Proposal: LT Tracking Project

Unknown

I have an idea on how we can track the distribution of 4/5* covers through LTs through community reporting. There is much uncertainty about the process and I'd like to put as many conspiracy theories to bed as possible while also improving peoples' confidence in exactly what they're grinding for or spending on.

I've tried this before and the results didn't work out because it depended on inherently inaccurate self-reporting.

Someone suggested that I use the "Last Legendary" thread, which would help correct the accuracy problem. But the data there suggests that there was an over-representation of people who only reported after successful draws.

Instead, I propose that we gather a pre-selected group of volunteers who will fill out a comprehensive survey of exactly what they drew over the course of, say, a week (by CSV or excel file? If there's too much interest maybe I'll set up some kind of Rails site).

This is potentially sensitive data, so we need to figure out a way to ensure privacy. If not enough people trust me with this information, maybe one of the mods will be willing to collect, verify, and anonymize the data.

I will then analyze the data and post the results (without personal identifiers) to these forums. Based on past data, I think that we may potentially have enough to make some interesting findings with as few as 200-300 draws.

The fields that I think would be interesting to have:

Type of LT
Token or CP-purchased
Cover Character
Cover Color
Time of draw

Is anyone interested in contributing data? Any other fields that I should collect?

Optimally, I'd like to do this at the start of the new season (so that the LT draw pool is consistent).

EDIT: If you are interested in contributing data, but don't want to publicly announce it, PM me or wait until I can arrange for a more trustworthy data collector?

CrookedKnight

Probably no surprise since I was part of the exchange that prompted this, but I'm happy to participate.

Unknown

Having some whales and superwhales on board would help data collection, but a mix of smaller spenders and F2P players would help determine any draw biases where spending is concerned, although I'm not exactly sure how to group people into these categories.

GrumpySmurf1002

Malenkov wrote:

The fields that I think would be interesting to have:

Type of LT
Token or CP-purchased
Cover Character
Cover Color
Time of draw

Is anyone interested in contributing data? Any other fields that I should collect?

Optimally, I'd like to do this at the start of the new season (so that the LT draw pool is consistent).

I don't keep track of tokens/purchased, but otherwise I have type, character, rarity, and color for every LT I've opened since inception. They're in the order I opened them, but not date/time stamped. I can tell you I opened zero tokens when odds were 5%, all have been done with 10% odds.

Edit: Point being I'm doing it anyway, so if you want the data I'll have it.

Unknown

GrumpySmurf1002 wrote:

Malenkov wrote:

The fields that I think would be interesting to have:

Type of LT
Token or CP-purchased
Cover Character
Cover Color
Time of draw

Is anyone interested in contributing data? Any other fields that I should collect?

Optimally, I'd like to do this at the start of the new season (so that the LT draw pool is consistent).

I don't keep track of tokens/purchased, but otherwise I have type, character, rarity, and color for every LT I've opened since inception. They're in the order I opened them, but not date/time stamped. I can tell you I opened zero tokens when odds were 5%, all have been done with 10% odds.

I think that the purpose of setting a strict timeframe is so that we can sidestep accusations of self-reporting inaccuracy since we can more or less validate data because rosters are publicly accessible (at least for those players who are in alliances) and so that the set of available characters is consistent.

If there's not nearly enough volunteers in a week, I'll fall back to just collecting whatever draw records people have kept.

mpqr7

I'll be happy to participate. I've been terrible at keeping track on my own (I mostly just open them and weep), but if you have a webpage or something, I'll enter my results in there.

GrimSkald

I already track the characters I pull out of each token - that's how I can tell you out of exactly 167 pulls I drew 9 5*s for an average of 5.39%. What this doesn't tell you (unfortunately,) is I have drawn exactly one since about the 60th or 70th pull, so my distribution has been weird. Interestingly enough, the actual characters I pulled are fairly evenly distributed - there are a odd few spikes (Carnage at 15, Mr. Fantastic at 14,) and low points (Thor and Invisible Woman at 4,) but everyone else whose been in it from the start is in the 5-9 range which seems about right.

I can track the additional data from now on - it would be very interesting particularly since I've been drawing so poorly lately. Green Goblin should get added any moment now, that would be a good place to start (not that it matters for me, though, since I won't be drawing Latest Legends save from tokens I'm given.)

Kojubat

I have been tracking my LT pulls for approximately a month, but not at token/color fidelity. I would be happy to increase my data collection and contribute.

Jam_Adams

I would love to see this. I've been pulling some really **** covers lately*, so I am wondering if pulls are truly random.

