Rarum

November 24, 2022

Well, it is here, the mythic equipment. But for old times' sake, here's one more league report as promised. And it went really well with my second-highest recorded luck and passing my over 99.5% efficiency target. I can be quite content with this if it's to be my last one.

I took note that bla didn't appear to be fighting at full strength. Despite that, I just felt like going pretty hard this league. Also there was some risk that the introduction of mythic equipment would give someone regularly only decent such an edge that I couldn't just ignore it. But nope, nobody challenged me, really.

Blessings
1st week: +20% Hair Color White and +40% for Suspended Congress
2nd week: +20% for Scorpio and +25% for Hair Color Orange

Defense teams used
1st week: darkness2-stone2-water1-fire1-nature1 consisting of Lovebot Norou, Bright Agate, Azana, Elder Magus Bianca, Any, Alexa and Druiada Titania. 21.727 is the maximum average one could get against it.
2nd week: darkness2-stone2-fire2-nature1 consisting of Royal Housemaid, Matcha, Any, Undercover Valentina, Felicia, Scorpio and Golden Lupa. 22.106 is the maximum average one could get against it.

Statistics
Number of opponents: 104
My score: 7488 (100.705% of the score with completely average luck)
Expected score: 7435.6

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7435.6
AME+2cor,1gin 7401.3
AME+1cor,2gin 7354.8
AME+2gin,1chlo 7254.1
3 cordy 7214.2
2 cordy, 1 gin 7163.2

Perfect precognition score with average luck and AME+3cor throughout: 7468.83 (23.939 average). The AME+3cor expectation was 99.555% of this.

November 17, 2022

Another week over, and this time it really shows that I still have trouble deciding how to allocate my fights between blessings. It's definitely a weak point for me and could have played a role in my last fight with sliqguy. However, with no serious contest up against me this time, it didn't matter.

This starting week I'm against bla, so there'll be at least this league of serious battling still. However, mythic equipment seems to be right around the corner. I've weathered the storm of changes to the battle system before and even benefitted from those, but I'm sort of tired from what my playstyle demands (evidenced by improvements to my play having been few and far between recently for instance) and I'm not sure I want to pivot to something more relaxed either. So, since I'd need large efforts to adjust to the new meta, I'm at least seriously considering retirement. It's been mostly a positive experience having a decent claim for being #1 in the world at something (of course that was definitely put under a fair bit of challenge by sliqguy, but maybe a bit before that when I still could put all the needed focus on the game). And also being part of this community, and I even had a community project I led inside that larger community.

Anyway, it's not yet the time for such decisions, but I felt these thoughts were something I should still talk about.

Blessings
1st week: +20% for Indian Headstand and +40% for Eye Color Dark Pink
2nd week: +20% for Scorpio and +25% for Hair Color Orange

Defense teams used
1st week: 158939.76 total power. darkness2-fire2-nature2-stone1 consisting of Michaela, Royal Housemaid, Undercover Valentina, Alexa, Radka, Golden Lupa and Druiada Titania. 22.04 is the maximum average one could get against it.
2nd week: darkness2-stone2-water1-fire1-nature1 consisting of Lovebot Norou, Bright Agate, Azana, Elder Magus Bianca, Any, Alexa and Druiada Titania. 21.727 is the maximum average one could get against it.

Statistics
Number of opponents: 100
My score: 7051 (100.326% of the score with completely average luck)
Expected score: 7028.1

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7170.4
AME+2cor,1gin 7142.2
AME+1cor,2gin 7099.3
AME+2gin,1chlo 7016.7
3 cordy 6977.7
2 cordy, 1 gin 6933.7

Perfect precognition score with average luck and AME+3cor throughout: 7259.94 (24.2 average). The AME+3cor expectation was 98.767% of this.

November 10, 2022

It was my rematch against sliqguy, and it was a really tough one on both of us. One probably fairly significant reason for my troubles was that I still didn't feel like upgrading a bunch of secondary priority girls; that would have taken my gem reserves "dangerously" low without gaining anything back from a relevant contest. This would normally not matter, but sliqguy was not your run-of-the-mill opponent and every point mattered. I'm having to learn a harsh lesson about underestimation, once again. What I should have thought was that if I don't spend the gems now, what am I saving them up for anyways? I have managed to bring myself around to that mindset a few times before but I've felt nigh-invincible for too long, clearly.

