How are cups available calculated?

Has anyone figured out and shared precisely how the difference between attackers' and defenders' cups translates into cups available for a win and a loss? We all have experience with this and a sense of how it works, but has anyone from the Nate Silver-ish forumers collected data and figured out the function, the shape of the curve? And is it the same at different cup levels (e.g., is an attack by a 1800 cup player on an 1800 cup town worth the same as a 3200 cup player on a 3200 cup town)?

Data on this should be easy to collect and of interest to many players. Looking forward to someone sharing an equation!

Quote:

---

Originally Posted by bhima

Has anyone figured out and shared precisely how the difference between attackers' and defenders' cups translates into cups available for a win and a loss? We all have experience with this and a sense of how it works, but has anyone from the Nate Silver-ish forumers collected data and figured out the function, the shape of the curve? And is it the same at different cup levels (e.g., is an attack by a 1800 cup player on an 1800 cup town worth the same as a 3200 cup player on a 3200 cup town)?

Data on this should be easy to collect and of interest to many players. Looking forward to someone sharing an equation!

---

1800 vs 1800 = 3200 vs 3200 => Yes. It's the difference of trophies between the two players that change the numbers of trophies available, not how much they have.

Data could be collected, but for what purpose ?

I have't seen any precise calculation. I would think some ambitious person with some spare time could go through their revenges, check the reward, then visit the player to see how many trophies they have.

And is there a max? Think I have had some 59 cups in a revenge. can it go any higher?

Quote:

---

Originally Posted by bhima

Has anyone figured out and shared precisely how the difference between attackers' and defenders' cups translates into cups available for a win and a loss? We all have experience with this and a sense of how it works, but has anyone from the Nate Silver-ish forumers collected data and figured out the function, the shape of the curve? And is it the same at different cup levels (e.g., is an attack by a 1800 cup player on an 1800 cup town worth the same as a 3200 cup player on a 3200 cup town)?

Data on this should be easy to collect and of interest to many players. Looking forward to someone sharing an equation!

---

I've put some effort into trying to figure it out.

Some things I have figured out are:

When looking for a target everyone knows that there is a number you can win and a number you can lose. The number you can win varies from 0 to 59 and the number you can lose seems to vary from 1 to 39. 0 to 39 might be possible but I haven't seen it even with a 700 trophy difference.

The amount also appears to be symmetrical. These numbers are somewhat made up, but as an example:

If you are at 1300 trophies and you fight someone else who is at 1300 trophies you should be able to win 30 or lose 20. Now if you are at 1300 trophies and fight someone else who is at 1200 trophies you might be able to win 20 trophies or lose 25. Note that for the 100 trophy difference the amount you can win dropped by 10 and the amount you can lose went up by 5. If you flip that so that instead of fighting someone at 1200 trophies you fight someone at 1300 trophies you will see an increase in the number you can win go up by 10 and the number you can lose go down by 5.

In short:

You: 1300 Them: 1300 Win = 30 Loss =-20
You: 1300 Them: 1200 Win = 20 Loss = -25
You: 1300 Them: 1400 Win = 40 Loss=-15

I've also noticed the amounts are not linear. For example a 200 trophy difference does not imply the trophy count for winning shifts by 20 and the for losing shifts by 10. It's a curve. It takes over a 700 trophy difference to reach the extreme end where a win is worth 59 trophies and a loss is worth -1.

There is also an additional twist that occurs under 1000 trophies. The number you win isn't the amount that is deducted from your opponent. This is how trophies are created. At an extreme case, if you have 0 trophies and beat someone else with 0 trophies; they will not lose any and you will still gain some.

I'd love to figure out the exact formula for the curve, but it's very tedious to try and collect data. Essentially you have to look at your trophy count. Then go through your revenge list. View a village to get their trophy count and then attack them to see the win loss rates.

Alternatively you can go to the attack screen and make a note of the win/loss amounts and their name and clan. Then exit the attack screen and search for them, view their village, and get their trophy count.

It could very well be that the reason SC does not list the trophy count of an opponent when attacking is so that collecting this data is much harder.