The Viceroy Returns

I've been tracking mine since the pull rate changed to 10% for a 5 star, when I first started opening them.
I have is a big long list of character & color, but I didn't track type or time pulled.
I do know how many total of each I've pulled, as I've been using the old mischiefmaker google doc token pull spreadsheet:

https://docs.google.com/spreadsheets/d/1HYgqFHHtQB7fWUkix2IGhbqDOVIapfX6AZdJOrpJcLQ/edit#gid=480895625

Not too many others have used this for LT results (I see you TazFTW!). In fact, I don't know how popular this is anymore to begin with. I still do if nothing else for my own OCD-ness and to cry over the terrible pulls I've had (check out my 12 Ant Man and 9 Mr. Fantastic pulls of Classic tokens!). I just barely over 10% 5 star rate combining both types.

I could easily keep track of these extra data points for however long you plan to grab data for this, or whatever you decide you want tracked.

Just let us know when you the start and end dates are.

Pongie

I would participate but I doubt I'll be earning many legendary token rewards now. It's all command points and I'm hoarding them till more of my 4* are championed and can use the covers.

I did recently opened 18 latest tokens hoping to get OML yellow before he rotated out.



x2




x2




and

OML

Fairly terrible as the only 4* covers I could use were cho and miles.

Kolence

I've started keeping track of every LT I open on January 17th. I recorded the date I opened it, the cover it gave (character, color, how it was used) and how did I get it. The list is up to 43 tokens as of now. Plus I remember what 5*s I got from the rest of the championed 3*s before I started keeping track of all the covers.

I'll keep tracking this for the foreseeable future as it's become sort of a ritual for me. If the data can actually help someone else, all the better.

TheXMan

I don't understand why this is sensitive data - you aren't going to be able to know exactly how they got the tokens. You can see people's rosters and kinda figure out who the whales are. I posted in the Suggestion forum that they should create an API where you enter a gamer name and see token history or even match history but didn't get any interest. Instead, we all have to track this manually. Lots of games show match history including things like K/D ratio making it obvious if someone is cheating. If it really is an issue, only let people see their own history.

Unknown

TheXMan wrote:

I don't understand why this is sensitive data - you aren't going to be able to know exactly how they got the tokens. You can see people's rosters and kinda figure out who the whales are. I posted in the Suggestion forum that they should create an API where you enter a gamer name and see token history or even match history but didn't get any interest. Instead, we all have to track this manually. Lots of games show match history including things like K/D ratio making it obvious if someone is cheating. If it really is an issue, only let people see their own history.

I meant that it may be sensitive because I'm going to ask people to record how exactly many LTs they're opening and whether or not those were cp-origin or token-origin. This is a strong indicator of how much people are spending on the game. If you seriously track event scores and rosters, you can kind of work it out. But I can appreciate people wanting whatever privacy that layer of obfuscation provides. In any case, I'm happy to do anything to gets people comfortable enough to share the raw data with me.

Cousin Simpson

Since aesthetocyst's 5* pull rate thread began I've reconstructed most of my pulls and am continuing to record them for the sadlulz (2 out of 32+ 5*s) and am happy to contribute to this community tracking project.

udonomefoo

I got a good chuckle out of tracking "per week". That's gonna be a really short list for me

Unknown

So far there's been maybe 10 people who have indicated that they'd like to contribute their LT data. This probably won't be enough data. Anyone else interested?

The fallback is just to collect LT logs, which I think would be a decently accurate alternative (although most people only record the cover and not the other data fields).

DrStrange-616

There are probably lots of people like me who thought, "good idea; I'll do it" but didn't post.

Cousin Simpson

DrStrange-616 wrote:

There are probably lots of people like me who thought, "good idea; I'll do it" but didn't post.

I'd also add that the number of people reporting their pulls in this thread suggests that plenty of them would participate.

Unknown

One of the reasons to do a time-limited, registration-gated test is because there was some criticisms from previous efforts that users could not be trusted either to give an accurate accounting of their pulls or an honest one. Doing it this way lets me validate at least a sampling of these results by comparing rosters before and after. I do think that this critique is way overblown and unfair, but I wanted to cut it off at the knees.

There's been a bit more explicit interest in contributing data, but now I'm thinking that I might ignore that last bit of criticism and just post the spreadsheet publicly and whoever wants to do it can just send me the results at the end of the period. And I might end up collecting individual players' historical records anyways as supplemental data.

Is anyone interested in any data fields that I've left out? The only thing that I think I'm intentionally leaving out is "people who spend money are treated better/worse by RNG than people who do not" because I specifically don't want to straight up ask about how much money people are spending. But I can probably do some analogous analysis based on number of draws.

Druss

Reading the forums I think I'm the only one not getting screwed by % draw rates for 5*'s. (17 out of roughly 200).

The trick to drawing a 5* is actually not to want them! I'm not keen on the whole PvP mechanic & don't wish to push my PvE scaling any higher, forcing me to only use the 5*'s, so I haven't levelled any of them - my OML is 4/0/2 so hardly usable without the yellow anyhow.

I wish the same could be said for my heroic token draw (just came off a 104 2* streak - which is comparable %-wise to being hit on the head by the moon).