Faker vs Tropyhunter – Analysis

What if Faker was a Trophyhunter? Well, I downloaded the last 239 solo-queue games of faker, calculated the trophies he would have achieved and compared the results to the average Trophyhunter. The average Trophyhunter probability is calculated from 9261 random TrophyHunter games where the Trophyhunter played on a solo-lane. The following is a collection of the most interesting differences between Faker and our Trophyhunters.

Trophies that Faker achieves much more often than the average Trophyhunter


dripping-honey (1).png

Sweet Honey

Have 180 cs at 20 minutes (Faker: 8%, Av.: 0.29%)

Faker is a great farmer. He is almost 30 times more likely to achieve this trophy, which shows how important the basics are.


eclipse.png

Darkness

kill 12 wards (Faker: 2.93%, Av.: 0.04%)

Faker plays around vision. Clearing vision will help your jungler to gank and will help you set up traps in the dark. Faker achieves this trophy an astonishing 73 times more otfen than the normal Trophyhunter.


baby-face.png

Nurturing

One of your teammates has most kills and assists in game. You assisted him 3 kills pre 10 minutes. (Faker: 1.26%, Av.: 0.09%)

Faker is 14 times more likely to achieve this trophy. Sometimes it is better not to take the kill for yourself but give it to the right champions – not only as a support. Faker often plays something like Lulu or Karma mid, so he knows kills on his teammates might be more useful.


hound.png

The Hound

Set up others to carry. Achieve 5 assists pre 10 minutes. (Faker: 0.42%, Av.: 0.05%)

Faker gets this ~ 8 times more often. This is very similar to the Nurturing trophy. Help your teammates and they will pay you back.


boss-key.png

Unlock The Beast

Score two kills in the 40 seconds before your team takes baron nashor. (Faker: 4.6%, Av.: 1.13%)

Always think what you can do with an advantage. Try to get an objective with it. Faker shows that he can enable his team to get a baron more often than other players.


Trophies that Faker achieves less often than the average Trophyhunter


cultist.png

Assassins Creed

Achieve seven solo kills (no assists). (Faker: 1.26%, Av.: 9.2%)

Wow! I didn’t expect that. Faker – the greatest midlaner of all times – doesn’t achieve as many solo-kills as the average-joe sololaner. However, if you think about it – this is probably due to his stronger level of lane opponents where solo-kills are less likely in general. Also – Faker plays a lot more with his team and wants to find his advantages there.


shambling-zombie.png

The Zombie

Score at least two kills being dead (ten seconds after death). (Faker: 0.42%, Av.: 2.66%)

Faker doesn’t play enough Teemo? 🙂


ninja-heroic-stance.png

Ninjas

You and a teammate score three kills with no other assist and no more than ten seconds between two kills. (Faker: 0.42%, Av.: 2.54%)

This is basically winning a 2v3. Probably Faker avoids situations like that in the first place instead of relying on an outplay.


demolish.png

Demolition Pear

Take down at least 5 turrets. (Faker: 0.42%, Av.: 2.26%)

Not so sure about this one. Probably in higher elo taking down turrets is done more  likely by the adc in a team effort or by some splitpusher and not by the midlaner.


dripping-blade.png

Lunatic

Score at least 20 kills. (Faker: 0.84%, Av.: 3.88%)

This is also interesting. If you think Faker in solo-queue is the always fed uber-player who runs around and kills everyone – you are wrong. It seems to become clear that his strength is much more in facilitating his teammates instead of being fed himself.


So far for the trophies. Once the new playstyle system is out, I will add Fakers playstyle too so you can compare your own playstyle to his.

 

TrophyHunter: Playstyle Analysis I

It’s been some time since our first release of a playstyle analysis section in TrophyHunter. Our goal back then was to analyse a summoners playstyle and assign one of 16 characters that describe the player. We based that analysis on these four characteristics:

  • aggression/defense,
  • earlygame/lategame,
  • soloplayer/teamplayer and
  • risk-taker/risk-avoider.

