# [Primer] Basic explanation of Trophies or Where do Trophies come from and why do I lose them?

New Subject
[Primer] Basic explanation of Trophies or Where do Trophies come from and why do I lose them?

OLD Subject
[Elo Primer] Where do Trophies come from?

[Typist: “See, when a mommy trophy and a daddy trophy…. “ grabs computer away from The Typist Ahem.]

Trophies in Empires use a method called the Elo rating system. The system is named after physics professor Arpad Elo. It originally designed for use in chess tournaments staring in 1960, so it has been fine-tuned for over 57 years.

==Predicting the Future!==
The Elo rating system is designed to predict the future! Similar to seasonal weather forecasting it assumes any trend will continue. So it will get warmer in the summer (gaining rating) or get colder in the winter (losing rating). Each time Elo rating math fails to predict the future, it will self-correct.

In weather forecasting, there will be a range of temperatures for any given day. However, you are unlikely to get snow on the Summer Solstice [Typist’s Note: “In an area currently experiencing the Summer Solstice, at a latitude that should not have snow at such time. Let’s be exact about this, eh?”] To clarify for the Typist, Summer Solstice in Miami. This is why your rating can fluctuate wildly even without your opponent messing with their Elo rating (see below) but will usually stay within a minimum and maximum range.

If Elo rating’s very complex math says +30 Trophies for a win and -30 Trophies for a loss, then based solely on your current rating and your opponent’s current rating, it thinks you have a 50% chance of winning. This is similar to knowing it will rain on half of the city, but not knowing where the rain clouds are and whether a particular person will need an umbrella.

If the win bonus is huge and the loss penalty is small, it predicts you will lose [Typist’s note: “Try anyway!”]. This is similar to it raining on 90% of the city. It thinks you will get wet! If the win bonus is small and the loss penalty is huge, it predicts you will win. This is similar to it raining on 10% of the city. It thinks you will stay dry.

==It’s Math!!==
This is literally that same as 2+2= 4 and 6-2=4. Elo rating math compares your two current ratings and makes a prediction. Every time two opponents with those two ratings match, it will always give the same exact win bonus and loss penalty. It does not care if either team deserves it’s current rating (see deliberately messing with your rating below). It does not care about your board during the match (summer will generally be hot and winter generally cold, but any day can be either). Elo rating cannot actually see the future, it is only math.

==Elo rating swings==
As a general rule, the weaker your team, the smaller your changes in rating. Too many higher teams are simply unbeatable and too many special skill are only available to higher teams. The stronger your team, the larger your changes in rating. You will fight powerful teams helpless against your special abilities and vice versa.

This is even more true if maxed out players drop in and out of player versus player (depending on special events in the game). Also a constant influx of new players caused by the game winning an award or an especially effective advertising campaign will also reinforce this general rule.

However, this becomes less true the more unbalanced the available heroes and the smaller the influx of new players. If Tiamat automatically wins raids for her team [Typist’s Note: “Hey, Devs. Can I have a Tiamat? Pretty please?”], only players who acquire, and fully level, Tiamat will have the highest Elo ratings. If their are no new players and thus no new teams, even the lowest rated team will level up their heroes.

==Deliberately messing with your Elo rating==
Just like a billiards hustler pretending to be a beginner until the mark bets a lot of money, players can deliberately mess with their Elo rating which messes with the system’s ability to predict the future.

Inigo Montoya: I admit it, you are better than I am.
Man in Black: Then why are you smiling?
Inigo Montoya: Because I know something you don’t know.
Man in Black: And what is that?
Inigo Montoya: I am not left-handed.
[switches sword to his other hand, and begins to fight far more successfully]
Man in Black: You are amazing.

Elo rating assumed a bunch of stuff (see assumption section) similar to the weather forecast assuming you are staying in your city and are outside a building. If you fly to the arctic from the tropics this is similar to using a 1* 2.20 defense team when your previous raid team was 5* 4.80. If you go inside your house and turn your cooling unit on maximum at midnight on the winter solstice this is similar to fleeing the raid when you have five heroes at full health and your opponent’s team has one hero at 2% health. Elo rating is totally useless, since a bunch of stuff happened it could not predict.

Elo ratings will always be worse than your morning weather forecast because players are always mucking with its basic assumptions.

==Elo rating deflation==
Whenever the players get access to more powerful teams, the game will experience Elo rating deflation.

For example when the Devs added the Barracks, players that did nothing lost Elo rating and those who maxed out their troops the fastest got (temporarily) a large increase in Elo rating.

This creates a temporary race to see who can max out the new advantage first. But eventually it will settles down as all players gain the new advantage. New players will have something else they need to increase their rating. So new player Elo ratings will tend to be lower, and increasing their rating will take longer than before the change.

