Profile of Siphon X.

I just worry about getting too granular.

Well, depends a bit of what point one tries to make, doesn't it? If the statistics end up being all over the place, yes, you won't know why that is.

However, it turns out (spoiler alert) that unlike ESL certain doctrines dominate for their factions. Now, if you have only a limited number of people contributing, it might seem like it's due to that (you know, because DevM is totally into German Infantry). If you add more people that also played on more maps, and they also all used German Infantry, you have a much stronger case that German Infantry is (perceived to be) damn strong right now.

Btw, right now I don't plan to include any win ratios for e.g. certain doctrines or tech choices because a) I don't think we have enough data to get much meaningful and b) it's a bit awkward to do

PS: If possible could an admin change the post title to "In Game Screenshots" Thanks

Ok

Ok, as we know there are units that fare better at short ranges, while other are better at long ranges. New maps typically try to have areas where both sorts of units can shine.

Now, it's quite obvious what the ideal area for a long range unit would be: Just a flat area without any obstacles. Conversely, CQC units should be happy in a heavily build up area (like a city), right?

Well, I think it's not as simple, and my feeling is that certain layouts are frustrating for both sides...

Here are a couple of sketches, how a build up area could look like:



The blue boxes could indicate buildings or other sight and shot blockers.

Example a) has a lot of shotblockers, but obviously is not a good map for CQC units because there lanes without any flanking routes, so a CQC unit would be spotted and shot at from a distance already.

Now, example b) has lots of flanking routes (this looks much like the original Stalingrad). Is it the ideal CQC map? I'd say no, because a unit parked on a corner of one intersection will still have long sight axes. Every unit trying to creep up has a high chance of being spotted. So, parking a long range unit at suitable corner would still make it hard to flank it without the other player noticing. However, as soon as the long range unit starts to move around, there is a chance of encountering a CQC unit around the next corner.

So, I'd say this kind of layout is actually quite frustrating regardless of what type of unit your are using. It's bad for long range units as they have to fear being demolished as soon as they move around the corner, but short range units have problems closing in to long range units undetected. So in effect both types of units would be parked somewhere, the CQC unit around a corner for an ambush, the long range unit at an intersection. Doesn't sound like a lot of fun, eh?

Now, c) with a staggered pattern in one direction: Ok, somewhat better, assuming that the bases are left and right. I varied the lengths of the blockers to not create new diagonal lines of vision. Still, the there are long lines in N-S direction, so getting from left to right without being seen anywhere will be tricky. So while maybe a bit more CQC-unit friendly than a), the problem is still there for both types of units when moving E-W.

In d) I tried to create a pattern that has no long lines of vision while still retaining at least three paths to get from left to right. To me this seems like a CQC unit heaven where long range units have little business of being around.

A positive side effect: d) is much closer to what European cities look like

Makes sense, or complete nonsense, what do you think?

Edit:

My hypothesis is merely that a layout like b) and some extend c) isn't as great for CQC units as one might think and in fact awkward to move around in for all sorts of units and therefore probably should be avoided.

If you you want to include a CQC part in a map, I'd say it should more look like d).

1. Personally, I would like to have a summary about what is the preferred light-vehicle meta play/counterplay.

For instance, which and how many light vehicles were fielded (was it a 222, a t70, 888)?
- What kind of non-vehicle counterplay was employed to counter the vehicles (e.g,, pgrens, guards, AT gun, Partysans)?
(and of course, who ended up winning, in the end)
- If it is trivial to get, ideally how much bleed the light vehicles ended up causing in the end (reinforcement / wipes).

I'll see what I can do there. It should be possible to do at least a progression of builds (like, "222", "Guards", "PAK", "T-70"). Which would be the relevant entities here?

Ost: 222, Panzerschreck upgrade, PAK, Puma, StuG-E (?), FHT (?)
SOV: Guards, AT Partisans, AT-Gun, T-70, PTRS upgrade (Tank Hunter)
OKW: Luchs, Puma, Raketenwerfer
USF: M-20, M-8 (lol), AAHT, Bazooka Upgrade (Cpt), Bazookas, Stuart, M1 ATG
UKF: AEC, Tank Hunter Section, ATG

Anything missing?

Re: Who ended up winning
If the game was lost by VPs ticking out, there is no info on this in the replay file. Also, not sure how meaningful that is unless the game ends at about 13 minutes.

Bleed is tricky, wipes are impossible to get. Reason being that the replay files only contain player commands, not what those commands resulted in. So, no record of a unit being damaged or wiped. "Reinforcement" is a command and thus is in the files, but "vault" (the backend of cohdb.com that I use to analyze the replays) doesn't decode which unit gets reinforced. Also, you have no idea, which unit did the damage, so...

2. Regarding teching choices, it would be nice to see which side-techs are the ones that are most commonly skipped.

Sure, no problem.

I wold say wpc only. The reason I say this is that the new maps being tested in the map contest tourney caused a shift in the meta as people learned how the maps functioned in the first place. Furthermore many of those maps were designed as a reaction to the current meta. Combining them would actually obsure the value of what the current meta is like.

I don't think the maps introduced a significant shift in meta. Actually, as long the maps are not well know, I figure most player will tend to stick even closer to what they think will work in most cases (aka, the "meta"). Then again, I probably would be more hesitant to use the newer replays, if they were just some weired ass maps, but 6 out of the 9 are now in the automatch pool, so...

Even if there was a shift: Ok, there might be some historical interest in what the meta was during WPC, but I guess for most people it is more interesting what the current meta is so they can either use it (or try to go against it).

I haven't looked at both in depth, but like Tric said: In terms of commander choices they are very, very similar. The top three OH/USF commanders are identical, the top SOV commander is identical, and the rest very similar, likewise for OKW and for UKF there are so few WPC replays that I actually would not do a review there are all...

You should still run analysis on both since this will provide some info as to how much maps affect the meta. If they don't differ then no harm in including it. If they do, you can see what changes (though 41 is small, but can't be a chooser) occur because of map difference. Though you would also need to consider the player difference as well, which will confound trends and make your conclusion a little more murky.

Actually, the player difference to me is actually the strongest argument for merging the two. As ESL showed personal preference played a very important role even for the top of the tops. The most replays for WPC are involving either Devm or Jesulin. So any analysis would be biased towards their preferences (which to some extend is justified, because apparently it worked out for them). However, neither of them participated in tric's tourney, so this gives a little more weight to other players like HelpingHans, Noggano, Talisman, Paul, Barton and Price.

Well, the RNG works like RNGs in computers normally do: They do spit out a (seemingly) unpredictable series of numbers, but they have to be initialized/seeded with a certain value. If that seed value is the same, the RNG will spit out exactly the same sequence of numbers.

And indeed, the replay file contains the RNG_SEED, the number with which the RNG generator is initialized for this specific match.

This means that when the replay runs and an RNG number is needed, the RNG will spit out the same number that it produced originally. So, if the shot penetrated the first time around, it will also penetrate the second time.