Mastery Points

[Anonymous]

@pointless
Yeah, I mean, I don’t usually actually pay attention when I autoplay these levels. I just have my phone nearby, so I see when the reward wheel appears in my peripheral vision, so I’m not missing much time between battles. I only paid close attention to times yesterday for information gathering purposes.
The only thing I grind manually is the arena. (Well, events too I suppose, if that counts as grinding.) Certainly, 13-5 and 17-5 are going to be best for manual mastery grinding, but that’s not going to get you very far unless you have waaaaay more patience than I do, hahaha!
I did test 9-7 vs. 13-5 for autoplay grinding last week, and 9-7 was better per minute. But that was a small sample and I only tested with my Witch/Rogues combo.

[Maxx_Matt]

@poptimus
I prefer auto-play games, too.
My previous post simply try to show numbers, indeed good numbers, for a “played” farming simulation. I’d like to make another step, now, playing for the same GOAL during autoplayed games and are if results are comparable or largely different.

I’ll show the subsequent “game frame” as soon i can, tomorrow, after having it played.

I edited the previous post, correcting form and grammar.

[burbman]

@maxxmatt – I tried your method of farming, and while very good for gathering snoutlings, it really did not provide much mastery when compared with what I can achieve using other methods. In fact, I think your mastery assumption is overstated by quite a bit, as mastery is awarded on the defeat of pigs, and your model only estimates the defeat of a few hundred pigs over five and a half hours. My only use for snoutlings is the dojo, and occasionally enchanting an item, and I can gain more than enough of them using the autoplay methods to keep up with the availability of dojo offers.

If I were to autoplay 9-7 for five and a half hours straight, I would improve the mastery of at least two bird classes by a full 10,000+ point level, and gain several thousand snoutlings.

[bchild]

@maxxmatt
Lv 36 pigs do give 8 mastery points.

@burbman
I agree. Autoplay of 9-7 for 5.5 hrs would net about 15,000 mastery for two bird combo I’m currently using (Tbird and Pirate). I was averaging 120 mastery every 2.5 mins. Which translates to 2880 points in an hour.

There are five 5-wave levels in cave 9. Has anyone else tried to compare which one gave more mastery? Snoutling earnings I figure are about equal. In my test runs 9-7 was slightly better than the rest.

[Sutekh137]

@poptimus

@pointless

@maxxmatt

@burbman

@bchild

pop, thanks for the help, sure, any data points are good. However, I am doing a pretty strict data set with mine, keeping track of how many classes and birds I have maxxed out and keeping running totals (with error bars) as I go. My spreadsheet is here:

https://docs.google.com/spreadsheets/d/1bbpM4fUkhjb_w6bZMnQQi6zT-MKBcCMkJMQmCKXl-P8/edit?usp=sharing

Should be fairly self-explanatory. The blue and red columns are average snoutling buys (currently 5.35) and error +/- (currently 0.447) as a running total. Right now the running total works well because number of maxxed classes and birds will be zero for a while longer (at least a few more days — may have Seadog and Wizard maxxed by end of week). I might hold off on that to get up to 100 data points of “pure” numbers, as 100 data points divides standard error down by a factor of 10.

Once I have a lot of “pure” data points I will start maxxing out classes, but be sure to NOT max out any given bird. I want to see if snoutling offers go below the calculated average as classes are maxxed out (fewer snoutling offers available) but the number of LC offerings remains the same.

As far as Cave 9, etc. — I’ll just keep my head down and grind. *smile* With as poorly as the last event went (I burned all my FE and several stamina and still only got 1 star — just awful) I am not holding out hope that I get much mastery from events at all. Going to have to do it via grinding and the Dojo. I rarely miss stamina usage, yet competition is fierce even around the 2-star threshold. I’ve never seen that before. I guess Rovio’s money grab is working, as there still seem to be plenty of folks fighting for scraps…

Thanks,
sutekh137

[burbman]

@sutekh137 – Interesting chart, Do you want data points to add to it, or should I track my own experience separately and we can compare end results?

I have been tracking my own results via a tally sheet bell curve, so I do not have them in order, but do have a track of how many times I have gotten each number of offers.

I just recreated your model, and came up with the following values for my last 49 dojo trips

AVG – 6.24
Med – 6
Mode – 7
Max – 13
Min – 2
St Dev – 2.7198
St Error – .3885

[Maxx_Matt]

consider me in for these spreadsheet Daily updates for sure, if you want to add parallel data to yours. let me know. my email massimo(dot)Mattioli(at)Gmail(dot)con

we can start reasoning all’ together around data models to apply, too

[Sutekh137]

@burbman

@maxxmatt

Separate sheets are probably fine, though feel free to make a copy of mine so we could later just paste stuff in. That means keeping track of number of maxxed classes and number of maxxed birds to stay consistent. Early on, though, we will all just be gathering data points on the “no maxxed” scenario to get a baseline (what I call the “pure” snoutling-offer average).

