# Are the summons based on your purchase behaviour? Is SG fair?

I have an issue with losing my trust in SG.
I have done in 2020 about 770 pulls. Out of it I got only 2 HOTM. With 1.3% it should be 10.
Probability to get only 2 hotm out of so many pulls is so low as: 0.27%. Only to 2 700 out of milion could this happen if probabilities are fair.

I am spender. During 2019 I was at probability as declared: 1.28%.

My doubts are: Am I not in marketing category of SG: “Do not give him HOTM, we know that he will buy new gems.”?
I am losing my trust in randomness of summons. I am not sure if SG is fair.

Guys, does anybody have similar experience?
I am really considering if it is not time to leave, without trust in developer it is hard to play.

@zephyr1, could you please ask SG stuff, if they have different categories of summons based on your purchase behaviour?

4 Likes

Odds of having your level of bad luck are low but your calculations show that it happens.

3 Likes

I recall a player that spent a lot of money and got nothing they wanted.
It’s a gamble no matter how much you spend!

3 Likes

Regardless of what you spend you’re still gambling, and nothing is guaranteed when you gamble.

1 Like

Just going to drop this in here:

Odds are odds… The likelihood doesn’t change… That kind of thinking is just plain and simple conspiracy theorist stuff.

All spending money does is increases the OPPORTUNITIES for the odds to play in your favour…

Think of it like a dice. If the goal is to roll a 6, let’s compare two situations.
Situation 1, the player gets one roll. They have a 1/6 chance of rolling the needed 6
Situation 2, the player gets 10 rolls. They have 10x opportunities at the 1/6 needed to get the 6.

In mathematics how this works out is:
`{x} = 1 - (1 - {chance of success}) ^ {# of rolls}`

Situation 1 = 1 - (1 - 0.1667)^1 = 16.667%
Situation 2 = 1 - (1 - 0.1667)^10 = 83.849%

Clearly person 2 has a higher liklihood of getting the 6.

So the same extends to the odds in the game… The odds are fixed for person A & person B. All that changes between the two is the number of opportunities that each person has to get the 5* hero.

I go into it a bit more in this thread:

11 Likes

No it should be a value in the range [0 - 770] with a tendency towards 770*.013=10 in the long run.

3 Likes

Guys, I am statistician myself. I know how probabilities work.

What I am saying is that 0.27% is damn small probability. I do not want to get simplistic answers that such low probabilities still happen. I know. Of course they do.

What I am asking is: are there any evidence that SG is / is not changing summon probabilities depending on how much you spend?
Have SG claimed not doing this? Could anybody ask them?
Or are there more such a stories? I mean like more than 0.27%, that would show it is not only conspiracy?

As a statistician you would recognise this style of curve:

Regarding your three points at the end:

1. Burden of proof is on the accusor not the defender…
Same as the proof of existence is on proving it IS there not that it isn’t (e.g. unicorns… It’s not proving that unicorns do NOT exist, instead you must prove that they do exist…)

2. I’m not sure if I can find a specific thing about them saying that they don’t adjust the odds but… Why should they? If they WERE adjusting odds it is called fraud & false advertising & theft… and is illegal… So no, I’m like 99.999999999% sure that they wouldn’t be doing it…
Additionally, coding the ODDS to be specifically tailored to a person’s spending history would be MASSIVELY difficult to do… And would require constant data referencing & collation which would cause stupid amounts of computing when you consider the sheer scale of SGG player base…

3. There are always stories of getting a crappy run of luck… There is also the other end of the curve where people are doing single summons only & get every HotM etc… Those odds are just as stupid…
There are also many F2P players who’ve never spent a cent yet have an amazing roster… By your accusation, they would never be able to get anything as the odds would get progressively worse and worse and worse until they start spending… Which they haven’t done…

5 Likes

Gurnvor, I recognize it. It is normal distribution. But how many cases hapens out of n pulls is not normal distribution.

Isn’t it…?

There are statistical outliers on both sides… As I said, there are stories of single pulls giving both the event hero (1% chance) and the HotM (1.3% chance) on a single free summon (0.013%)…

There are also stories like yours where it’s the inverse… Lots of pulls for not much success…

Doesn’t change the facts that there are outliers to any RNG based odds…

Same as in RL… I knew someone who won the lottery not just once but twice… Then there are people who’ve done the lotto ever week for 40 years and never won a dime…

If SG keeps different probabilities for different players, does not mean that overall probability is not 1.3% as they promisse. So nothing illegal.
Very easy to programm. Imo.
Why would they do it? It would maximize their revenues. Players who spend a little get hotm and are motivated to spend more. (And there are quite a few such stories). Players who spend more, might be continuing to spend, because they did not got the hotm n

2 Likes

No it is not. Normal distribution is symetric. Here you cannot go bellow zero. You cannot pull -1 hotm. Not possible.

1 Like

Well don’t think of the x-axis as being # of HotM pulled, think of it as the “personal % chance”

Median point is 1.3%, distribution around that.

In any case, my points still stand.

1. There is no evidence
2. It is illegal to advertise “odds” in game but then manipulate those odds on a per-person basis
3. There is substantial evidence of the bell-curve where there are the “lucky” and the “unlucky”
1 Like

Ok, still not symetric. You cannot get bellow 0% probability. Man, I am sure, what we are talking here about, is not normal dittibution. Cannot be.

setting the math aside… the operational logistics of identifying spending clusters of players to deprive of hotm in such a manner that improves consumer spend from that target cluster isn’t all that viable.

I suspect that they are focused primarily on measuring your spend over time to get an 18-24 mo. average revenue per user for each cohort (month) of installs and then adjusting their marketing spend and targeting along those lines.

that said 770 pulls and no results sucks… sounds like you need a beer and not a statistics refresher.

5 Likes

there is only really one distribution in mobile games and it’s the power distribution (the blue line) which reflects the 4-7% of players who actually spend in a game and the 1% of players who represent about 80% of the total revenue generation of a game and also the players at the top of our leaderboards with all level 30 troops and special heroes.

I suspect if OP doesn’t have all of the heroes wanted, it’s not a function of the normal distribution, it’s a function of simply not spending enough to top the power distribution.

Your point is obviously different and focusing on the possible outcomes based on the available odds but while it is correct, it is not really diagnosing the situation fully. Clearly 770 pulls over the course of 3 months isn’t enough for 10 hotm- what we really need to know is the \$\$ spent behind the 770 pulls and how that \$\$ compares to players with the 10 hotm. (especially if we are talking the monthly ones and not the Atlantis summons)

4 Likes

It is not. It is binomial distribution as there are 2 cases with each pull:
1 you get the hotm 1.3%
2 you don’t get the hotm 98.7%

n = number of pulls, p= 1.3%

1 Like

I’ve asked before, and they’ve said no.

This was me asking this question during the March 2019 AMA:

6 Likes

I hate how indirect their answers were. I don’t want to be skeptical, but they really skirted around the explicit nature of the question.

We do know that not all deals are offered to all players. That suggests that they do market things differently based on unique player data. Where it goes from there we have no idea, but it at least makes you wonder what else that data is/could be used for.

To be clear, I’m making no particular claims — just laying out the facts.

7 Likes