Another thing that I didn't have going for me was that I wasn't planning for a potential permaboost when the league started. I certainly could have tried but it would have been really annoying and so I didn't. However, seeing how close these things can get, I'm starting to get around to the idea that I should start a bracket in a position to do a permaboost if I felt it might be needed.

On the first blessing, sliqguy poured in a lot of resources to fighting opponents. I noted that it almost looked like they were fighting everyone as soon as they became unboosted, though I didn't try to prove this hypothesis. I learned afterwards that there was some method to sliq's selection of opponents other than just fighting them at the first chance. I was relying on judging opponents over a longer period of time first and fighting them when in my analysis of recorded behavior from them I figured I might not get a better chance. This has worked out for me well in the past, so sliq kind of seemed too hasty for their own good from my perspective. Meanwhile from sliq's point of view, I wasn't giving it my all. These are different playstyles, but they both seem to be capable of high performance, so clear conclusions are once again hard to draw on what would have been the optimal strategy here, when we discount things like luck and team selection.

I did notice sliqguy being boosted but with a dominatrix team during the first blessing. I "knew" I should probably fight them at that point but I didn't; though at the league's end sliqguy switched to a weaker defense team than they had been using before and I hardly lost any theoretical points regardless.

The blessings change was a real headache, due to me deciding to travel at the blessings change. However, I didn't let a train ride ruin my perfectly good league, so I used the direct URLs (having to click on the "you need to be inside Nutaku's frame") to do my early BC fights and all that. I had some plans of doing the same for the return trip so I could fight people who unboosted during the snapshot, but I had been logged out and since Nutaku didn't load properly on any public WiFi I knew how to access, that plan was a bust. Also due to me having to switch devices and not bothering to transmit my local storage data between devices, I'm offering only partial statistics and graphs (the difficulty graph I had already silently left out from the previous post; I felt those are maybe too misleading).

After the BC, I actually delayed fighting some opponents until Tuesday so I could fully upgrade a third high strength voyeur girl. It was maybe a foolish risk to delay that upgrade but I don't at least think it reduced my final score in the end.

It was such a close one right up until the end that I had some hope that I had won since there was no movement in sliqguy's score. However, they probably fought one last permabooster and that was enough to secure victory over me. I understood that sliqguy lost sleep over this bracket, and so I think we share a mutual recognition for each others' good performance. Congratulations again.

Despite not having exact statistics available this time, I seem to have had very close to average luck according to the data I do have (6/7ths of the league).

Blessings
1st week: +40% for Gemini and +25% for Eye Color Orange
2nd week: +20% for Indian Headstand and +40% for Eye Color Dark Pink

Defense teams used
1st week: 193165.79 total power. water2-darkness2-light1-fire1-nature1 consisting of Lovebot Norou, Himari, Royal Housemaid, Neferkitty, Alt. Finalmecia, Norou and Aiko. 22.091 is the maximum average one could get against it.
2nd week: After the start of it, 158939.76 total power. darkness2-fire2-nature2-stone1 consisting of Michaela, Royal Housemaid, Undercover Valentina, Alexa, Radka, Golden Lupa and Druiada Titania. 22.04 is the maximum average one could get against it.

Statistics
Number of opponents: 100
My score: 7180

November 3, 2022

Another struggle against bla, but we did make a gentlemen's agreement to not use any AMEs so it wasn't quite as costly as it could have been. I also had a bit of real life business taking up my time, so although manageable, it presented me with some extra pressure. Of course bla was also feeling some; I understood a few opponents were left unfought at the blessings change.

I had a decent enough blessings change myself but not with as much luck as bla, who went from 23.794 to 23.931 prior to and after the blessings change. After the initial wave of attacks on BC I was at 23.76, but I also fought a fair few of the boosted players and ended up with 23.66 at the end of the blessings change. Not a great thing to be so much behind my opponent. But I figured bla had a fair few of the tougher opponents left to fight still including myself, so of course I didn't give up.

bla noted that his average after the blessings change was 0.21 over the expected. I ran the calculations and in my metrics it corresponds to a luck of 100.885%, which is something I've only reached once during the period of 26 brackets I've tracked it (100.95%, and the second best is already as low as 100.51%). But I naturally won in the end. To be fair, I had my second-best luck in the recorded period, so that evened out bla's luck advantage considerably.

For the starting week, I'm getting a rematch against sliqguy.