At this point I think snoutling offers are closer to 5 than 6, but I shouldn’t second-guess using sparse statistical evidence. *smile* EDIT: Just was your edit — 6.24, eh? And you have no birds maxxed out? I guess I have simply been having a tough run of Dojo trips for my average to be that much lower…

Thanks,
sutekh137

[burbman]

@sutekh137 – I also tracked 314 trips last round of grinding, and came up with the following:

Avg – 5.78
Med – 5
Mode – 4
Max – 17
Min – 0
St Dev 2.5264
St Error – .1426

[burbman]

@sutekh137 – You may be having a rough run of it, but a couple trips in the 10+ range will bring your numbers up pretty quickly. Again, I do not know how my averages looked over time, as I just input the data from my tally sheets.

[Sutekh137]

@burbman

Yeah, a 17 would be nice. *smile* I am at 5.37 after a recent 6, so I’m really not that far out from approaching 6 or so… Absolutely no reason not to trust your large data set. I will waste no time in maxing out classes once I hit 100 data points or so, as that is where the information will become more interesting.

Thanks,
sutekh137

[Anonymous]

@sutekh137
I’ve got a similar spreadsheet of my own going. I’ll modify it to fit with yours and will share it with you once we have a useful amount of data with respect to offers with some maxed out classes/birds. Shoot me an e-mail at gschenendorf[at]gmail[dot]com and I’ll reply accordingly.

My data is also focusing on % of Snout offers vs. % of LC offers after the initial round of offers, and so far I’ve noticed a pretty stark contrast: Roughly 85% of the initial offers for me have been Snoutling offers, but then only roughly 57% of subsequent offers.

For what it’s worth, for now: In 18 dojo visits since I started tracking, I’ve had an average of 5.89 Snoutling offers per visit, with a max of 10 and a min of 2.

[Maxx_Matt]

@burbman

@sutekh137
@poptimus

@pointless

@bchild

Avg # 5.8333
Med # 6
Mode # 7
Max # 9
Min # 3
St Dev # 2.2286
St Error # 0.9098

these data started from two data I’ll keep track of dojo trips until classes maxed out their mastery level.

I’m keeping track of the new factor underlined by @poptimus.
I’m sure that dojo offers are dependent by a “random” function dictated by a misteriosi “seed”. Not sure if our work Will evidence something robust or objects e, but It is worth a try.

Other factors could be:
– missing birds classes (i don’t own seadog, stoneguard and frostsavage) so my offers should bè.more streamlined
– should we take notes of the classica frequency among the offers available?
– is it a time-related preference (maybe nighty offers could differ from Morning ones).

[burbman]

Hey all, enough of this math stuff! We are two weeks into the current grind, How is everyone doing? Gotta keep morale up, and recognize progress as it is being made. This middle part of the grind is where it is easy to lose sight of the progress that is being made, and focus on just how much needs to be done.

Personally, I have achieved 80 of the 240 mastery upgrades that I actually am working on.

Red has all classes at M73
Chuck has four at M74, and two at M73
Matilda has all classes at M73
Bomb has four at M74, and two at M73
And the Blues have managed to find enough mastery to get one class to M73 and the other five to M72 despite my ignoring of them as a playable class for the time being.

[burbman]

@maxxmatt – in your case, where you do not have all of the classes, I would treat the classes that you do not have as being maxed out on @sutekh137’s spreadsheet. Essentially you are giving us data points for what the model will look like when we have some classes maxed out, but no full families. My hypothesis is that you will see a (slightly) lower number of snoutling offers as you have ~10% fewer than a full set of classes, but still have all five families plus the LC for all offer. The hypothesis presented by @poptimus would be that you would have exactly the same chance at snoutling offers, as the determination of snout vs. LC would be set prior to the game generating the specific offer.

By the way, how many data points do the numbers you put up represent?

[angryde]

@burbman

Amazing progress, mate, grinding all those puppies up to the mid-70s… already!

I’ve really gotta get my act together… I’m lagging woefully behind! XD

[Maxx_Matt]

@burbman
I embrace your idea: not having three classes can be compared to have Them maxed out. Cool, for statisticamente porpouses. :-)

While I’m grinding for maatery since january, I kept track of values only for a week, with only 31 data points. I’ll update them in a Daily basis and I’ll keep you in the loop.

This Caves’+Dojo strategy, is paying well. I had at least One or two maatery level gained only in this brief span of time.

I can assure you that my minimum “hystorical” Dojo offer is 0 while maximum Number have been 12. They occurred only once.

[Anonymous]

FYI, for those who may care, I’ve decided to max out my Witch and Rogues (and maybe another class or two) in order to start getting dojo info when I have less than 30 classes left to go sooner than later, to see if/how offers are distributed differently from when I have all 30. It’ll waste some mastery I win from events, but whatever. It’s not like I won’t make it to all-M80 before the next level cap increase anyway. Might as well collect some useful data on the way!

[Sutekh137]

@poptimus

I plan on doing this in a big way, getting at least 8-10 classes maxxed out across four birds (all but Matilda — I still want to benefit from saving a last bird to max out all of its classes at once at the end of this).

I am trusting the data we have for when nothing is maxxed yet, so the whole point of this (for me) is in seeing how snoutling offer probability changes in these other scenarios.

