COH2stats.com match and player statistics 2.0

Is Fortified Armor such crap that Jaeger Armor just superior? The ISU doctrines are pretty competitive with each other, Jaeger Armor just... isnt.

Well Hull down on tanks like the elefant don`t seen that useful so the scope is more preferable, smoke too if you need to use it on the elefant, is because most likely you are already surrounded and up ''puddingcreek'' with no paddle, command tank migth be viable with new command patch but rigth now just ins't. the only mostly useless thing in jaeger is the mine(mostly nobody uses haltrack to lay it).

Solid work.

Unfortunately the match details does not have this information. It has only player profile ids. So when processing each match I would have to look up each player. And I can't store this information in our own database, because that might not be up to date - for example when you have new team. So as I see it the only way would be to look it it at at the Relic API. But I am afraid that would be too much stress on the Relic API and they wouldn't like it. For example 1k matches 4v4, would require 8000k API calls for the player profiles.

What does the API give you back exactly?
Scraping the leaderboards might also be possible. Call the site for one player in the team, if the names/IDs of the other players show up, count as AT, otherwise they are randoms. There might be issues with changing. E.g. the current Soviet#1 in 1v1 http://www.companyofheroes.com/leaderboards#profile/steam/76561198018614046/standings -> the HTML code includes the steam IDs of his team mates.

Overloading the calls is indeed an issue, but maybe you can space those out? One call every 10 min would give 144 checked ranks per day. After a week there will be 1000 games annotated.

But maybe that's not that necessary after all.

Totally missed this thread. Thank you for creating this website. It's great to finally have some proper insight into statistics, even if their reliability isn't that high (yet).

Map statistics
- We could generate stats per map, so we could see win rate for each faction on each map and see if
some maps are unbalanced for some factions

Please add this. I think the high disparity in win ratios for Axis vs Allies in teamgames has a lot to do with laney maps like Red Ball, While Ball and Hamburg being the most played maps, all of which obviously favour HMG 42 openings and the heavy tank destroyers late game because there's no way to flank on those maps, and disfavour the use and effectiveness of Allied light vehicles. These factors seem reversed in 1v1 with its more open maps. It would be very interesting to see if the win ratios vary less wildly on other maps.

Is it possible to see it for months?I guess yes but only for future?

And i want to see it before the last unit patch. So when Isu had IL2 bomb.
Or is that not possible?

How it works about top 200?

When 1 single guy is in top 200 that it get count?
Atm you often see this when you search allies 4vs4...


Surprise axis won

I'm not sure how it's done right now, but can there be a filter added so only matches get counted where all participants were top 200 as opposed to just one of them?

Would cut down the number of games, but get rid of the worst of the stomps.

Or games where the average rank was 200 or better for both sides.

Is it possible to see it for months?I guess yes but only for future?

And i want to see it before the last unit patch. So when Isu had IL2 bomb.
Or is that not possible?

How it works about top 200?

When 1 single guy is in top 200 that it get count?
Atm you often see this when you search allies 4vs4...


Surprise axis won

If I remember correctly. If you are top 200 in any faction, you are in respective mode charts (all of them).

And yes. Matchmaking is screwed up beyond belief. Best case is to play in the evenings (GMT +2) when West Europe is playing. Then the matchmaking actually works because higher brackets play. I once played a game at 00.00h (10 PM British time) and got matched with 2x 5000+ ranks vs some rank 4 guy and 2 rank 5000+ (year ago, I remember one of those rank 5000+ had 1 win and ~40 loses with OKW). Basically it was me vs him. It was a good 1v1 (3v3) game which ended with us winning because I decided to leave the VP and completely drive off the other 2 from their VPs. Rage ensued and they surrendered. So yeah, it's all about when you play and in what timezone. If you get lucky, you will get a 2v2 game. Seldom are good 3v3 games where everyone knows what they are doing.

But it's not really MMs fault. It wants to get an average rank game.
Sum(axis ranks)/(# of players) == Sum(ally ranks)/(# of players)

I'm not sure how it's done right now, but can there be a filter added so only matches get counted where all participants were top 200 as opposed to just one of them?

Would cut down the number of games, but get rid of the worst of the stomps.

This could be possible. I have snapshot of the leaderboards for each day. I could search in it for their steam ids.