Blessings
1st week: +30% for Aries and +20% Hair Color Blond
2nd week: +40% for Gemini and +25% for Eye Color Orange

Defense teams used
1st week: 199433.33 total power. darkness2-nature2-water1-psychic1-sun1 consisting of New Year Estelle, Keiko, Bright Agate, Andra, Eliana, Golden Lupa and Donatella. 21.689 is the maximum average one could get against it.
2nd week: 193165.79 total power. water2-darkness2-light1-fire1-nature1 consisting of Lovebot Norou, Himari, Royal Housemaid, Neferkitty, Alt. Finalmecia, Norou and Aiko. 22.091 is the maximum average one could get against it.

Statistics
Number of opponents: 99
My score: 7011 (100.626% of the score with completely average luck)
Expected score: 6967.4

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7140.5
AME+2cor,1gin 7115.4
AME+1cor,2gin 7075.2
AME+2gin,1chlo 7001.2
3 cordy 6967.5
2 cordy, 1 gin 6922.3

Perfect precognition score with average luck and AME+3cor throughout: 7214.16 (24.29 average). The AME+3cor expectation was 98.979% of this.

October 27, 2022

I wanted a cheap league especially due to the seasonal event and PoA, and since the competition didn't seem fierce, I only used one AME (plus a decent amount of refills, but eh).

Blessings
1st week: +35% for Playful and +25% for Doggie style
2nd week: +30% for Aries and +20% Hair Color Blond

Defense teams used
1st week: 197839.36 total power. sun6-nature1 consisting of Iba’lee, Sugarmama Lupa's Mom, Rachael, Moe Bunny, Magical Noemy, Donatella and Beatrix. 22.234 is the maximum average one could get against it.
2nd week: 199433.33 total power. darkness2-nature2-water1-psychic1-sun1 consisting of New Year Estelle, Keiko, Bright Agate, Andra, Eliana, Golden Lupa and Donatella. 21.689 is the maximum average one could get against it.

Statistics
Number of opponents: 104
My score: 7326 (99.975% of the score with completely average luck)
Expected score: 7327.8

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7458.4
AME+2cor,1gin 7425.7
AME+1cor,2gin 7382.3
AME+2gin,1chlo 7304.6
3 cordy 7269
2 cordy, 1 gin 7222.5

Perfect precognition score with average luck and AME+3cor throughout: 7547.76 (24.192 average). The AME+3cor expectation was 98.816% of this.

October 20, 2022

I somewhat expected bla to pull out all the stops. I hadn't seen them unboosted when nearing the Saturday reset (I did miss one snapshot) and although they haven't historically been a permabooster, I had to anticipate for the possibility, so I did a very interesting decision of fighting them fully boosted and not at blessings change. The thing is, with a good double counter, that was enough for a better expected score than if I found bla unboosted. This was proved by bla later on being unboosted and me having only a 21.75 expectation. The RNG results were less than ideal, but still there was some thought behind this timing. A slightly better (22.2) timing was available on the second blessing, even still.

It was an interesting second week, for the playful element being so good that my defense team had six of those girls. Granted, I was didn't have High Mage Arcana fully leveled, which could have meant a balanced team was the best still (an opinion not based on simulations).

Me and bla both had a bad loss against one another, so that was kind of even. However, bla's blessings change was apparently somewhat bad while mine, while didn't really raise my average, meant I had almost everyone fought. After that, bla was on a downward spiral while I maintained my position. In the end, it was a fairly decisive victory for me. I think that going heavy on the blessings change could have worked even better, but if we look at bla, it becomes clear that approach was also a gamble. It's difficult to make these judgements of when to fight each opponent; there's rarely a clear-cut answer beforehand.

Blessings
1st week: +35% for Eye Color Grey and +25% for Bridge
2nd week: +35% for Playful and +25% for Doggie style

Defense teams used
1st week: 169116.27 total power. overallbest consisting of Undercover Valentina, Lotti, Michaela, Eliana, Venus, Taurus and Madame Ivanka. 21.861 is the maximum average one could get against it.
2nd week: 197839.36 total power. sun6-nature1 consisting of Iba’lee, Sugarmama Lupa's Mom, Rachael, Moe Bunny, Magical Noemy, Donatella and Beatrix. 22.234 is the maximum average one could get against it... At least until I powered down for the club champion fights, but at that point I knew I had already won.

