You may have read my previous post about my raid stats. This is the next edition. As costumes are now available, it felt appropriate to close another tracking period, analyze the data, and see what I can learn.
Raid Stats
August 26, 2019 - November 25, 2019
Record: 1,995-126, a 94.1% winning percentage (up from 93.1% in the previous period!)
~23 raids a day; one wonders if I am sleeping enough.
Attacking Configuration
Configuration
Previous Period
Current Period
Mono
0%
0%
4/1
4.1%
10.2%
3/2
57.5%
35%
3/1/1
38.4%
54.7%
Observations:
I may take a deeper look at why my 4/1 usage went up, especially since Iām winning only 90.7% of those raids.
In the previous period I noted that I was clearly underusing Gravemaker. One way I addressed that was to put him in more 3/1/1 raids where he and Mother North were paired with another elemntās 3tack.
I win more in 3/2 than 3/1/1 (95.4% and 93.8% respectively). Improving the minor stack seems smart, but I suspect Iād lose too much by pulling my single most powerful hero (Mother North) and āforcingā a 2tack. Itād be different if I had an even close to comparably powerful hero in whatever element. For now I regard this difference as small but am making a note that itās an area I may want to revisit in the future.
Defense Tank Color Prevalence
Color
Previous Period
Current Period
In Top 100
Red
17.3%
20.2%
14.34%
Blue
31.0%
32.4%
16.3%
Green
8.6%
8.3%
2.85%
Yellow
19.2%
18.3%
27.79%
Purple
23.8%
20.8%
38.71
The table shows how often I battled a tank of each color and the share of the Top 100 each color has, per Titan Mafiaās most recent reporting period.
Observations:
Iām not sure thereās much new here. I continue to readily tackle blue teams because of Mother North. As I noted last period, green tanks are rare, and are probably increasingly so as you move into the Top 100.
Anecdotally, Iād say I saw fewer Guinevere tanks, noticeably so. N.B. Iām not saying I fought fewer - Iām saying I wasnāt offered as many in my matchups.
Win Rate Against Tank Color
Color
Previous Period
Current Period
Red
90.8%
93.9%
Blue
94.9%
95.2%
Green
98.0%
92.0%
Yellow
94.5%
93.6%
Purple
89.6%
93.7%
Observations:
One of the decisions I made from the previous tracking period was to emblem Kiril and Frida. That appears to have dramatically improved my results against red tanks.
Green merits some explanation. Basically I started thinking, ālol - free ironā every time I saw one. I think I was overly aggressive and foolish, trying to tackle them with weak-ish teams. Once I realized this was happening, I reined it in a bit. I also emblemed a few red heroes. Doing a quick split between the first half of my attacks and the second seems to show this response was effective: 90.6% win rate in the first half, 93.7% in the second half.
My win rate against purple tanks (read: Kunchen) improved quite a bit. I used Li Xiu in my previous period, and replaced her with Jackal +18 for this one. In fact, that shift was the whole reason I closed the previous period; I wanted to easily be able to see Jackalās impact.
Offensive Hero Usage
Hero
Current Period
Previous Period
Mother North
100%
100%
Gravemaker
44.8%
17.3%
Seshat
40.1%
24.5%
Evelyn
35.0%
54.9%
Onatel
33.5%
45.0%
Caedmon
32.2%
See Note
Leonidas
32.1%
37.9%
Mitsuko
31.4%
37.6%
Proteus
27.5%
24.0%
Frida
25.0%
23.1%
Kiril
24.2%
12.2%
Grimm
21.0%
22.9%
Jackal
20.7%
0%
Observations:
One of my major goals from the previous periods review was to increase Gravemakerās usage. Obviously that goal was met.
Several core heroes - Evelyn, Onatel, Leonidas, and Mitsuko - saw their usage decline. I think this is because Gravemaker and Seshat took on more prominent roles, but I think the back end of my roster also stepped up (I am basing this partly off of data not shown in the table, when Boldtusk, Wilbur, Sabina, etc. were plugged in).
Caedmonās usage is basically unchanged: in the previous period I was testing both him and Buddy (and ultimately went with him to match Evelynās mana speed) - they combined for ~33% usage in the previous period.
It is clear that I now view a heavily emblemed Kiril as a more viable option, leading to an increase in his usage.