We created threshholds for each of these parameters so everyone was evaluated as aggressive or defensive, as risk-taker or risk-avoider. Combining these four parameters we get 16 possible combinations. Each combination then resolves to one defined playstyle. It looks like this in the app:

palystyle.png

An interesting question now is how these playstyles are distributed among the players. Below is a picture showing the distribution of the 16 playstyles:

playstyle distribution 1.png

As you can see the playstyles are very unequally distributed. There are four main playstyles that amount to 3/4 of all players. Then there are 4 minor playstyles and the rest is almost non-existent.

The interesting thing about it is that each of the four characteristics in itself is balanced so that approximately half are risk-takers and the other half risk-avoiders. The problem is that the characteristics are highly correlated between each other. In our case – a risk-taker was almost always also a teamplayer and an aggressor. That is of course very bad design because the amount of achievable playstyles was heavily reduced like that.

 

What is the new approach?

We decided to change two things. First of all the playstyles should be more defining of the players main characteristic. If the player is ultra aggressive then this should be the defining attribute for the players playstyle. In the old system it didn’t matter if you were ultra aggressive or just slightly aggressive. The new system will have each playstyle character have a main attribute that mostly describes the summoners playstyle.

The second thing we wanted to change is that the player should have more clarity of how his playstyle is calculated. Especially if you think that the playstyle that we computed doesn’t fit your real playstyle, you can now at least understand how we came up with it.

So what do we do?

Luckily we have a huge trophy system already in place. Why not use it here? We do!

But first of all, let me give you the new attributes that we came up with:

  • Lethality
    • special kills, target selection.
  • Wisdom
    • vision game, leadership (team achievements), kill participation.
  • Combat
    • playmaking in fights. Usually trophies where it is not important that you are the one getting the kill.
  • Capability
    • your ability to make yourself fed in raw stats: farming, kills, gold, xp.
  • Power
    • raw damage numbers, crowd control, kill streaks.
  • Robustness
    • defensive abilities: few deaths, tanking damage, kda.
  • Cooperation
    • enchanting others: assists, heals.
  • Cunning
    • objective focus: turrets, baron & drakes.

 

Some attributes may seem to overlap a bit like lethality, combat and power, but we had to find  a way to deal with the high amount of trophies rewarding kills in some way. If we put all of the kill-trophies into the same attribute then that attribute would have the highest score for almost all trophy hunters. So we decided to split the kills into the categories described above. Lethality: trophies where you made some sort of play to get a kill, combat: trophies where you made a play leading to kills (you don’t need to be the killer), power/capability: raw kill stats.

Each trophy will give you points for one attribute depending on its difficulty. For example the trophy The Elephant (Don’t die for 20 minutes) will give you points for the robustness attribute. The trophy Key Targets (kill the target with the most gold at that point in time three times) will give you points for the lethality attribute.

We will also keep the playstyle characters from before. Each of these 16 characters will have one primary attribute and three minor attributes. For example The Owl will have wisdom as a primary attribute and the three secondary attributes intelligence, cunning and lethality.

We now calculate the sum of the points that you have in the secondary attributes plus three times the points that you have in the primary attribute. The result will tell how good this playstyle fits you. We do this for every playstyle character and the one with the most points will be assigned to you. I will add an example once we go online with this stuff.

 

Some more details ( for math addicts)

The points for each attribute will have asort of decay, so your playstyle will remain flexible. Each time you played a game, the old score for that attribute will be multiplied by 0.95 before the new points will be added. This means that a game that you played 13 games earlier will have only half the influence as the current one. A game that was played 25 games ago has approximately 27% as much impact as the current one and a game 50 games ago only has approximately 8% impact compared to the current one.

 

Trophy points will be adjusted as well

When balancing this new playstyle system, it became obvious that the current points given for a trophy were not balanced well. The difference in points didn’t properly reflect the difficulty in trophies. I will get into that in the next post.

Expect these new playstyle changes to go online around the end of this month.