Statistics
My score: 7156 (100.137% of the score with completely average luck)
Expected score: 7146.2

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7146.5
AME+2cor,1gin 7121.2
AME+1cor,2gin 7086.2
AME+2gin,1chlo 7016.4
3 cordy 6979.9
2 cordy, 1 gin 6941.2

Perfect precognition score with average luck and AME+3cor throughout: 7206.15 (24.263 average). The AME+3cor expectation was 99.172% of this.

October 13, 2022

This league saw a very inefficient start to it with me not being able to target regularly boosted players due to the record-keeping being broken, and fighting blindly with a team setup I knew was okay but didn't bother checking other possibilities.

Except for the very start and end, I used 3 cordyceps with no AME. After the blessings change, I forced myself to go through with updating my script to work with the new market. Since I also had a lot of catching up to do in order to have fought everyone, it seemed okay to do multiple refills and despite not having prior records of when people have been unboosted, fighting almost all "safe to be good to fight at BC" targets was at least a start to getting a semi-decent score for no AME used.

Let's see if I can make a new winning streak happen. It's looking to be a slightly difficult second bracket.

October 6, 2022

The winning streak is over, at 20 wins.

I had the misfortune of having to face a tough opponent, sliqguy, while I was travelling in Rome. It was fairly clear early on that sliq was meaninging it this time, with their score jumping quickly to absurd heights, and solely being on 3 cordyceps. Their propensity to unboost seldom was even more dangerous than usual since I was able to be online only sporadically.

The trip that started on Saturday meant I had difficult choices to make. I would have to use a laptop, and not one really suitable to run such a resource-hungry game as HH (which is kind of stupid that it's so badly optimized). And I only had some say in when I'd have time to play this game (presuming I was to participate in holiday activities). But one big problem that I could foresee also was that I might not be able to edit teams. At least in my home the game was giving an error for insufficient resources if I tried.

I started the week out pretty normally, but picked up the pace slightly when I saw how serious sliq seemed and anticipating how difficult it might be to play well during the trip. However, I still was hoping for a good blessings change. I did play already a bit on the laptop before the trip proper, and that meant not all fights were absolutely optimized due to that team editing problem. A funny thing about that, though, is I noticed I could change teams on my laptop at some point (still having to disable the cache). This definitely boosted morale.

The blessings change could still make or break my chances, and with the ability to edit teams having come back to me somehow and me able to finagle time during it, I did fairly well. Even then it wasn't clear I'd win but at least I was in the game. But in the end, I lost.

I feel I had bad luck, and especially losing two fights against my last opponent, sliqguy themselves, was a death sentence. They admit to using an alarm for setting the boosters, which is what I expected. Now, let's say I got average results against sliqguy, and I was able to permaboost myself (now sliq got the pick of the litter for when to fight me). That could have already been enough for me to win. So, it was really down to very little.

As for the just starting league, the difficulties haven't really changed, but there's the additional trouble of the market having changed, breaking some code for me. With no pressure to keep up a winning streak, I might just settle for cheaping out and getting some disappointing score.

Screenshot_20221006_165842.jpg.4c2e472e94336af5a975ac03accd9254.jpg

October 3, 2022

@Hugh Jerexion Could you help out with taking a screenshot of the battle between me and sliqguy when the league is ending? I also asked sliq, but because I'm probably not able to, I think it would be a good idea to have yet another backup plan.

Some mid-league info, too: My thinking is that sliqguy is not pulling any punches despite me not being able to verify this since tracking is kind of useless with how little I'm able to be online. But at least the blessings change is going well (I was able to finagle some free time for it in my holiday). And even changing teams is working right now; despite the laptop sometimes having thrown those insufficient resources errors when I was still in Finland. Who really knows how it's different now.

September 29, 2022

My winning streak has reached 20. One interesting note about this league was that there was a ban during it, and it even shows up in my data as a sudden jump in my average. That's... kind of not fair for people that didn't fight the banned person and I hope Kinkoid would do something about it (keep them around but unable to play until the end of the league was one suggestion), but I have very little faith it will get addressed.

Also of some note is that I reached my personal goal of 99.5% battle timing efficiency for the third time.