Win Rate by Offensive Hero
Hero
Win Rate
Seshat
96.1%
Scarlett
95.7%
Evelyn
95.4%
Caedmon
95.3%
Onatel
94.4%
Proteus
94.3%
Leonidas
94.3%
Gravemaker
94.1%
Mother North
94.1%
Boldtusk
94.0%
Sabina
93.8%
Mitusko
93.7%
Jackal
93.6%
Kiril
93.4%
Grimm
93.3%
Frida
93.0%
Rigard
92.9%
Wilbur
92.2%
Falcon
89.9%
Observations:
Sabina deserves a mention here! She was replaced by a second Seshat midway through the period, but not before posting a 105-7 record. Very serviceable. And she had no emblems, FYI.
It is clear to me that Rigard needs emblems to keep up - he struggled even with low usage (meaning I only put him out in situations I thought were ideal for him).
This table reflects my hiccup against green tanks. Falcon and Wilbur didnāt have emblems but I was bringing the, to the table; they dragged Boldtuskās win rate down. Scarlett only took the field once emblemed, so her win rate gives me optimism about my red stacks for the next period.
Defensive Hero Usage
Hero
Usage
Lianna
21.4%
Marjana
18.1%
Sartana
17.2%
Kunchen
16.7%
Vivica
16.6%
Evelyn
16.5%
Magni
16.5%
Joon
16.1%
Gravemaker
14.1%
Onatel
12.8%
Observations:
No conceptual change from the previous period - in fact itās the same 9 plus Onatel replacing Richard (who was 11th in this period).
Again, itās worth noting the prevalence of TC20 heroes, particularly the snipers.
Win Rate by Defensive Hero
Hero
Win Rate
Mother North
12.2%
Zimkitha
10.7%
Poseidon
10.5%
Gravemaker
10.3%
Mitsuko
9.2%
Kunchen
8.8%
Ariel
8.6%
Thoth-Amun
8.5%
Tarlak
8.3%
Kadilen
8.1%
Hel
8.0%
Leonidas
7.3%
Santa Claus
7.5
Boril
7.4%
Observations:
I continue to think that Thoth-Amun and Kadilen are much stouter defenders than many believe.
I required 40 appearances to get into the table, though many heroes look potentially worrisome in smaller samples. One I knew about is Alice - she would actually top the table. One that surprised me is Inari, who would be right behind her. Both were 35 battles so Iām not ready to draw a conclusion yet - just something Iāll keep in mind.
Boril is the highest placing qualifying four star. I have to go pretty far to find the next one: Proteus at 5.9%. Meaning all the four stars other than Boril posted below average win rates.
Thorne posted a sterling win rate against me in the previous period. This time he only notched 2.4% in 164 fights. Itās very possible I wasnāt looking out for him before and, once aware of him, I dealt with him better / sooner this time.
Letās look at the TC20 snipers: Sartana 6.6%, Lianna 5.9%, Joon 5.6%, Magni 5.4%, Marjana 4.7%. Theyāre Diamond worthy, but hardly superstars. Possibly this is one of the many manifestations of power creep.
Action Items
Iām mostly happy with how Iām doing against each element tank, especially because I think I addressed the red vs. green situation midway through the period. But it is worth monitoring. If I dig deeper here itād probably be by specific hero more than by element (e.g., maybe Guinevere has become less problematic now that I have two Seshats, but perhaps I want to improve against Onatel).
Raid configuration is a potential area for more research. There are at least two things I should consider: investigating why Iām losing more with 4/1, and if it is worth āforcingā 3/2, at the cost of benching Mother North or making the minor 2tack weak against the tank.
Rigard needs emblems. I have stopped feeding Mother North to help him, especially now that I have his costume.
@IvyTheTerrible. Iāve got to ask a few questions. Your win rate seems quite high. I just looked at my tower and looked at my raids (neglecting when I was defense). I was 19-8 for a 70%+ win rate. About 24% less than yours. So I wonder what are the differences.
Do you cup drop? I do not. I try to stay as high as possible. I like the challenge and accept that I have less wins.
Do you reroll? I do not. I believe I can beat any team. Even if I see a team very challenging, I go for it. Also, I donāt believe in rerolling for resources. By the time I use ham to reroll, I might as well just battle.
If you lose, do you rematch? I do. Similar to number 2, I donāt flee.
Just digging more into the stats to get a better understanding and hopefully make this thread more worthwhile for others.
Ignoring revenge raids, I have an attacking win rate of almost 91% (1132 raids, so a 95% confidence interval of about 89%ā92%).
I donāt cup drop.
I donāt reroll, but if Iām filling a chest, I might forego a particularly challenging raid and take a revenge instead. (For the particularly challenging raids I do take, I have a win rate of 82%, so if I took all, my total success rate would indeed have been lower. I donāt have an estimate at hand though.)
I donāt rematch. If a defense defeats me, thatās fair.
