Jump to content
You need to play a total of 5 battles to post in this section.
Lert

The return of Lert's tier 5 experiment

78 comments in this topic

Recommended Posts

23,989
[HINON]
Supertester
19,896 posts
13,384 battles

Many of you will remember my tier 5 experiment, and are probably wondering what happened to it.

 

Well, in short, I put it on hold for a while. First the Graf Spee missions handing out a ton of free tier 6's for everyone thus 'messing up' MM and skewing results, then RL interfered a bit, I was burnt out on the game for a while and took a break, but today I resumed the experiment.

 

As a refresher:

 

This experiment sees me playing 100 solo battles in a tier 5 ship and keeping track of the battle's tiering, server population and win or loss. This is far from a scientific experiment and is not intended to prove or disprove anything. At best, it might show a tendency, a trend.

 

My goal is to see how 'bad' the T5 matchmaking is. It's conventional wisdom on this forum that T5 MM was 'ruined' with WG's T4 MM change from unprotected to protected. But is it really? I often see numbers about how 'T5 ships have to go up against T7 ships 70% of the time' and I want to see if that is really the case.

 

Before the experiment was put on hold I had collected the following 30 datapoints:

 

1) 6.2k pop, tier 7 battle, victory
2) 6.5k pop, tier 5 battle, victory
3) 6.8k pop, tier 7 battle, defeat
4) 7.3k pop, tier 5 battle, defeat
5) 7.6k pop, tier 7 battle, victory
6) 8k pop, tier 7 battle, victory
7) 9.2k pop, tier 5 battle, defeat
8) 9.6k pop, tier 7 battle, defeat
9) 10.6k pop, tier 6 battle, victory
10) 11.3k pop, tier 7 battle, victory

 

Tier 5 battle: 3 / 10
Tier 6 battle: 1 / 10
Tier 7 battle: 6 / 10

 

11) 13.4k pop, tier 5 battle, victory
12) 13.6k pop, tier 5 battle, defeat
13) 13.7k pop, tier 7 battle, victory
14) 13.8k pop, tier 7 battle, defeat
15) 14k pop, tier 7 battle, defeat
16) 14.4k pop, tier 5 battle, victory
17) 14.7k pop, tier 7 battle, defeat
18) 14.8k pop, tier 7 battle, defeat
19) 15.2k pop, tier 6 battle, defeat
20) 15.4k pop, tier 7 battle, victory

 

Tier 5 battle: 6/20
Tier 6 battle: 2/20
Tier 7 battle: 12/20

 

21) 9.6k pop, tier 5 battle, defeat
22) 10.9k pop, tier 7 battle, victory
23) 11.6k pop, tier 7 battle, victory
24) 11.9k pop, tier 5 battle, defeat
25) 12.3k pop, tier 7 battle, victory
26) 14.2k pop, tier 5 battle, defeat
27) 14.6k pop, tier 5 battle, victory
28) 14.9k pop, tier 7 battle, defeat
29) 15.2k pop, tier 5 battle, defeat
30) 15.4k pop, tier 7 battle, victory

 

Tier 5 battle: 11/30
Tier 6 battle: 2/30
Tier 7 battle: 17/30

 

Today the following 10 datapoints have joined the results:

 

31) 6.0k pop, tier 5 battle, defeat
32) 6.2k pop, tier 7 battle, defeat
33) 6.3k pop, tier 7 battle, victory
34) 6.4k pop, tier 7 battle, defeat
35) 6.4k pop, tier 7 battle, victory
36) 6.5k pop, tier 5 battle, defeat
37) 6.7k pop, tier 5 battle, victory
38) 7.1k pop, tier 7 battle, defeat
39) 7.8k pop, tier 5 battle, defeat
40) 8.3k pop, tier 6 battle, victory

 

The results so far:

 

Tier 5 battle: 15/40
Tier 6 battle: 3/40
Tier 7 battle: 22/40

 

I will try to keep this experiment going. I need only 60 more datapoints to complete the experiment, if everything goes well that should take about a week at 10 a day.

 

<Edit>

 

Fifth batch done:

 

 

41) 6.3k pop, tier 6 battle, victory
42) 6.6k pop, tier 5 battle, defeat
43) 6.7k pop, tier 7 battle, victory
44) 6.8k pop, tier 7 battle, victory
45) 7.0k pop, tier 6 battle, defeat
46) 7.3k pop, tier 6 battle, victory
47) 7.4k pop, tier 5 battle, defeat
48) 7.7k pop, tier 6 battle, defeat
49) 10.4k pop, tier 5 battle, defeat
50) 10.6k pop, tier 5 battle, victory

 

<edit>

 

Sixth batch done:

 

 

51) 7.7k pop, tier 5 battle, victory
52) 8k pop, tier 7 battle, victory
53) 8.2k pop, tier 5 battle, victory
54) 8.3k pop, tier 7 battle, defeat
55) 8.5k pop, tier 7 battle, victory
56) 8.7k pop, tier 5 battle, victory
57) 11k pop, tier 6 battle, victory
58) 11.1k pop, tier 6 battle, defeat
59) 11.4k pop, tier 7 battle, victory
60) 11.6k pop, tier 7 battle, victory