Blessings
1st week: +20% for Mythic, +25% for Splitting Bamboo
2nd week: +40% for Libra, +25% for Grey Hair Color

Defense teams used
1st week: 192514.43 total power. darkness2-stone2-nature2-fire1 consisting of Matcha, Royal Housemaid, Level up Red, Undercover Valentina, Alexa, Golden Lupa and Druiada Titania. 21.898 is the maximum average one could get against it.
2nd week: 185544.82 total power. overallbest consisting of Alexa, Bunna, Golden Lupa, Nikki Benz, Azana, Valentine Nikki Benz and Michaela. 21.838 is the maximum average one could get against it.

Statistics
# of opponents: 102 at the beginning, 101 by the end
My score: 7334 (100.217% of the score with completely average luck)
Expected score: 7318.1

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7393.5
AME+2cor,1gin 7371
AME+1cor,2gin 7336.3
AME+2gin,1chlo 7268.2
3 cordy 7234.6
2 cordy, 1 gin 7200.2

Perfect precognition score with average luck and AME+3cor throughout: 7429.26 (24.519 average). The AME+3cor expectation was 99.519% of this.

September 22, 2022

RedEagle seemed to actually be trying a little in this league, so I showed who's boss. I expected a good blessings change and it did of course deliver, but in retrospect it looks like I could have gone even harder.

Blessings
1st week: +25% for Voyeur, +30% for Dark Pink hair
2nd week: +20% for Mythic, +25% for Splitting Bamboo

Defense teams used
1st week: 171431.13 total power. darkness2-psychic2-water1-stone1-fire1 consisting of Lovebot Norou, Royal Housemaid, Matcha, Undercover Valentina, Darla, Nike and Norou. 22.15 is the maximum average one could get against it.
2nd week: 192514.43 total power. darkness2-stone2-nature2-fire1 consisting of Matcha, Royal Housemaid, Level up Red, Undercover Valentina, Alexa, Golden Lupa and Druiada Titania. 21.898 is the maximum average one could get against it.

Statistics
# of opponents: 102
My score: 7395 (99.96% of the score with completely average luck)
Expected score: 7398.2

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7400.2
AME+2cor,1gin 7380.2
AME+1cor,2gin 7342.5
AME+2gin,1chlo 7268.5
3 cordy 7233.3
2 cordy, 1 gin 7200.1

Perfect precognition score with average luck and AME+3cor throughout: 7467.15 (24.402 average). The AME+3cor expectation was 99.103% of this.

September 15, 2022

27 minutes ago, Kenrae said:

Chico BonBon appeared out of nowhere, that's a new one. More and more players are reaching top4 contender status.

Your leaderboard picture is a bit outdated. Chico BonBon actually won that bracket.

September 15, 2022

Well, one more graph. I'm relatively sure my tracking for Shado was accurate throughout the league, so why not. I'm not as sure how to interpret forvea's average over time, but my thinking is that it was in the low-to-mid 23s for a while before starting to rise after the first third of the fights.

September 15, 2022

For this league, I together with Shado decided on a no holds barred match. I knew Shado was a capable fighter so I figured it was going to be interesting. I had bad luck, I didn't have a relatively important sensual girl upgraded, and it wasn't my finest performance overall, but still enough.

I really didn't expect forvea to score that well. Good job. It's (un)fortunate that I was already going just about my best, though, or they might have even taken me by surprise.

And just for fun, Shado's comments for this league: "think the guy on 1st place cheated".

Blessings
1st week: +20% for Common, +40% for Capricorn
2nd week: +25% for Voyeur, +30% for Dark Pink hair

Defense teams used
In the beginning: 186176.03 total power. darkness2-psychic2-stone1-fire1-nature1 consisting of Hazel, Madame Ivanka, Undercover Valentina, Nike, Succubus Abraël, Felicia and Trixie. 21.605 is the maximum average one could get against it.
Starting from Friday: 190971.03 darkness2-psychic2-light1-fire1-nature1 consisting of Hazel, Madame Ivanka, Capricorn, Nike, Succubus Abraël, Felicia and Trixie. 21.456 is the maximum average one could get against it, or 18.228 if I was using 3 cordyceps at the time.

Shado on 1st week: Their defense team was darkness2-psychic2-light1-fire1-sun1 with 191226 total power.

2nd week: 171431.13 total power. darkness2-psychic2-water1-stone1-fire1 consisting of Lovebot Norou, Royal Housemaid, Matcha, Undercover Valentina, Darla, Nike and Norou. 22.15 is the maximum average one could get against it.