That is why your name was @AirHawk and not AirDataš. Opps! It had been changed.
Honestly, other than the amount of money spent when I was a p2w, I never keep track of anything else related to E&P.
Records are mean to be broken, losing streak means improvement is needed.
Then again, itās just a game and I just want to enjoy playing during my spare time.
Kudos to @IvyTheTerrible for having all the patience to records all those data. It wasnāt an easy task for most.
I guess anyone not into mono is probably having more patience.
I do NOT cup drop (challenge you to a duel, demand satisfaction, etc. etc. ). However, I would feel Iām being disingenuous to not talk about my defense. Itās Mother North+13, Gravemaker+8, Aegir+0 (but maxed), Onatel+0 (again, maxed), Seshat+ 9. I consider that a pretty weak defense for the Diamond Arena. I know that I personally find it very inviting if I ever see a tank with no emblems. So my cups probably hover lower than yours. Counterpoint: it is the best defense I can field right now and does keep me around typically 2,500-2,550 overnight.
Reroll is a tricky question. I do not reroll by using the reroll button. Like you, I find that a waste of resources. But I do like to clear my tower revenges, to the extent I have flags (again, demand satisfaction). So, when presented with a matchup I can either fight them or do a tower revenge. After that fight, when I check my next matchup I have effectively rerolled. But for me itās really more about the fact that I want to hit my tower revenges anyway.
Yes, I do rematch. Unless circumstances donāt allow it: e.g., Iām now out of flags and the next time Iām online I try to do the rematch but that person is online; Iām not going to wait to catch them later - I just do a tower revenge instead and am sad about it. It did take me three tries against a few defenses in this sample. And I donāt flee from raids. Only war before a heal / revive / field aid.
@FrenziedEye, Iām just vastly inferior with mono, compared to a 3tack. I think thatās also probably a factor in my 4/1 struggles, which is why Iām hesitant to investigate more. Itās likely more about me as a player than about an inherent weakness of the configuration. Also, in many / all elements, putting a fourth and fifth matching element hero on the field is a steep drop off from the heroes Iām removing.
I have used mono some in raid tournaments (where you almost have to, at times), and I see how it requires different plans and moves. But Iām far from mastery. I remember @Halifax (well known mono advocate) saying it took them a few months to get good at it. Iād probably have to commit to an immersive mono only program for a while and just eat the shallow learning curve.
Tracking
@AirHawk, I screen shot the Vs. screen before each fight. I then take a shot at the end IFF itās a loss. Then I batch those into an Excel file, recording the configuration, each team power, the heroes used on each side (but not level or emblems), and the result.
I have tabs set up to crunch all that data and display everything I presented in OP. It really doesnāt take that long to enter the data now that Iāve already written the formulas. I do occasionally need to add new heroes, but that doesnāt take long. Upkeep is really minimal; I definitely spend more time exploring the data in new ways, which is really fun for me. It also took me way way longer to format the post than it did to sort the data I needed in Excel!
I may eventually migrate it to Access, since thatās really a better platform for what Iām doing, but Iām kind of meh in Access, whereas Iām very strong in Excel.
I just batched yesterdayās raids into my file: 10 minutes. And that included adding a new hero, a quick error check that the process went correctly (which is, of course, automated - I just have to look at the error check fields I wrote), and a brief perusal of the data (I want to closely monitor how my Costumed Rigard is faring). Plus, I was listening to podcasts the whole time.
Point being, now that I built the infrastructure, the time investment is minimal. Iām sure I spend more time on the forum in a given day.
Thatās an interesting approach. I may try that. I typically go through all revenges. If someone is on, I do a regular raid (as I like them in order) but check when they leave. I then revenge again til no more. Then regular raids.
Thatās a good percentage you have there. Iām not that high but havenāt tracked enough to have good stats (maybe I am that high - doubt it though).
I forgot to include this. I thought itād be fun to post pictures of the two highest team powers I faced. Both were wins on the first attempt. I benefited greatly from Mitsukoās reflect in both fights. I also think both teams are using an unconventional tank; I would re-arrange both if they were my defense.
Iām aware of the rule against naming and shaming - which I feel Iām clearly not doing here - but Iāve never been completely clear on the community accepted best practice, so I erred on the side of caution. It feels different to me if I post in Fighting Familiar Names, since I (kind of) know that person.
Youāre not the only one who errās on the conservative side. There is no defined rule so I think you have to go by the ādo what makes senseā rule. If anything, Iād say youāre praising them - āWow, look how strong they are. Iām happy I could beat them.ā
This is just my personal opinion. Hereās what I posted earlier on this topic.