 

<Edit>

 

Seventh batch results:

 

 

61) 8.4k pop, tier 5 battle, defeat
62) 8.5k pop, tier 7 battle, defeat
63) 8.7k pop, tier 7 battle, victory
64) 8.8k pop, tier 6 battle, victory
65) 9.5k pop, tier 6 battle, defeat
66) 9.6k pop, tier 7 battle, victory
67) 9.7k pop, tier 5 battle, defeat
68) 9.8k pop, tier 7 battle, defeat
69) 10.1k pop, tier 7 battle, victory
70) 10.3k pop, tier 7 battle, victory

 

<Edit>

 

Eighth batch results:

 

 

71) 5.1k pop, tier 6 battle, defeat
72) 5.5k pop, tier 5 battle, victory
73) 5.9k pop, tier 7 battle, defeat
74) 6.5k pop, tier 7 battle, victory
75) 7k pop, tier 7 battle, victory
76) 7.3k pop, tier 7 battle, defeat
77) 7.6k pop, tier 5 battle, defeat
78) 7.9k pop, tier 7 battle, victory
79) 8.4k pop, tier 7 battle, defeat
80) 8.8k pop, tier 7 battle, defeat

 

<Edit>

 

Ninth batch results:

 

 81) 9.2k pop, tier 7 battle, defeat
82) 9.4k pop, tier 7 battle, victory
83) 9.8k pop, tier 5 battle, victory
84) 10.1k pop, tier 5 battle, victory
85) 10.2k pop, tier 7 battle, defeat
86) 10.4k pop, tier 7 battle, victory
87) 10.5k pop, tier 5 battle, victory
88) 10.7k pop, tier 5 battle, defeat
89) 10.7k pop, tier 5 battle, victory
90) 10.8k pop, tier 7 battle, victory

 

<Edit>

 

Final batch results:

 

 

91) 6.5k pop, tier 5 battle, victory
92) 6.8k pop, tier 7 battle, victory
93) 7.1k pop, tier 7 battle, victory
94) 7.6k pop, tier 7 battle, victory
95) 8k pop, tier 5 battle, defeat
96) 8.2k pop, tier 5 battle, defeat
97) 8.4k pop, tier 5 battle, victory
98) 8.6k pop, tier 5 battle, defeat
99) 8.9k pop, tier 7 battle, defeat
100) 9.2k pop, tier 7 battle, defeat

 

Tier 5 battle: 36/100
Tier 6 battle: 12/100
Tier 7 battle: 52/100

  • Cool 16

Share this post


Link to post
Share on other sites
[KNTAI]
Members
3,130 posts
7,923 battles

I like the Texas. My favorite T5 ship. It has nice turning, great AA, and decent guns. Get T7 MM and many BB's? Use your 10 19km guns to bring in the hurt from afar. Any cruiser of any tier won't matter, as well. Great ship all-around.

 

...That said, besides the Texas, and maybe the Kongou, I wouldn't really play T5's anymore (A case COULD be made for Gremmy and Kami though).

Edited by Excield

Share this post


Link to post
Share on other sites
2,247
[DAKI]
WoWS Wiki Editor
7,613 posts
6,838 battles

This sounds interesting, do you plan on doing the same at T6?

Share this post


Link to post
Share on other sites
148
[DOOP]
Members
856 posts
2,435 battles

Get in a Kami and damn the results. Full speed ahead..:izmena:

 

Share this post


Link to post
Share on other sites
23,989
[HINON]
Supertester
19,896 posts
13,384 battles

do you plan on doing the same at T6?

 

Probably not. Feel free to run your own tier 6 experiment if you want to see the results of such.

Share this post


Link to post
Share on other sites
23,989
[HINON]
Supertester
19,896 posts
13,384 battles

Are you going to test divisions too?

 

How much divisioning messes with MM is:

 

1) Outside the scope of this experiment and

2) One needs a base line value to compare against before you do something like that anyways.

 

So, no. All 100 battles will be solo.

Share this post


Link to post
Share on other sites
2,247
[DAKI]
WoWS Wiki Editor
7,613 posts
6,838 battles

Probably not. Feel free to run your own tier 6 experiment if you want to see the results of such.

 

Could be worth a shot :hmm:

Share this post


Link to post
Share on other sites
3,713
[HINON]
Members
9,687 posts

 

How much divisioning messes with MM is:

 

1) Outside the scope of this experiment and

2) One needs a base line value to compare against before you do something like that anyways.

 

So, no. All 100 battles will be solo.

 

Solo is the only way to go, all the cool kids do it. :B

 

vxdaYZD.gif

  • Cool 1

Share this post


Link to post
Share on other sites
361
[D12]
Beta Testers, In AlfaTesters
3,112 posts
9,230 battles

You should see if the Win % of the captain effects it.  After your baseline of course.

Edited by Nachoo31

Share this post


Link to post
Share on other sites
2,369
[-K-]
Members, Beta Testers
3,104 posts
6,455 battles

So 55% of the time in a tier 7 battle so far, over 40 games.  The results at 100 will be interesting.  Thanks for doing this.

