MMR Data

[qan]

I would like to map out all of the kills/deaths based on the coords but not sure if anyone has the coords of the actual arena itself. Ballpark estimate is [6590-9791, 5342-8865].

I know BIET did this at one point for DDs. I can't remember for certain, but I think found that more deaths actually occur slightly to the right and bottom of the arena center, meaning the team that spawns top left is probably more naturally aggressive at the start of a round through that positioning. Odd, because radar covers that area if you're firing to the bottom right. Well, whichever way it was biased, we ended up deciding to randomize spawns so that a team would not have a consistent turf advantage. I think the data may have even showed that the first team tends to win slightly more on average than the second. BIET might still have some of the data. It was really interesting. Think there's a forum thread somewhere.

If you'd like to calculate arena coords yourself and see if there are trends in MMR, go to your Continuum folder and look in /zones/SSCU Trench Wars. All .lvl files (corresponding to arena names) are downloaded there. Then open it with the DCME map editor, found here: https://trenchwars.org/index.php?v=downloads Then to get pixel position in the arena, multiply the tile coord by 16. I could just pull it up myself but you seem interested enough to put in the small amount of effort needed to see it.

The position will be off a bit due to how the bot tracks players in the arena. It changes who it spectates every 200ms, so with an arena of 10 players, at the absolute worst, a position will be 2 seconds stale. However, a death by a player is more likely to show an accurate position because at least one other player is within the vicinity (meaning when the bot is spectating them, it will receive the dying player's position packets, too), so you're usually looking somewhere below 1 second of position staleness for positions recorded on death.

[cubone]

I know BIET did this at one point for DDs. I can't remember for certain, but I think found that more deaths actually occur slightly to the right and bottom of the arena center, meaning the team that spawns top left is probably more naturally aggressive at the start of a round through that positioning. Odd, because radar covers that area if you're firing to the bottom right. Well, whichever way it was biased, we ended up deciding to randomize spawns so that a team would not have a consistent turf advantage. I think the data may have even showed that the first team tends to win slightly more on average than the second. BIET might still have some of the data. It was really interesting. Think there's a forum thread somewhere.

If you'd like to calculate arena coords yourself and see if there are trends in MMR, go to your Continuum folder and look in /zones/SSCU Trench Wars. All .lvl files (corresponding to arena names) are downloaded there. Then open it with the DCME map editor, found here: https://trenchwars.org/index.php?v=downloads Then to get pixel position in the arena, multiply the tile coord by 16. I could just pull it up myself but you seem interested enough to put in the small amount of effort needed to see it.

The position will be off a bit due to how the bot tracks players in the arena. It changes who it spectates every 200ms, so with an arena of 10 players, at the absolute worst, a position will be 2 seconds stale. However, a death by a player is more likely to show an accurate position because at least one other player is within the vicinity (meaning when the bot is spectating them, it will receive the dying player's position packets, too), so you're usually looking somewhere below 1 second of position staleness for positions recorded on death.

Thanks Qan. I know I've looked at some of BIET's stuff and yeah, pretty much exactly what I'm wanting to look at for MMR.

[cubone]

Been messing around with kill coords and plotting on map. Here's an example:

[LeeRomeno]

Okay, was kind of bored and have parsed all of the event data so I have coords, times, team coords, etc for every kill/death for every game. There's a lot of data and a lot of possibilities but we'll see how motivated I remain. If anyone has a data point they are curious about, let me know and I might look into it.


I decided to look at the average distance for all deaths (in jav). This is calculated using the X and Y coords of the killer and player killed and then calculating the Euclidean distance between those points.


Dear Thesauce aka WALL-E,

It gives me great pleasure to inform you that you have been awarded the "Open Your Eyes Award" brought to you by Braille Institute. Your despicable performance in the field has set a new standard for the competition.

Your achievement of being killed from the furthest average distance away is a testament to your lack of skill, poor strategy, and various disabilities. Your ability to die at a distance is truly remarkable.

On behalf of the entire Trench Wars community, I congratulate you on this remarkable feat. Your dedication, passion, and sportsmanship are an inspiration to us all.

We look forward to seeing you continue to excel in the upcoming seasons and push the boundaries of what is possible in Trench Wars.

Once again, congratulations on this well-deserved recognition!

Sincerely,

Cubone

i wonder if i should switch to higher resolution lol. this is embarrasing

[cubone]

i wonder if i should switch to higher resolution lol. this is embarrasing

I've kind of hit a wall for now in terms of entertainment value, but I was thinking about looking at kill distance and comparing to reso to see how many kills were actually strays and how many were aimed. My reso is 3400x2000 or something and I have a fair amount of R3/R4, but I can see like 95% of the arena on my screen.

