Replay Analysis Part 1.5: Matchmaking Spread

Welcome to Part 1.5 of my World of Tanks Replay Analysis, where I revisit Part 1 and update the data with more accurate and filtered statistics.

I am now using a different method to filter the data, excluding many points of rogue data and focus only on the Standard, Random Battles. The original data set comprised of 63,555 replays and with the new method of filtering Standard, Random Battles, this figure has dropped to 50,252 (minus 13,303).

Not all, but a large proportion of the removed replays will be Skirmishes and Ranked Battles, followed by other game modes, such as Team Training and special events. This is the data I was keen to remove in the first place, but had not found the (in hindsight clearly obvious) way to filter it out, which is why I am republishing it.

You can read more, including Part 2: Battle Duration and Part 3: Battle Results by clicking here.

How do I define Matchmaking Spread? - Click to expand

There are six possible spreads: -2, -1, -1/+1, =, +1 and +2. At all times during the analysis, -2 means in the replay analysed, the player was bottom tier by two; -1: bottom tier by one; -1/+1: mid tier; =: same tier; +1: top tier by one; and +2: top tier by two. For example, -2 could mean the player was in their Tier 8 vehicle matched with Tier 9’s and 10’s and = could mean the player was in their Tier 6 vehicle matched with only Tier 6’s.

By coincidence, there are six main colors in XVM: Red, Orange, Yellow, Green, Cyan and Purple. In the infographics, the spreads will be colored the same for continuity: -2: Red, -1: Orange, -1/+1: Yellow, =: Green, +1: Cyan and +2: Purple.

Unique Conditions

In Random Battles, it is not possible for:

  • Tier 1 to be +1
  • Tier 1, 2, and 3 to be -2, -1/+1, or +2
  • Tier 10 to be -1 or -1/+1
  • Tier 9 and 10 to be -2

Low tiers do not play on every map. Wargaming change the map rotation every so often and don’t shout about it, making it very difficult to separate these battles from the data. However, I feel this is not vitally important, as the focus is likely to be on Tier 6 and above.


Reading the data...

Of the 50,252 battles...

With a focus on Tier 6 and above:

Most common spread

28.89% of Tier 6 battles were -2,
34.62% of Tier 9 battles were -1,
27.81% of Tier 9 battles were -1/+1,
56.37% of Tier 10 battles were =,
26.07% of Tier 9 battles were +1, and
26.27% of Tier 10 battles were +2.

Least common spread

19.13% of Tier 7 battles were -2,
19.36% of Tier 8 battles were -1,
5.68% of Tier 8 battles were -1/+1,
0.24% of Tier 7 battles were =,
6.73% of Tier 8 battles were +1, and
2.82% of Tier 6 battles were +2.

Most common spread by tier:

Tier 1: =,
Tier 2: -1,
Tier 3: =,
Tier 4: -1,
Tier 5: -1,
Tier 6: -2,
Tier 7: -1,
Tier 8: =,
Tier 9: -1, and
Tier 10: =.

Least common spread by tier:

Tier 1: -1,
Tier 2: +1,
Tier 3: +1,
Tier 4: -2,
Tier 5: +2,
Tier 6: +2,
Tier 7: =,
Tier 8: -1/+1,
Tier 9: =, and
Tier 10: +1.

What data stands out to me?

  • Of 2,864 Tier 7 battles, only 7 were of equal spread.
  • Tier 8 was (almost) equally +1 or +2 and -2 or -1.
  • 40.5% of Tier 8 battles were of equal spread.

These statistics are fascinating - be honest, how many times have you sat there complaining that you've been bottom tier every game? The data contradicts this statement!


Reading the data...

Not including Province and Widepark, as they are exclusive to lower tiers, and Safe Haven, as it is a new map...

Most common spread (by percentage of battles on each map)

-2: Berlin (16.97% of all Berlin battles were -2),
-1: Pearl River (23.57%),
-1/+1: Berlin (12.90%),
=: Abbey (39.04%),
+1: Ruinberg (22.79%), and
+2: Paris (19.28%)

Least common spread (by percentage of battles on each map)

-2: Mines (7.59% of all Mines battles were -2),
-1: Abbey,
-1/+1: Mines,
=: Ensk,
+1: Berlin, and
+2: Pearl River.

On all maps, the most common spread was =.

-1/+1 is least common on most maps. The Exceptions are Berlin, Empire's Border, Ghost Town, Karkov and Pearl River, where +2 was the least common.


Reading the data...

  • =, +1, and +2 were more frequent on NA than EU and -2, -1 or -1/+1 were less frequent.
  • 29.21% of Tier 10 battles for NA are +2, but only 16.87% are for EU.
  • 42.30% of Tier 10 battles on NA were at equal matchmaking, whereas only 30.46% were for EU. 
    • The majority of NA battles were at Tier 10, which may cause a disparity when comparing the Spread frequencies.

Reading the data...

  • Of the 29,475 replays on the EU servers, 17,630 were on EU1, 11,684 on EU2 and 161 on EU3.
  • More battles were +1 / +1, =, +1 or +2 on EU2 (11.50%, 30.78%, 15.32% and 7.77% respectively) than EU1 (10.91% 30.31%, 13.77% and 6.63% respectively) and less were -2 or -1 on EU2 (12.83% and 21.80% respectively) than EU1 (15.36% and 23.02% respectively).
  • Ergo if you want =, +1, or +2, play EU2!

I would very much like to increase the data size and out of all of the statistics I have discovered in this analysis, this is the one I find most interesting - to the point where I am now routinely playing on EU2 (EU1 was my preferred server simply by habit). I have noticed more top tier battles on EU2, but the mere notice of these battles is likely influenced by seeing these statistics!


Thank you for reading. This concludes Part 1.5 of the replay analysis. I have at least one more to go for the Range and will then revisit each category and dive much deeper into the statistics, focusing on very specific areas, such as preferential matchmaking, average battle duration per vehicle class etc.

Leave a Reply

Generic selectors
Exact matches only
Search in title
Search in content