the slow response; still overseas.) Since 2.0 is here now and we have new data, I think it'd be awesome if we can crack this algorithm once and for all, though it will definitely be tough.
@Maki Thank you for retesting with new QCK fodder :) The exactness of 1.5x looks promising, I wonder if it's that way for all enemy QCK levels or if it happened to hit that exact value.
For the Blue Tang Hippogriff, do you remember if it was Blue x2 + Cardinal x1, or the other way around? It's seemed to me in the past that each actor contributes to the turn order and there can be slight differences depending on the number of each QCK.
@RageAsylum I would love it if you could PM the logs please. :) Thank you!
---
I think there could also be a slight difference between having the QCK to start with vs. Haste to reach a certain QCK level. (We'd have to test it with low dragon QCK levels to be able to match the Hasted value.) I have no written proof, but I seem to remember this based on observation. (There are cases where I'll Eliminate the same-QCK enemy that hadn't moved yet vs. one that just moved, and the thing that just moved didn't get another turn where the one that hadn't moved did.) Might be worth revisiting.
I'm thinking of writing a tool where we can test our various theories for algorithms and just plug in data to verify whether we can find out anything from that particular algorithm. (If there are any edge case mismatches, we'll have to redo it.) I need to think some more on how to simulate KOs to remove the actor from the list of QCK actors. (I want the data entry to be relatively painless.) I'll put some more thought into that before publishing a tool. There would be some sort of simple text format with key words for submission and something that allows you to easily load the actors. (Probably the known QCK enemies would be saved and you'd just need load a slight thing from a text file for the dragons on your side.)
Once this thing is written, it shouldn't be hard to add more theories or tweak current ones to search for a match.
@Maki Thank you for retesting with new QCK fodder :) The exactness of 1.5x looks promising, I wonder if it's that way for all enemy QCK levels or if it happened to hit that exact value.
For the Blue Tang Hippogriff, do you remember if it was Blue x2 + Cardinal x1, or the other way around? It's seemed to me in the past that each actor contributes to the turn order and there can be slight differences depending on the number of each QCK.
@RageAsylum I would love it if you could PM the logs please. :) Thank you!
---
I think there could also be a slight difference between having the QCK to start with vs. Haste to reach a certain QCK level. (We'd have to test it with low dragon QCK levels to be able to match the Hasted value.) I have no written proof, but I seem to remember this based on observation. (There are cases where I'll Eliminate the same-QCK enemy that hadn't moved yet vs. one that just moved, and the thing that just moved didn't get another turn where the one that hadn't moved did.) Might be worth revisiting.
I'm thinking of writing a tool where we can test our various theories for algorithms and just plug in data to verify whether we can find out anything from that particular algorithm. (If there are any edge case mismatches, we'll have to redo it.) I need to think some more on how to simulate KOs to remove the actor from the list of QCK actors. (I want the data entry to be relatively painless.) I'll put some more thought into that before publishing a tool. There would be some sort of simple text format with key words for submission and something that allows you to easily load the actors. (Probably the known QCK enemies would be saved and you'd just need load a slight thing from a text file for the dragons on your side.)
Once this thing is written, it shouldn't be hard to add more theories or tweak current ones to search for a match.
@Sylvandyr
If I remember right, the "hadn't moved vs had" thing shows up pretty frequently with the psywurm waves in the mire. I do not believe that depends on haste, at least in that case.
If I remember right, the "hadn't moved vs had" thing shows up pretty frequently with the psywurm waves in the mire. I do not believe that depends on haste, at least in that case.
@Sylvandyr
If I remember right, the "hadn't moved vs had" thing shows up pretty frequently with the psywurm waves in the mire. I do not believe that depends on haste, at least in that case.
If I remember right, the "hadn't moved vs had" thing shows up pretty frequently with the psywurm waves in the mire. I do not believe that depends on haste, at least in that case.