[banzi fuq]

Random ideas
-john wick award: most kills in 3 minutes
-jerk award: most kills in spawn area
-sitting duck: most deaths in spawn area
-deadly duo: player combos with high win %

[cubone]

Random ideas
-john wick award: most kills in 3 minutes
-jerk award: most kills in spawn area
-sitting duck: most deaths in spawn area
-deadly duo: player combos with high win %

Already had this one up.



The (S)pawner Award

Presented to Pawner, the undisputed champion of not being able to escape spawn!

In a game where most players can make it to base, pawner has managed to turn dying in spawn into an art form. With snail-like reflexes and zero spatial awareness, they have managed to catch more bullets than 50 Cent.

Some might call it luck, but we know better. Pawner has failed to hone their skills, using every spawn as an opportunity to survey the battlefield, gather intel, and ultimately die 0.3 seconds later.

So let it be known: if you're looking for someone who knows what base looks like, Pawner is not your person. Congratulations on being the ultimate Spawn King/Queen of Trench Wars!



All of (S)pawner's Kills/Deaths in spawn:

[Jessup]

The spawn zone in javs is atrocious. Why we could put the effort in to fix wb spawns but still have this tiny boxed in area for javs is really odd to me. PS... stop using your selective stats to do trolling cubone.

[Kassius]

i wonder if i should switch to higher resolution lol. this is embarrasing

It's about flicking your eyes to the radar often, but not gazing at it, whilst you know you have a second or two of relative safety vs on-screen opponents. Keeping lateral momentum high, and ensuring if enemies are closing from off screen, you try to match their momentum away from them as soon as possible to help avoid being vulched.

[Tiny]

-jerk award: most kills in spawn area

I want this so badly.



Also, can we get the MMR Rating to left of the name like in TWDT with the ratings on the score screen please?

[cubone]

I want this so badly.



Also, can we get the MMR Rating to left of the name like in TWDT with the ratings on the score screen please?

Spawn kills is easy. The table above is specifically spawnkills where both people (killed and killer) were in spawn area. I filtered out any kills where at least one person was outside of spawn (e.g. straying into spawn from base).

The problem with MMR rating is it changes constantly and it's not in any of the data that I have. To get MMR rating, I would have to manually lookup each person through :MMRBot:!rating player. If the ratings are in a data table or something, it would be simple to add them to the data.

If I had MMR ratings by date, I've thought it would be interesting to look at who people kill the most. For example, maybe someone has a good KD, but they really only kill low-rated players versus someone who kills mostly people at 3500+.

[Cape]

Win percent in highest rated games would be a cool stat.

- Highest multi-kill % per game

[cubone]

The next award goes to goes to someone who is no stranger to MMR awards. The Trench Wars Humane Society presents the King of the Strays Award, given to thesauce for his complete inability to aim and reliance on a little old fashioned luck. Mercede$ and YOGI were hot on his tail, but thesauce has proven to be the superior strayer. Not surprisingly many of the names on this list are stalwarts on many of the other lists which are generally measures of poor performance.

The math of this is for each kill, taking the X and Y coords of both the player killed and the killer and then looking at the killer's screen reso to determine if the player they killed was actually on their screen or not. Nothing fancy here.

Also, reminder to spay or neuter your pets.

Edit: Pct_vs_League indicates the percentage of kills from strays compared to the league average. Everyone here is overly reliant on strays compared to the overall league average.



And here are the kills from the recipient (who has quit this zone because it's garbage yet continues to hang out here) of this very prestigious award:

[cubone]

Win percent in highest rated games would be a cool stat.

- Highest multi-kill % per game

What do you mean by highest multi-kill %? Percentage of someone's kills that are 2fers, 3fers, etc?

[ag$]

Already had this one up.



The (S)pawner Award

Presented to Pawner, the undisputed champion of not being able to escape spawn!

In a game where most players can make it to base, pawner has managed to turn dying in spawn into an art form. With snail-like reflexes and zero spatial awareness, they have managed to catch more bullets than 50 Cent.

Some might call it luck, but we know better. Pawner has failed to hone their skills, using every spawn as an opportunity to survey the battlefield, gather intel, and ultimately die 0.3 seconds later.

So let it be known: if you're looking for someone who knows what base looks like, Pawner is not your person. Congratulations on being the ultimate Spawn King/Queen of Trench Wars!



All of (S)pawner's Kills/Deaths in spawn:

ofc best, tiny, and riverside top this
They love laming spawns
lmao