==Switching from weather similes to sports simile==
Elo ratings in an MMO are usually designed so a person first starting out has a rating of zero. This is similar to a person playing football for a club. You know the basic rules, have the basic equipment, and have practiced with others. Elo rating of 1000 usually means you can qualify to play for a top-division football league. Elo rating of 2000 usually means you can qualify for the World Cup finals. Elo rating of 3000 usually means you were one of the players on the winning team for the World Cup.

==Why use Elo rating at all?==
Even when players deliberately mess with Elo ratings, your account’s highest Elo rating will be a pretty good indicator of how well your can raid using your best team. So all players trying to get the highest ratings in the game will tend to get matched with similar rated players if those players are also trying to get the highest rating.

No system is perfect, even with players not messing it up. [Typist’s Note: “It worked fine in Beta! How did they break it so fast?”] But Elo rating has many years of success under many player versus player and teams versus teams sports and games. Like weather forecasting it cannot predict the future, and it has its problems, but it gets the job done.

==Assumption Section==

Elo math assumes:

There is significant punishment for not trying to win every match. Several systems have evolved to deal with this since Elo ratings were introduced in 1960, because humans.

Each player plays the same number of matches. This is just not practical for a MMO, so you will get some weird numbers from frequent raiding players and infrequent raiding players.

Nothing has changed since the last time a match was played. This actually loses an MMO money, so you will get some very very weird spikes in Elo rating (up or down) from week to week.

7 Likes

Ok, understand Elo (tend to play red/black in roulette with trend…not Elo,but wth).

Where do trophies come from? (Daddy)

There is one other assumption in an ELO system:

That the two players in a match are playing the same game with the same chance to win, and that the match is (at least primarily) a game of skill. (The skill being what ELO systems measure.)

Raiding doesn’t meet that criteria.

The defender:

• Not actively playing.
• Putting together a defense team is more a measure of their roster than their skill using it, although you can make a better or worse defense with the same roster

The Attacker:

One natural consequence of this disparity: When you are actively raiding, your trophies go up. When you log out and are defending, your trophies go down.

I would say that there is a “resting” defense team trophy level–a given defense team will tend to fall to a certain trophy level, and stay there.

I think that the active raiding level is mostly a measure of how many raids you’ve done since you were at the resting defense level. (If you don’t believe me, find one of the topics that states pretty much anybody can get to #1 as long as they are willing to burn 5-6 raid flasks to get there.)

Anyhow, to my mind, all these flaws put together actually make raiding fun, even if my actual trophy level is pretty random and noisy.

2 Likes

# Elo ratings- it rates

No.

His math did not really care if you were playing a game of skill or not. If you were the reigning world champ or a beginner ( see lip balm cheating for an interesting anecdote ).

Elo assumed that the player’s rating and the win/loss/tie between two matched players would both be a bell curve ( Why edge cases like 2400+ trophies do not work well).

Instead of assuming equal skill, or equal starting conditions, Elo assumed that no one could know a players’ current Rating so the formula has several variables that hard adjust when a player first starts ( see Notes- new player). The fact that Elo needed to add such variables to speed up early rating changes, and more importantly to dampen late career rating swings, shows how difficult it is to measure anything about us humans.

…this variable is the first casualty with MMO which is where MMO go to heck very very quickly.

Between players skipping raiding while they power up teams, to mismatched good attacking heroes/ good A.I. Defense heroes to Elo deflation like troops, Barracks, Classes, Elo’s several assumptions had no chance of being applicable.

# Raid/ War season

Some MMO use seasons to compensate for new players and rating deflation. MMO that do use seasons often reset the season on a timetable, after X number of new users, after introducing a new type of power up, etc.

If Small Giant Games has seasons, the new season would start with the activation of classes.

# The Perfect is the enemy of the good ( wiki link)

But people keep using Elo’s basic formula, with a few tweaks like participation points or rating decay or seasons, because 80% of the time, it works. While the next best thing does not always work 80% of the time and takes much more work ( see poor Small Giant Games attempt to improve on Elo’s math with their war score, which is morphing closer to Elo’s math with every iteration).

# Nightly Defense

Actually my trophies always go up at night.

Gryphonkit, my wife, reached #30 on the Global Top 100 player list from players revenging against her reaching #58 on the Global Top 100 player list. She screen shot at #58, I logged in to check and screen shot at #30. It did not last.

Often your trophies go down after a big attacking streak because players can reroll if matched with you ( plus valid points people, including you, have mentioned on the forum about attacking versus defending).

# Choosing what to level

A contributor is that a very good attacking hero is seldom a very good defending hero.

Unlike other games, a freshly summoned 4* 1.1 hero is not very useful, and with titans being the best way past 3*/4* ascension item paywall, few players level up a very good defender until they have one to two very good rainbow attack teams.

Example: I still use 8/8 4* 4.70 Wu Kong on defense, because my other defense choices are worse. ( Edit: Just replaced with my Sigrunn 1*+20 for the week ).