How to approach it? 1v1, 2v2, 3v3, 4v4 ??
There are several approaches

1. Only one player in the match is in top 200 -- current implementation
2. At least 1 player from each team needs to be in top 200
3. At least 50% of players needs to be from top 200
4. All the players needs to be in top 200
5. something else?

Or games where the average rank was 200 or better for both sides.

I can't get the ranks for all the players in the match So I can't really make average rank.
The previous stats coh2charts.com he had huge benefit of direct access to the relic DB

Overloading the calls is indeed an issue, but maybe you can space those out? One call every 10 min would give 144 checked ranks per day. After a week there will be 1000 games annotated.
But maybe that's not that necessary after all.

The issue is that every day new matches are played and added (the amount of games is insane). Everything needs to be done in a single day so you are ready to process matches next day.

I am little bit scared of Relic. I think it was also reason the last project was shutdown because it was "stressing" their servers too much.

It's not official API. I don't have dev API key or something, it's basically "hacking" their API which they use to display the statistics which you can see on their site and also in-game. They could do some actions - which could make really hard to get this data. To be honest I have no idea how much we are stressing their system. Maybe they did not even noticed it


--------------------------------------------------------------------------------

Additional note to their API. It was not designed to provide a lot of data. It's designed to support the function COH2 currently have in-game / on their site. This makes everything more complicated and the process is fairly ineffective.

Hey, pagep, great work!
Here is an article regarding player rank, in case you have questions regarding ladder placement and skill level:
https://www.coh2.org/news/54445/the-ladder-1v1-matchmaking
Search by phrase 'About a year ago, Relic introduced'

Hope it helps!

Doesn't modes with tiny playerbases (3v3 specifically) have the potential to skew this if those players games are being used in other games?

Here is an article regarding player rank, in case you have questions regarding ladder placement and skill level:
https://www.coh2.org/news/54445/the-ladder-1v1-matchmaking
Search by phrase 'About a year ago, Relic introduced'

That's awesome. However the data does not add up. I can see level with each player, but it's number 0-300, for example:

alias "solaris"

country "ua"

leaderboardregion_id 0

level 300

name "/steam/76561198074006051"

personal_statgroup_id 4686731

profile_id 3153275

xp 18785964

For this player http://www.companyofheroes.com/leaderboards#profile/steam/76561198074006051/standings

I see that level is great. However it's displayed only on the player profile. Not with each match

Doesn't modes with tiny playerbases (3v3 specifically) have the potential to skew this if those players games are being used in other games?

Stats for each mode 1v1, 2v2, 3v3, 4v4 is calculated separately.

Stats for each mode 1v1, 2v2, 3v3, 4v4 is calculated separately.

So I'm just trying to understand this statement:

Basically all games which includes any player who is in any top 200 in the leaderboards . So for example player could be top 200 in 1v1, but we proceed all his matches, so if he played some 4v4, those would be proceed too.

Does that mean if I'm a top 200 player in 3v3, my matches would be counted when I play 1v1? Or just the 3v3 games I play that day? What if I'm top 200 with Wehr in 3v3 and play Soviets in 3v3? Is it still counted?

Does that mean if I'm a top 200 player in 3v3, my matches would be counted when I play 1v1?

Oh yeah. You are right. Those can affect the stats! However what I see from the stats. Usually when someone is really good with one faction they are also pretty good with other too

Oh yeah. You are right. Those can affect the stats! However what I see from the stats. Usually when someone is really good with one faction they are also pretty good with other too

I singled out 3v3 because its a tiny mode. My rank is literally 4 levels higher in 3v3 than my 1v1 and 2v2 ranks despite barely playing it. I think its really dangerous to have a top 200 player in one mode affecting other modes.

I am little bit scared of Relic. I think it was also reason the last project was shutdown because it was "stressing" their servers too much.

It's not official API. I don't have dev API key or something, it's basically "hacking" their API which they use to display the statistics which you can see on their site and also in-game.

Additional note to their API. It was not designed to provide a lot of data. It's designed to support the function COH2 currently have in-game / on their site. This makes everything more complicated and the process is fairly ineffective.

Wait is it possible that you make the problems? Because since yesterday alot of people complain about drops, lags, bots not reacting after 1 person dropped.

Or was it just relic self who fucked ip something in the maintenance?

And about the ranks question. Well hard to tell, about how many people should be in top 200.
But the acutal version 1 makes not much sense as you can see in other threads.

