Vault/LT Token "Streakiness"

Coconut99

So, I've always been under the impression that grey/gold tokens normally had some "streakiness" to them, and as a result, I would usually save all of my tokens up until the end of each PvP season, then pull standard tokens until I got a gold (unless it took forever, in which case I'd try again later), and then go pull all of my gold tokens (barring tacos). No clue whether or not that actually had any impact, but it generally netted me some decent covers (at that point, I'd be pulling like 80-90ish gold tokens).

Anyone know whether vault tokens and LTs/CP buys also have any streakiness to them? Is it still worth saving them up, or now that all PvE events are vaults, should I just forego saving up tokens entirely and just pull everything, LTs included, as I get them? Saving up vault tokens already has issues since they go random-fill once the event ends, anyway...but was wondering if there was any consensus.

Thanks!

Cousin Simpson

Streakiness is randomness.

Unknown

How did you come to have this impression?

Coconut99

Hearing from others in multiple (large/respectable...at least at the time) alliances. Always kind of felt like it was the case to me, also, but that's also probably just a feeling based on the difference between pulling all at once versus over the span of a month.

simonsez

Coconut99 wrote:

but was wondering if there was any consensus.

There will never be a consensus about the truth behind pull rates, but I think there's probably a consensus that stockpiling, in and of itself, doesn't help.

Malcrof

simonsez wrote:

Coconut99 wrote:

but was wondering if there was any consensus.

There will never be a consensus about the truth behind pull rates, but I think there's probably a consensus that stockpiling, in and of itself, doesn't help.

100% agree here, hoarded for a few weeks, opened yesterday , 7 20cp tokens, 1 duplicate char, and not back to back, but needed all 7 (2 5* drought over).

Stony

I do something similar to OP. I save LTs and CP until I pull a 4* from a heroic/event token. Then I pull 3-5 LTs. It seems like they've been better since doing this. But then again, probably just confirmation bias. It has made things more fun though!

meekersX

It's a common fallacy that random distributions should result in even results, even in small samples. Contrary to this expectation, streakiness is to be expected of proper randomness. This is closely related to the Gambler's fallacy.

Cousin Simpson

I've been thinking that a video showing 50 actual coin tosses would be useful in demonstrating random distributions.

carrion_pigeons

Vaults should change their odds with each pull, so unlike Heroic and LT tokens, there actually is a way to optimize.

If you have a bunch of tacos, open them until you pull the first thing you care about, and then stop. You get 10 tokens per vault, so if you want to balance out taco usage with taco flow, then you should identify the 30 best rewards in a DDQ vault and wait until you open one of them, and then stop and wait for the vault to reset in ten days.

Every time you open a "good" pull, the odds of opening another go down, and every time you open a bad one, the odds improve.

Coconut99

So...I think the "consensus" (such as it is) that I'm hearing is that any streakiness is confirmation bias, and to just go ahead and pull everything as I get it (or whenever, the point being there's no reason to save them up)?

PeterGibbons316

carrion pigeons wrote:

Vaults should change their odds with each pull, so unlike Heroic and LT tokens, there actually is a way to optimize.

If you have a bunch of tacos, open them until you pull the first thing you care about, and then stop. You get 10 tokens per vault, so if you want to balance out taco usage with taco flow, then you should identify the 30 best rewards in a DDQ vault and wait until you open one of them, and then stop and wait for the vault to reset in ten days.

Every time you open a "good" pull, the odds of opening another go down, and every time you open a bad one, the odds improve.

I don't necessarily agree with this strategy because it doesn't take into account the diminishing value of the "good" pulls as you progress over time while you horde.

Consider someone transitioning to 3* - they pull the token they want and then stop, over the next two months they cover max that hero, and they are left holding a token that previously had a lot of value as a potential needed 3*, but now represents just 500 ISO or something similar.

Calnexin

carrion pigeons wrote:

if you want to balance out taco usage with taco flow,

This deserves to be the title of a book.