This. Well definitely noisy and fun. Less random than most realize.

Niantic’s Ingress learned the hard way that defender advantage is boring. Then the relearned the lesson with Pokémon GO gyms.

All things being equal otherwise, if attackers do not have a huge advantage over a defender, the game gets boring and stale very quickly.

This is both a consequence, and a encouragement, for using A.I. To defend in game. Since A.I. Is more likely to lose and defender is more likely to lose. Since defender has a disadvantage, letting the A.I. Handle the inevitable losses makes the game more fun ( why yes, defending in live PvP like Book of Heroes, or Hearthstone, can be even less fun for many users so live PvP often relies on participation points and rating decay formulas ).

# Lip balm cheating in chess

They don’t really know how I cheat, but they are “sure” that I cheat. Krivonosov was telling stories about nowadays’ high-technology, and so I heard my lip balm became the most suspicious…

Real life example of high rating players assuming Elo’s math is broken so there must be cheating.

# New player

As ih8sens said, systems adjust for new players by giving them a provisional rating which limits their effects on the established pool while giving the provisional rating the ability to fluctuate wildly in an attempt to quickly estimate the correct rating. This can be seen on this site (or just about anywhere else) when a new player wins a game and gains hundreds of rating points while an established player may gain less than 10.

OK, let try that again.

The math doesn’t assume much of anything, it can and is applied, and the math still “works”

However for the ELO score to be a meaningful measure of how good players are, there are further requirements.

The asymmetry between attackers and defenders messes with it badly.

I’d conclude you don’t raid to improve your trophy score, or you are somehow in a very different circumstance than me or the people I’ve been playing with since I started to pay attention.

These days I am usually at 2700-2850 cups after I finish filling a raid chest, and after I log off / in the morning I’m generally around 2500-2600 cups. (Haven’t fallen out of diamond in months) This was true back when I was in silver, and tried to raid up to open my raid chests in gold if I could.

Actually if I raid up into the top 100 (It is pretty often), as soon as I log out of the game, I’ll get hit 10~20 times in a few minutes or half an hour. That generally knocks me from 2800-2900 back to the ~2600 level.

I generally re-roll for two things in my opponents: Winning more cups than I lose if I win the match, by 2:1 if possible, and something that I like my odds of beating. (I don’t reroll for the second reason much anymore). In the past I would reroll for more food/iron, but I don’t now. I try to hit all my revenges.

I would say that few players choose to level up a good defender until they have quite a few good attackers.

Free/cheap players often don’t have the luxury of choosing which 4* or 5* hero to level up and thus level up the ones they have, whether they are good attackers, good defenders, or even not particularly good at either.

And some heroes are good attackers and bad defenders (Wu Kong), some heroes are good defenders and poor attackers (Guinevere), but some heroes are good at both. (Gravemaker or Boldtusk).

The degree to which players have leveled up good defense heroes is … a lot more random in that regard. (They could have leveled up Aegir thinking he was a good defensive hero, when he’s not really)

3 Likes

Based on inquiries on these boards, people seem to care quite a lot about their defense. Personally I think raid defense is over-valued. War defenses, however, are important for holding up one’s responsibilities in an alliance.

4 Likes

# To Raid of Not to Raid

You are not wrong.

But in my personal social cycle, I read the forum and forty+ of my friends do not. Roughly a quarter to half of my friends skip raids.

# To Tank to Zero or Not to Tank to Zero

Before I joined, 50% of my alliance tanked trophies to ZERO for easy chest filling. After raid loot adjustment it dropped to 25%. After Season 2, dropped to 0% ( go Devs ).

# Raid defense

Gryphonkit, my wife, agrees. She thinks the best raid defense is a very good raid offense.

# War Defense

In my alliance I still defend with Wu Kong 4* 4.70 since my other options are worse.

Until I have three good rainbow 5* 4.80 teams, I cannot justify spending 3* / 4* ascension items on a defense hero.

When players are leveling up 4*.

I’ve always advised titan heros first then level up others once you have a solid titan team in each color.

Even a decent 4* war defense can be made with titan heros. Just depends on what progress level your alliance is at and the strength in opponents you face.

In other words i 100% agree with your logic

2 Likes

I agree on that too in general…but…Heroes are all different, and how effective they are against various map/quest levels/monsters, titans of various colors, raids against various other heroes, on raid defense, and on war defense isn’t the same.

I’d suggest starting with ones that are all-around useful, preferably good against all colors of titans. But you have to pick from the heroes you actually have, so you may not manage that.

The good news is that most of the heroes which are effective against titans are also useful in other areas. But there are a few one-trick ponies out there, as well as a few zero-trick ponies

1 Like

Subject
Latest edit- Giraffe/ human hybrid simile

Why cups are higher in revenge than regular win?

Cups are not always higher in a revenge. It depends on where your cups are at the time of the revenge.

In my case, 100% incidents.