Thanks,
sutekh137

[burbman]

@Poptimus, @sutekh137 – I will take the opposite tack, and begin working to level full families up at (nearly) the same time. That way I should be able to get good data for 4, 3, and 2 full remaining families to go with the pure data.

[burbman]

Three weeks into the current grind, with an event on the horizon that for those lucky enough to get onto a winning team has the potential to level up two full bird families, plus half a level for everyone else. A great chance to make up for lost time in the process. As stated above, I have shifted tactics to try to level full families up one at a time, and chose Bomb to be my first bird to max out. As of now, I have achieved 33 of Bomb’s level increases, and should be looking at reaching max for his classes in a couple of weeks.

Overall, I have obtained 107 of the 240 mastery increases that I am working on, plus 16 for the blues from the dojo for a total of 123 out of 300.

Red – Three classes at 75, two classes at 74, and one class at 73
Chuck – all classes at 74
Matilda – all classes at 74
Bomb – Three classes at 76, three classes at 75
Blues – Four classes at 73, two classes at 72

[burbman]

@sutekh137 – I just took a look at your table, and it appears that our dojo data is converging nicely. I now have 106 “pure” data points.

Avg 5.80
Med 5.50
Mode 4.00
Max 13.00
Min 2.00
St Dev 2.4628
St Err 0.2392

[Sutekh137]

@burbman

Aw, you kept a link to my sheet. *smile*

Yes, my numbers have gone up, but apparently my data set is a little more wild — I don’t think I will get down to that standard error bar within 22 more data points.

But yes, we are close. I think the “truth” is somewhere below 6.0, so not sure if they tightened up on that or if the rough 6.5 figure from previous data gathers was just incomplete or taken in error (i.e. not “pure” data points with no classes maxxed).

Thanks,
sutekh137

[Anonymous]

@burbman @sutekh137

Brief update on my research so far, plans for future research, and current mastery levels. There’s also a question for the two of you in the planning section:

– In 78 dojo visits, I’m averaging 5.68 Snoutling offers per visit. Initial offers are 82.9% Snoutlings, and replacement offers are 56.2% Snoutlings.

– I’m not terribly far away from maxing out Paladin, Stone Guard, Lightning Bird, Thunrderbird, Witch, Rogues, and Skulkers. I’m then going to continue collecting the same data as above, and will also start collecting data on which birds are getting the Snoutling offers, to see if/how the number of maxed out classes for each bird affects the probabilities. I’ll have three birds with two maxed classes, Matilda with one, and Bomb with none, and it will be that way for a long time. (I’m going to be careful to not grind any other classes up to 80 for as long as can, both to maximize future event rewards, and to maximize my data sample size.)
Hopefully I’ll have a large enough sample when I’m done to show whether or not there are any differences between which birds/classes get Snoutling offers after some have been maxed out already. As previously discussed, @burbman’s theory has been that each maxed class diminishes the overall chances of Snoutling offers proprtionally. My theory is that it doesn’t diminish the chance of Snouts being offered, and that the remaining classes should continue to get offers on an even basis, so I’ll get relatively more offers for Bomb than for all other birds, and more for Matty then Red/Chuck/Blues. Or, maybe the data will instead show that once “Snout” is determined, it’s then evenly distributed by bird, such that I’ll get more offers for each of Red/Chuck/Blues remaining classes than for Matty’s and Bomb’s, but that the offers per bird will add up toughly evenly.
Regardless of the outcome in terms of whose theory was right or whatever, I’m just looking forward to knowing how it actually works. Hopefully I’ll have enough data to actually answer that question!

– My question for you two guys (and anyone else who wants to chime in, for that matter): Should I max out all of one birds’ classes sooner rather than later, to see how it affects things? Obviously LC offers for that bird won’t be an option anymore, so it could theoretically increase the Snoutling offer ratio. I’m leaning towards doing this, actually.

– Current mastery ranks:
Red: 72/71/71/71/76/76
Chuck: 72/76/72/72/78/72
Matilda: 71/71/72/71/71/78
Bomb: 71/71/71/72/71/72
Blues: 72/78/72/72/76/72

All 71s are very close to 72, except Avenger for some reason. Wizard is pretty close to 73 as well.

[burbman]

@poptimus – I am currently working toward maxing out all of my bomb classes (3 @ 77 & 3 @ 76 right now, and will re-spin for 5,000 all + 5,000 bomb from the event – my team is up by ~20,000 points right now), at which time I will begin gathering data on only having four bird families available, then three and two in time.

The one thing that concerns me though, is that I will still be dealing with small sample size issues with the lesser number of families. I currently have ~130 data points in my “pure” log, and am averaging 5.87 snout offers per reset, @sutekh137 is averaging about the same with around 100 data points, but you are approx. two tenths of an offer lower with a similar data set size.

When I combine my data with @sutekh137’s, I come up with a standard deviation of about 2.5 which is large enough to mask any statistical variances between all but the most extreme examples we will be able to gather before eventually maxing out the entire flock. Even if we can get 100 data points with each potential scenario, the changes are simply not going to be significant enough to rise above the noise.