Calnexin

Coconut99 wrote:

So...I think the "consensus" (such as it is) that I'm hearing is that any streakiness is confirmation bias, and to just go ahead and pull everything as I get it (or whenever, the point being there's no reason to save them up)?

The caveats there is whether you draw something you didn't want, but could still use

If you're trying to build character Y, you may want to cash in an LT for the chance you get a cover. But you might get character X, who's also worth building. You might also get character TA, who is worth 1000 Iso.

It makes sense to save if you're not in a position to roster everyone you might possibly want to. Otherwise, smoke 'em when you've got 'em. I cash in my tacos immediately, usually hoping for Iso, HP, or that lucky 4*/LT. It's usually Hawkeye, though.

Dayv

Pattern recognition is a hell of a drug.

Related: https://www.youtube.com/watch?v=yL_-1d9OSdk

Malcrof

DayvBang wrote:

Pattern recognition is a hell of a drug.

Related: https://www.youtube.com/watch?v=yL_-1d9OSdk

Thanks, now i am having Chic-Fil-A

simonsez

Coconut99 wrote:

So...I think the "consensus" (such as it is) that I'm hearing is that any streakiness is confirmation bias, and to just go ahead and pull everything as I get it (or whenever, the point being there's no reason to save them up)?

To clarify my post, no and yes. I'm convinced that the streakiness is due to something like every pull being drawn from an invisible vault, which is why so many people see clustering. But since you'd never know what's in your invisible vault at any point in time, stockpiling isn't going to help. It could actually hurt if you cash them all in when your vault has garbage.

carrion_pigeons

If there were really an invisible vault, people who hoard tokens wouldn't see as much variance as they do. Clustering is at the heart of randomness; there's no need to resort to explanations like an invisible vault.

simonsez

carrion pigeons wrote:

If there were really an invisible vault, people who hoard tokens wouldn't see as much variance as they do. Clustering is at the heart of randomness; there's no need to resort to explanations like an invisible vault.

There is, when everyone who posts pulls shows a statistically unlikely number of pairs and triplets. Show me the people posting pulls that show all this variance you speak of. Until I got bored yelling at the sky, every time someone posted a set of pulls, I would reply with the probability of getting whatever sort of clustering appeared in those pulls. I've yet to see anyone reply to one of these posts with "congratulations, you avoided statistically improbable clustering"

carrion_pigeons

simonsez wrote:

carrion pigeons wrote:

If there were really an invisible vault, people who hoard tokens wouldn't see as much variance as they do. Clustering is at the heart of randomness; there's no need to resort to explanations like an invisible vault.

There is, when everyone who posts pulls shows a statistically unlikely number of pairs and triplets. Show me the people posting pulls that show all this variance you speak of. Until I got bored yelling at the sky, every time someone posted a set of pulls, I would reply with the probability of getting whatever sort of clustering appeared in those pulls. I've yet to see anyone reply to one of these posts with "congratulations, you avoided statistically improbable clustering"

That's because clustering *isn't statistically improbable*. That's my point. Getting an even distribution of covers would be the surprising thing.

Let's say you ran two tests. The first test, you saved up 300 heroic tokens or whatever and opened them all on the same day. The second test, you got the same tokens but opened them as you got them.

The reason to conclude that there is an invisible vault would be if the former test gave you a *more* even distribution than the latter test, not a less even one. The reply you should be looking for from people is "congratulations, you got a spectacularly lucky lack of clustering", which of course never happens because it's both really improbable and also not very notable.

simonsez

carrion pigeons wrote:

That's because clustering *isn't statistically improbable*. That's my point.

Then your point is invalid. This isn't subjective. It's just math. If you pull 10 tokens, the probability of getting 3 MrF blues, and 3 Falcrap yellow is ridiculously small, if every cover actually did have an equal chance of being pulled. And when practically every person who posts pulls shows clustering that is unlikely, I stop giving the benefit of the doubt.