So if you really want to have only top 200 all should be top 200. I know in 4vs4 you will get less information, but in 1vs1, 2vs2 the data should be very well and good!

Oh yeah. You are right. Those can affect the stats! However what I see from the stats. Usually when someone is really good with one faction they are also pretty good with other too

In 3v3 and 4v4 there are tons of players who are terrible (like 45% wr) but have one or two bloated 50-150 ranks because they grind or get carried a lot. Odds are that they do terrible in other modes or with other factions. I hope those wouldn't influence the stats too much.

Wait is it possible that you make the problems? Because since yesterday alot of people complain about drops, lags, bots not reacting after 1 person dropped.

No I don't think so. The stats API should be completely separated. Also I am talking to the API for only ~30 minutes/day. 2 AM UTC time. Also the scrapping is running from 8th of March the same.

Oh yeah. You are right. Those can affect the stats! However what I see from the stats. Usually when someone is really good with one faction they are also pretty good with other too

1v1s play entirely differently from every other game mode so its rare(some exceptions) you'll find primarily team game players transitioning to 1v1 and instantly being good at it. some factions also play entirely different, ex, USF playstyle is the polar opposite of Ostheer. while true to some extent (if you are top 200 as one faction you are better at others), Its not always the case.

All that being said, when analyzing statistics between games all the players involved should be top 200, while its rare team games meet this category the games that make it will be infinitely more valuable when basing balance decisions off of it.

That's awesome. However the data does not add up. I can see level with each player, but it's number 0-300, for example:

alias "solaris"

country "ua"

leaderboardregion_id 0

level 300

name "/steam/76561198074006051"

personal_statgroup_id 4686731

profile_id 3153275

xp 18785964

So if I understood you right there no information about the rank or ladder position in that piece of data. Perhaps someone from coh2.org knows how to do that? They have this tool for it https://www.coh2.org/ladders that contains almoust everything leaderboard does.

In 3v3 and 4v4 there are tons of players who are terrible (like 45% wr) but have one or two bloated 50-150 ranks because they grind or get carried a lot. Odds are that they do terrible in other modes or with other factions. I hope those wouldn't influence the stats too much.

(3v3 experience)
Not necessarily. When I started playing competitive, I had 13 wins in a row. Started at rank about 150. Then I had another 3 or something wins in a row of games I didn't even play because somebody DCed on enemy team. Got to rank 100something. Then I had about 10 loses in a row. A couple of DC, flaming teammates and just bad games from either my part or we played badly as a team. Loses but good games if you know what I mean. After that I oscillated between 100 and 300. Got to top 20 on a couple of occasions, even won vs top 10 players, but again dropped to mid 100s. Rinse and repeat for past year and now I'm stuck at 100. Rank 1500+ really do have trouble dealing with MG42s or pak howis or mortar encampment. But TBH, I haven't encountered a lot of players in top 400 that don't know what to do. Most of the time it was just bad decisions at the worst of times that cost you the tank or two and kaput. That or picking commanders that do not have a response to an enemy build (most of the times it was no offmap for lefh spammer).

My biggest success in COH2, being a casual player, was that top 20 position. Winrate? 52%
Were there games that I got carried? Sure
Were there games where I carried? Sure

I would drop the "tons" adjective. Because in top 200, that would mean around 150 are getting carried.... By whom? Who is carrying those people? In a 3v3 game, who is constantly carrying those people? And furthermore, if they are getting carried, it must be from some top 50 or sth player. Where are all those top 50 players? Also, the MM is designed so that the ranks are always averaged against the teams. If there is a top 50 player on your team, that means there is a top 50 player on the enemy team as well, or even better.
Next, it's a 3v3 game. If you have such a big liability in your team (the person getting carried), then you really have to carry the s*** out of the game to win. Because a liability in 3v3 means a lot of veterancy for the enemy and not a lot of damage from your team.

And lastly. These are random 3v3 games. Can you say with utter certainty that the person you PNGed is really a bad player? 50% winrate is pretty much the definition of "you lose some, you win some". In a game of 6 or 8 RANDOM people that's hell of a winrate. Especially considering the fact that if you drop to rank 600+, you're going to have a hell of a time getting to rank 500. Constantly getting paired with flamers, kids or people that have 2 left legs for arms.