Share this post


Link to post
Share on other sites
591
[WNSTN]
Members
2,707 posts
16,020 battles

 

Probably not. Feel free to run your own tier 6 experiment if you want to see the results of such.

 

I might do this...my stats be damned!

Share this post


Link to post
Share on other sites
Beta Testers
604 posts
3,501 battles

Those results are even worse than I expected to see, and I thought it was pretty bad already.

 

Umm 22. at 7 3 at t6 and 15 at T5. add the 3 to 15 and you get 18 v 22.

Many people cry +1 MM or same tier only you can safely say that you can combine T5 and T6 MM.

A 4 battle difference at 40 games is not a big issue. I cannot wait to see the next 60

right now its 60% max teir vs 40% at acceptable levels. Sample size is still too small but right now not that big of a difference.

Edited by Battleship_Constitution

Share this post


Link to post
Share on other sites
Members
858 posts
1,808 battles

Those results are even worse than I expected to see, and I thought it was pretty bad already.

 

How do you figure that???  Lert's numbers so far are that he was top Tier 15 of 40 tries, or almost 38% of the time.  That's much better than what the doomsayers say is happening.

Share this post


Link to post
Share on other sites
1,851
[V_KNG]
Beta Testers
9,364 posts

 

Umm 22. at 7 3 at t6 and 15 at T5. add the 3 to 15 and you get 18 v 22.

Many people cry +1 MM or same tier only you can safely say that you can combine T5 and T6 MM.

A 4 battle difference at 40 games is not a big issue. I cannot wait to see the next 60

 

 

No... but if the baseline is valid, the real difference here is the short-sheet approach of going to T6 - only 3 battles. 

Put another way - you're more likely to play T7 matches than T5 which is more likely than T6. If anything - one would have "hoped" for the 22 count to be in the T5 group. 

Share this post


Link to post
Share on other sites
23,989
[HINON]
Supertester
19,896 posts
13,384 battles

The purpose of this experiment isn't to determine how many battles at what tier is alright and what's 'too many'. It's just to create a base line of what to expect.

 

Also, I'd welcome any data anyone else collects, the more datapoints the better.

Share this post


Link to post
Share on other sites
Beta Testers
604 posts
3,501 battles

 

No... but if the baseline is valid, the real difference here is the short-sheet approach of going to T6 - only 3 battles. 

Put another way - you're more likely to play T7 matches than T5 which is more likely than T6. If anything - one would have "hoped" for the 22 count to be in the T5 group. 

 

still to early, willing to bet the MM balance out to 48-52 Again think coin toss. The more you toss the closer you get to 50%. Again +1 MM should be add to at teir MM only for the simple fact most players already want this in MM.

Share this post


Link to post
Share on other sites
354
[SYN]
Members
1,399 posts
15,711 battles

Interesting, so far not as bad as I thought.  Curious about time of day of your testing.  I'm guessing if you play at peak you get uptiered less.

Share this post


Link to post
Share on other sites
23,989
[HINON]
Supertester
19,896 posts
13,384 battles

Curious about time of day of your testing.

 

Outside of peak hours, for the most part. Hence me collecting population numbers, which you can see in my results.

Share this post


Link to post
Share on other sites
283
[HC]
[HC]
Beta Testers
1,539 posts
10,162 battles

Well, at one point MM was going to force 1 top tier game out of every 3 if possible. I don't know if that's still the case. but looks like it's happening by those numbers.

Share this post


Link to post
Share on other sites
Members
1,114 posts
5,917 battles

Well, at one point MM was going to force 1 top tier game out of every 3 if possible. I don't know if that's still the case. but looks like it's happening by those numbers.

 

I've been recording my numbers for quite some time now and it doesn't work like that. In fact, it's very streaky. I started recording my numbers out of curiosity after going 16 games in a row without a top tier game.

Share this post


Link to post
Share on other sites
1,088
[OPG]
Beta Testers, In AlfaTesters
2,926 posts
10,485 battles

I have also continued my experiment of tier 10 matchmaking and after several thousand battles I have yet to be bottom tier. I'm starting to get suspicions that tier 10 matchmaking is biased so that they will never be bottom tier.

  • Cool 1

Share this post


Link to post
Share on other sites
7,594
[NMKJT]
Beta Testers
21,775 posts
3,895 battles

Most of my time lately has been spent in tier 10 introducing battleships to the things they hate the most

 

But when I get home tonight I'll throw in the tier 5 data I've gathered so far from the previous thread, for you, Lert.

Share this post


Link to post
Share on other sites
315
[DIEBL]
Members
1,413 posts
15,672 battles

Not that Im any pro like you guys but Ive done same test 65% tier 7 matches 20% tier 6 and 15% tier 5 matchups that's about 200 games I think I was at was over a couple weeks but it seemed like hard times for tier 5 for sure had to play very cautiously. was playing a lot of Okhotnik, Fujin , Mutsuki ,Furrataka, Kirov,

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×