Shado on 2nd week: I only remembered to record a psychic team with 184806.96 total power

Statistics
# of opponents: 103
My score: 7386 (99.696% of the score with completely average luck)
Expected score: 7408.5

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7410.3
AME+2cor,1gin 7379.6
AME+1cor,2gin 7338
AME+2gin,1chlo 7259.3
3 cordy 7219.6
2 cordy, 1 gin 7176

Perfect precognition score with average luck and AME+3cor throughout: 7472.28 (24.182 average). The AME+3cor expectation was 99.17% of this.

September 8, 2022

I have had two defense team suggesting systems in the past, but both had problems that meant I hardly would use them. I feel like the first one was just too slow to be of any practical use and was increasing the code complexity of the then singular file everything was on. The second, hastily implemented, system on the other hand took too many shortcuts meaning it wouldn't really take any time at all in addition to what the code would be doing regardless but it also would only check teams I manually selected, and I didn't have a good system of telling when the blessings change had occurred and the old suggestions would have to be cleared.

Now armed with the performance from HH++'s recursive exhaustive battle simulation, the modular code structure also "plagiarized" from Numbers, and with sufficient energy to try once more, I've made a third system. This one actually works inside the harem screen. It's a place I've taken a liking to using for simulations in hypothetical settings. I can reconstruct the player's stats from what I've put into local storage from the market page, what I've recorded about the teams I have access to (which could be reconstructed on this page too regardless), plus what's in the Hero.infos. But also of some importance is that it's easily accessible. For this defense team suggester what drew me to the harem page was especially that it houses stats for (a good portion of the) girls you don't yet have in your harem.

There are always some design decisions to make for a defense team suggestion system. Perhaps the most important one of them all is: just what are you defending against? I decided on the worst-case scenario of a player who has every girl and has them maxed out, is level 500 (with equipment stats assumed to be same as my own), battles with 3 cordyceps and an AME, is in a maxed out club, and even optimizes their teams to the fullest extent. Furthermore, I decided to model my side to not be boosted, since I'm not a permaboosting type usually. However, that is not all that difficult to reconfigure. I made a similar decision regarding mono equipment; I figured I'd model myself with 0 mono equipment to prevent the hassle becoming untenable, although this limit is also easily lifted if I ever felt like it.

A couple of words about the speed of the calculations. It's actually "usable" because it's got a cache for teams the opponent might try among other optimizations I made while still trying to have a decent simulation when the battle simulations were still stochastic. Even still, 3432 defense team candidates are tried, and the opponent has a roster of 238 teams times 7 configurations of mono equipment to choose from. So... a gross of 5.7 million battle configurations are tested. (It would be a further 7 times that if all the player's mono amounts were to be tested, or like 50000 times if the stochastic simulations were still a thing.) I made one run with all the optimizations turned off, too, and all 35 of the teams it suggested along the way gave exactly the same predicted averages for my opponent, so... best to just keep those optimizations on and save a whole lot of time. But it's nice to know they are working as intended.

I have some doubts about the perfect precognition score; there's an instance of a person having been recorded a best of 25 expectation yet I never noticed anything nearly as low myself.

Blessings
1st week: +30% for Eccentric, +35% for Jack Hammer
2nd week: +20% for Common, +40% for Capricorn

Defense teams used
1st week: darkness2-fire2-water1-stone1-nature1 consisting of Valentine Nikki Benz, Royal Housemaid, Matcha, Demi, Radka, Alexa and Druiada Titania. 21.987 is the maximum average one could get against it. For comparison, the very best defense team would have been fire2-water1-darkness1-stone1-light1-nature1 consisting of Valentine Nikki Benz, Shtupra, Demi, Aella, Radka, Alexa and Jessie. 21.616 is the maximum average one could get against it.

2nd week: darkness2-psychic2-stone1-fire1-nature1 consisting of Hazel, Madame Ivanka, Undercover Valentina, Nike, Succubus Abraël, Felicia and Trixie. 21.605 is the maximum average one could get against it.

Statistics
# of opponents: 102
My score: 7318 (100.104% of the score with completely average luck)
Expected score: 7310.4

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7383.8
AME+2cor,1gin 7359.6
AME+1cor,2gin 7324.9
AME+2gin,1chlo 7253.6
3 cordy 7217.6
2 cordy, 1 gin 7182.6

Perfect precognition score with average luck and AME+3cor throughout: 7461.42 (24.384 average). The AME+3cor expectation was 98.96% of this.