Here is some of the data I collected:

My trophies	Win	Loss	Their trophies	Trophy diff	ABS of win amt - 30
2338	0	39	1295	1043	30
1353	0	39	567	786	30
2338	1	38	1713	625	29
2338	6	35	1981	357	24
872	13	27	654	218	17
872	15	25	691	181	15
872	15	25	692	180	15
872	16	24	707	165	14
872	21	21	766	106	9
1283	21	25	1177	106	9
1353	21	25	1251	102	9
1353	24	23	1283	70	6
872	24	20	806	66	6
1353	24	23	1287	66	6
1332	25	22	1283	49	5
1353	28	21	1332	21	2
946	29	18	939	7	1
1353	29	20	1352	1	1
1283	30	20	1283	0	0
1332	30	20	1332	0	0
1283	30	19	1287	-4	0
872	31	16	884	-12	1
1353	35	16	1412	-59	5
1353	36	15	1426	-73	6
1283	40	12	1412	-129	10
1353	42	11	1507	-154	12
1353	42	11	1509	-156	12
1353	46	9	1562	-209	16
1353	47	8	1588	-235	17
1353	47	8	1589	-236	17
1353	49	7	1615	-262	19
1283	51	5	1592	-309	21
1283	52	5	1613	-330	22
1283	54	3	1678	-395	24
1353	55	2	1797	-444	25
1175	56	2	1678	-503	26
872	57	1	1412	-540	27
1096	57	1	1678	-582	27
946	58	1	1613	-667	28
997	58	1	1678	-681	28
974	58	1	1678	-704	28
946	59	1	1678	-732	29

Thanks, Dahimi. Very useful and exactly the sort of insights I was hoping for. I also would love to see the exact curve, which behaves different at both ends: the extra cups added below 1000 cups and the different matching that occurs above, I think, 3000 (at some point, it goes from +/- 200 cups to +/- 500, but I don't know how gradually the matching range increases).

Here's the rub: One person going through their revenge list is either making a few observations at different cup levels (if they are rising or falling) or a bunch of observations at one cup level. We know the trade-off is not linear, but I wonder if the curve is the same at all cup range. We assume it is, but it might not be.

I hope one of the data-minded forumers takes up this challenge. If someone does, a bunch of us could provide data. Perhaps someone could step up and solve this by creating a google spreadsheet to allow people to enter data. We'd get a huge sample pretty quickly, I think.

I can't take credit for this as I found this posted awhile back on the Forum. Unfortunately I accidentally deleted the note on my phone where I stored the initial formula and I was unable to relocated the original forum post. If I remember correctly it went something like this:

3 star = 30 trophies + (Defender's trophies - Attacker's trophies) / 12.

The formula would then work something like this:

Scenario 1:

Defender has 1500 trophies and you have 1300.
Difference is 200
200/12 = 16 and change
30+16 = 46

so if someone at 1300 trophies 3-stars someone with 1500 they would earn 46 trophies.

Scenario 2:
Defender has 1300 and attacker has 1500
Difference is -200
-200/12 = -16 and change
30 - 16 = 14

so if someone at 1500 trophies 3-stars someone with 1300 trophies they would earn 14 trophies.

Using this formula it would appear that players would be matched up with others roughly +/- 400 trophies from their own.

Quote:

---

Originally Posted by Stinkpickle63

I can't take credit for this as I found this posted awhile back on the Forum. Unfortunately I accidentally deleted the note on my phone where I stored the initial formula and I was unable to relocated the original forum post. If I remember correctly it went something like this:

3 star = 30 trophies + (Defender's trophies - Attacker's trophies) / 12.

The formula would then work something like this:

Scenario 1:

Defender has 1500 trophies and you have 1300.
Difference is 200
200/12 = 16 and change
30+16 = 46

so if someone at 1300 trophies 3-stars someone with 1500 they would earn 46 trophies.

Scenario 2:
Defender has 1300 and attacker has 1500
Difference is -200
-200/12 = -16 and change
30 - 16 = 14

so if someone at 1500 trophies 3-stars someone with 1300 trophies they would earn 14 trophies.

Using this formula it would appear that players would be matched up with others roughly +/- 400 trophies from their own.

---

That doesn't jive with the collected data. For example:

Scenario 1: Defender has 1588 trophies, attacker has 1353. Difference is 235. 235/12 = 19 and change so you'd expect the number to be 49 or 50. The table above indicates 47.

Scenario 2: Defender has 1678 trophies, attacker has 1096. Difference is 582. 582/12 = 48 and change so you'd expect the number to be 78 or 79. It is 57.

Also the function you have provided is linear when the data suggests the function is a curve.

I think its possible to lose +1 if your at -1 trophies

Quote:

---

Originally Posted by Frogsdead

I think its possible to lose +1 if your at -1 trophies

---

Yeah I think being at -1 trophies is an anomaly. Perhaps a rounding error or some such. I'd probably treat it as a special case.