Introducing WNX
By Goldflag
WN8 has been the most significant World of Tanks player skill metric since I started playing the game in late 2014. I remember religiously tracking my stats on Wotlabs every single day in 2015-2018. Everyone wanted their overall WN8 to reach the 2450/2900 threshold to become purple. The E50 and M46 were the most popular stat padding tanks, and even today nearly 10% of my account's total battles are in the E50. WN8 was so important to people that some even rerolled their accounts just to have better overall stats.
We always knew the lack of assistance damage in WoT's public API made WN8 an extremely flawed metric. Despite this, WN8 was still a lot better than Personal Rating, so it was what everyone used.
However, the relevance of WN8 has slowly faded over the years. Marks of excellence — which actually included assistance damage — became what elite players played for. Wargaming also replaced PR with the much better WTR, though it hasn’t gained mass adoption because WG has kept the implementation proprietary. But most importantly, WN8 just wasn’t that good at telling you how good a player was because it’s so easy to game. You still had to go look at other stats to get an accurate picture.
What happened to WN9?
WN9 was developed around 2016-2017, but it was never implemented for two major reasons.
- WG hadn't added assistance damage to the API, so no matter how much you improved the formula, you could never make any significant improvements to WN8
- The creators of the WN series of stats and the community stat trackers had quit the game. Vbaddict died in 2017, and most of the other sites ceased further development on the stats side.
Assistance damage finally gets added to the API
Wargaming unexpectedly released tank-level assistance damage in 2025. I didn’t find out about this until 2025, but when I did I immediately knew I had to create a WN8 replacement.
The Tomato.gg mod does track assistance damage (as well as a myriad of other stats never found in the API), and I considered adding a WN8-like stat that included assistance damage for Tomato.gg mod users at times. But I didn't think true adoption would happen unless it was available for the vast majority of players who didn't run the mod.
WNX Details
You can see more detailed information on expected values and formulas here.
Better expected value calculations
WN8’s expected values are based on what an above average player (~75th percentile) would score in a tank. I’m sure the designers had their reasons for this, but I’ve made the expected values of WNX true to what the name implies — what the 50th percentile player should achieve.
Clone tank comparison
WNX’s expected values should also be more accurate than the ones XVM produces for WN8. This can be seen through the expected values of clone tanks that have different levels of average player skill. The WNX methodology is better at normalizing player skill.
| Tank Comparison | WN8 | WNX | Winner |
|---|---|---|---|
| Stone Sentinel vs. Bisonte | 1462 vs. 1375 (6.3%) | 1119 vs. 1119 (0%) | WNX |
| Mars vs. Progetto 46 | 1348 vs. 1278 (5.5%) | 1151 vs. 1130 (1.9%) | WNX |
| Monkey King vs. 121B | 1783 vs. 1813 (1.7%) | 1476 vs. 1458 (1.2%) | WNX |
| Hurricane vs. CS-63 | 2075 vs. 1814 (14.3%) | 1546 vs. 1565 (1.2%) | WNX |
| Miel vs. Bourrasque | 1348 vs. 1283 (5.1%) | 1155 vs. 1150 (0.4%) | WNX |
| 50TP R. vs. 50TP pr. | 1354 vs. 1315 (3.0%) | 1119 vs. 1117 (0.2%) | WNX |
| Peregrine vs. IS-3A | 1425 vs. 1354 (5.2%) | 1198 vs. 1167 (2.7%) | WNX |
| Mauerbrecher vs. VK 168.01 P | 1227 vs. 1245 (1.5%) | 1045 vs. 1075 (2.9%) | WN8 |
| Skorpion vs. Skorpion G | 1399 vs. 1360 (2.9%) | 1229 vs. 1209 (1.7%) | WNX |
| Alembic vs. Astron Rex | 1396 vs. 1357 (2.9%) | 1159 vs. 1128 (2.7%) | WNX |
| T26E5 P vs. T26E5 | 1182 vs. 1179 (0.3%) | 1028 vs. 1020 (0.8%) | WN8 |
| Defender vs. Obj. 252U | 1340 vs. 1371 (2.3%) | 1156 vs. 1152 (0.4%) | WNX |
Removal of winrate and decap
We didn’t include winrate because it's already the primary non-WN8 metric. Winrate shows how many games you win and WNX should give you context about how well you played, win or lose. Let’s keep them independent! The other reason is that WN8 is used as a single-game or session level metric just as often as it is used as a last 1000 game or overall metric, and winrate ends up being noise unless you have a large sample of battles.
Decap is absent from WNX because it has always played such a small role in the game. The typical player averages 1 decap point per game. It's never going to provide any meaningful data to the calculations.
Similar scaling to WN8
The way WN8 scales is ingrained into everyone's head and we don't want to change that. I've scaled WNX to line up with WN8 as close as possible — close enough for them to use the same color scale. If your WN8 is 1500, chances are your WNX will be very close to 1500. If you never spot, it might be 1300. If you spot a lot, it might be 1700.
New color scale
We’ve also updated the color scale for winrate and WN8 to better reflect WoT in 2025. We're still using the old Wotlabs colors. You can see the color scale here.
What's next
The version of WNX we released today is a rough beta. We’re listening closely to community feedback and everything is still up for change.
As for WN8, it won't be going anywhere since it will always have its place because of its historical value.
On the site, I've updated the main overview of player stats to include WNX. Most other pages will still only show WN8, but expect that to change slowly over the next few months.
We only started collecting assistance damage data around June 20th, so all battles before that will show up with 0 assistance damage. This means that unless you've played 1,000 battles since June 20th, the recent WNX displayed on your profile is lower than the real number.