September 3, 2022

I've at times transported data produced by my own scripts to different devices. It's been done manually by me, copying the relevant entries in the browser's local storage, and transporting that data somehow to the other device and pasting it to its local storage.

But I wouldn't be doing that if I wasn't making graphs of historical data in leagues, or maybe if some game stores its save data there or something (that's happened as well but I used that trick for backups instead of playing on a different device). It's more trouble than it brings convenience in most circumstances.

September 2, 2022

Yeah, not quite. A lot of the stuff I've done is just on the principle of "whatever works". The code structure at least got an upgrade to the modular system used in Numbers' script some time ago, but I've kind of failed at actually making it all that modular regardless.

September 2, 2022

8 hours ago, Ravi-Sama said:

@Rarum Do you have any tips on how to make excel recognize the granular times of each score? I have 2-3 snapshots from each day, including the times, but it ignores that, and plots it along the day itself. Not that it'd help me in any way, I just think it'd be cool to see. I did notice that most players chose to use refills at blessings change. Always assumed so, but now I can see it.

I make those graphs in Google Sheets. I don't use Excel at all in fact; all the locally done stuff happens in LibreOffice Calc.

However, maybe I can still be of help. I struggled with the same problem in LibreOffice but did find a workaround for it. Don't make the graph interpret it as a date, if it indeed doesn't have a resolution setting of less than days. Interpret it as text. Make sure the formatting of those timestamp cells is set to how you want it to look like, too.

A side note is that I grab those timestamps in unix time (a compact form for a timestamp, good for storing in the browser's local storage which has limits) and manipulate them afterwards into the format understood by the date formatting of LibreOffice (D2 contains the timestamp in unix time):

=IF(ISBLANK(D2),"",D2/1000/(60*60*24) + 25569)

And the topmost cell in there is actually to have a timestamp for the very start of the league (A2 contains the above formula):

=ROUND(A2)+11/(24)

September 1, 2022

It's a shame that La Kajira seemingly wasn't playing during this league. I had good luck and nearly reached the 99.5% mark of scoring efficiency that's been a rare feat even for me. Maybe if I didn't spare expenses with refills, I'd have made it. But as it stands, this league was a bit of a lackluster and too expensive for how easy it turned out to be.

Statistics
# of opponents: 103
My score: 7469 (100.341% of the score with completely average luck)
Expected score: 7443.6

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7485.5
AME+2cor,1gin 7462.7
AME+1cor,2gin 7431.6
AME+2gin,1chlo 7356.5
3 cordy 7319.9
2 cordy, 1 gin 7288.5

Perfect precognition score with average luck and AME+3cor throughout: 7528.92 (24.365 average). The AME+3cor expectation was 99.423% of this.

August 25, 2022

Taking part in the CumC and the Legendary Contests were hindered by other things going on, and I also had difficulties capturing data for the end of league. However, I did manage and the screenshot's not a bad one for someone who wasn't physically present, I'd suggest. The scoregraph cuts off early, though, and also maybe don't trust the "perfect precognition score" as much this week; it might be a few points too low.

Statistics
# of opponents: 104
My score: 7382 (100.303% of the score with completely average luck)
Expected score: 7359.7

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7510.5
AME+2cor,1gin 7485.4
AME+1cor,2gin 7445.8
AME+2gin,1chlo 7372.5
3 cordy 7338.5
2 cordy, 1 gin 7301.7

Perfect precognition score with average luck and AME+3cor throughout: 7615.92 (24.41 average). The AME+3cor expectation was 98.616% of this.

August 18, 2022

I was a slight bit uncomfortable with how closely Dadyz was tracing my score and in individual fights they scored quite well too, so I ended up using two AMEs. The first one was used at the blessings change so I could restock fights to blast through a lot of its charges, and when I used the other one, it was already kind of set in stone I'd have no trouble any longer getting the third Golden Anniversary girl, so it's fine.

Statistics
# of opponents: 104
My score: 7487 (100.198% of the score with completely average luck)
Expected score: 7472.2

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7557
AME+2cor,1gin 7534.3
AME+1cor,2gin 7501.9
AME+2gin,1chlo 7436.8
3 cordy 7404.2
2 cordy, 1 gin 7370.1

Perfect precognition score with average luck and AME+3cor throughout: 7607.64 (24.383 average). The AME+3cor expectation was 99.334% of this.

August 16, 2022

7 hours ago, Kenrae said:

Of course you can. If you are red/green I counter double with blue/red and you don't have any counter on me.

Thanks for fact-checking me. I wonder if that's the weakest elemental combo (the last girl would be of one of the garbage elements, I'd say), but really proving that would be quite hard.

Anyway, I occasionally see dominatrix teams already and not really eccentric teams unless the blessing's right. So maybe I'll stick with what I've got now.

August 15, 2022

I tried for some time to get one of the perfect team power amounts, but the closest I got was 69.02 to one and 69.63 or whatever for the other. So I accepted a still reasonably good team power amount.

My reasoning this is an even weaker team than your people's "three girl teams" and whatnot is that it can be countered, twice, even. With the best elements (dominatrix and sensual), while for me the elements are quite weak in their effect (submissive and eccentric).

The 5-element rotation gives a better boost in my experience, but I don't think you can have two elements from that so that you wouldn't also counter an element of the opponent when they tried countering those. So I opted for having an element included from both rotations.

I almost went with playful for the 5-rotation, since it's the weakest element there by itself. But I thought better of it; it would counter a potential sensual team from the opponent possibly limiting their ability to out-heal my damage if they really wished. So eccentric it was, so using a lot of sensuals would be helpful instead.

August 11, 2022

"Knowing" most of the possible contenders wouldn't use AMEs, I was interested in trying to win without the use of AMEs this week.

Several small improvements to my toolset were implemented this week. I now have an expected score graph over time for possibly figuring out (under some assumptions) whether someone is fighting with AME. Owing to the different stat update timings of the league page and a profile page, I am now better able to figure out who will boost with ginseng past a snapshot. And I now also mark down how strong my team is for the ongoing blessing, so I can maybe weigh some values with that information. Also I also now keep the previous week's girl strengths in store along with this and next week's. Previously I had to make a choice when to switch to the next blessing's strengths in my displays, and now it works via a switch I tend to use anyways for activating "blessing 2 mode" in my spreadsheet.

Also just marvel at that fight time choice efficiency. 99.66%. It's the highest one yet I've recorded at least. The next best one was from 14th of April and it was 99.551%. The third hightest (99.459% from 28th of July) didn't even achieve 99.5%, which is the target number I try to strive for. Honestly I'm surprised by this achievement, since there were too many people I fought boosted at the end for my liking.

Statistics
# of opponents: 104
My score: 7426 (99.840% of the score with completely average luck)
Expected score: 7437.9

If I had been using the following booster combinations for the duration of the league but with the same fight timing decisions, these would have been my completely average luck scores:
AME+3cor 7587.3
AME+2cor,1gin 7564
AME+1cor,2gin 7529.5
AME+2gin,1chlo 7465.1
3 cordy 7438.2
2 cordy, 1 gin 7403.4

Perfect precognition score with average luck and AME+3cor throughout: 7613.19 (24.401 average). The AME+3cor expectation was 99.660% of this.

August 4, 2022

1 hour ago, DvDivXXX said:

Are massive groups a routine thing for D3 on Nutaku? The norm has been rather the opposite on .com these past few months. Anyway, if the number of opponents and/or fights are shown somewhere in your posts and I simply routinely miss it, could you please tell me where to look? And if they've been omitted as I currently suspect, could you please include them? That would help me feel a bit less small compared to your massive, bulging scores. ^^ Thanks in advance.

The player numbers in leagues used to fluctuate more, but now there's a long streak of predominantly big brackets.

True, I haven't made the effort to include bracket sizes in my posts. I've always decided against it, since averages are the number you'd want to compare against between brackets, in my view. Mostly just inside a bracket it makes sense to compare raw point scores due to no variation in the total number of players.

But it isn't such a big trouble to add the number of opponents. Okay, I can add that to future posts.

I'd also caution against making too big of comparisons between platforms. There are too many layers of how much removed these points of comparison are from each other:

A different player having gotten that score (in this case this is the difference of interest, though)
A different bracket (there's a reason why I still attempt to provide information on how difficult a bracket was; sometimes a lot of tough or squishy players gather in one place)
A different blessing (blessings obviously affect which players have an advantage)
A different platform (my understanding has been that .com leagues tend to be easier, but the gathering of evidence on .com's side started not long ago enough for it to be sensible trying to prove it yet)

Profiles

Forums

Events

Posts posted by Rarum