lft am demo

Playing in Newcomer/Amateur this season? We're interested in compiling a frag video this season! We're going to attempt making a NC/AM S13 Frag video and with your help, we can make it happen!

DM me demos with ticks to any insane frags (airshots, etc.) and we'll be cooking real good. We NEED MORE CONTENT!

Hey y'all, I was inspired by this spreadsheet and decided to make one for our division. With lots of help from sol, vohh, and Gabiduo, here it is! Feel free to check it out, but please don't take it too seriously; this was just for fun c:

unbelievably based spreadsheet

tf2 team is the best tf2 team

the silver lining to getting obliterated by catgirls is I get to see if my spreadsheet is properly HUMBLING me

I was away for week 1, so not sure why I have scout stats on your spreadsheet.

I must have unconsciously substituted Diop's stats for yours; it's fixed now. Most of my time working on this was very early in the AM.

As for week 2, I'll be updating stuff slowly as teams play more of their matches.

This is actually pretty cool. How do you determine team/player rankings?

vibes

For medics specifically, using the stat of total deaths instead of deaths/minute naturally biases you against close/long games since you will naturally die more even if you are actually surviving more situations than a faster game. For medic players it's better to lose 0-5 fast and die 5 times with good heals than to lose 4-5 and die 9 times with good heals, despite the fact you likely performed better in the second case.

Also since the most important stat for combat classes is KDA, the soldier stats are kinda silly right now. Every team has a low scoring soldier since every team has a roamer.

We take the logs from each match (usually two matches per team) and average each player's DPM and KA/D (HPM and total deaths for medics). Using the maximum and minimum values for each stat (separated by class), we normalize the averages to a scale of 5 and add them together, generating a player's score out of 10.

"Considering deaths/minute over total deaths may not necessarily be a bad idea, but it does have some problems of its own. Putting aside having to add a time variable for each case, there are instances where people play short games w/ low deaths and long games w/ high deaths. This prompts the question of 'how do we separate these?'

Performatively, you can equate long games w/ many deaths to short games w/ few deaths. In essence, short games = less data to work with so even if someone gets rolled, it's not really clear how their performance shines, which is why we just settled for overall deaths as it's a fairly good hallmark for med performance." -Sol

Although DPM and KA/D are weighted equally in our calculations, the flank roles will naturally perform "worse." The immediate solution that comes to mind would be separating the roles within classes. For example, this would generate the roamer scores relative to each other, and the scores would appear more consistent with the rest of their team's.

Theoretically, I could use the "healing received" stat to differentiate who plays what, but for now, having a lower scoring scout/soldier is a consistency across all teams. If enough people demand, I'll consider separating the roles, but doing so would be no small feat.

Please understand that this is a personal project I issued to myself for fun. In doing so, I sparked interest in some friends to work alongside me. Thanks for all of the feedback so far, keep it coming c:

LF more criticism and viewpoints to further modify this. I hope this post is RGL-safe as I'm still banned lol

Will there be combined stats for all games played or is it just week by week?

For now, it's week by week. Though possible, adjusting our current functions to account for the data in 16+ matches as opposed to 2 is a nightmare. Never say never, but as it stands, a combined season page would likely require a complete rewrite.

https://www.youtube.com/watch?v=RymXiada3uE

It really comes down to if I play 2 hours of q3practice with freeman or if we just bite the bullet one night

EDIT: this can and will happen (amphi ctf and the edits to account for a season-wide comparison). While ball is life, I can get this approved by the wife

Formulating a niche metagame comes before formulating the totally real and objective skill levels of the boys.

soldier stats should be based on the number of rocket jumps they do in a match

on a more serious note, maybe I'm alone in this but I find dpm a much more useful stat when compared to dtm, maybe a ratio of the two would give interesting stats

I had a go.

There are a bunch of issues that will come up when trying to do this, namely that if you're doing the lookup based on a static team list, those players don't always play - so you're gonna be running into stats being recorded under Player A, when backup Player Z is actually playing, which is gonna result in stolen valor. #salutedatroops

I tried to set it up to be a little more automated, and to allow for all data to be pasted into one tab, and that the lookups are done based on the log, and thus the name of the player in the log.

Orange Headers = Formulas that should auto populate over time
Yellow Headers = Formulas that will have to be dragged down
Blue Headers = Calculated Cells that will have to be dragged down
Green Headers = Cells that have to be manually entered

Log data gets pasted into column M on the 'DATA' sheet. I'm sure there is a way to automatically pull this if you have the log url (JSON API that doesn't exist by default in Google Sheets??) but I am a monkey and don't actually know how to use a computer.

If anyone wants to chat about spreadsheets lmk ;))

Very impressive, thanks DJ BIG GYG. We gave it the old college attempt but master class came through. Yea lol we cba for the log scraping, so we started it off with just testing if raw pasting could even work the way we wanted. I guess from here it’s just fixing up scores, maybe toggling options to see how heals received/DTM or other variables can better represent performance in a stat-driven world

https://www.maxim.com/wp-content/uploads/2021/10/american-psycho-business-card-scene-promo.jpg

Nah in all seriousness this is incredible. You can really see the Mister Milano mentorship shining through in your optimizations. This is turning out to be an invaluable learning experience so I appreciate everyone's enthusiasm on this topic c:

IM PLAYERS - We are also doing IM/Div-1! Please DM me clips on discord siyo_. (or nightfaul on discord) and if we like it, we'll put it in our folder.

Thanks!!

MATCHES ONLY TOO BTW

Lets go tf2 team!

Just thought I'd make a quick update post.

• Thanks to ggglygy, there's new automation pertaining to gathering match pages/logs/data
  
• All players should be accounted for (despite the difficulties determining who's a main-roster player)
  
• The standings are now a combined live depiction instead of a week-to-week comparison

My goals for this past week have been figuring out the automation and imports, so how we determine scores has not yet changed. That being said, if anyone's interested in helping us optimize the sheet further, feel free to leave your feedback below or dm melons_ on discord c:

ALSO PLEASE LMK IF I GOT YOUR ROSTER WRONG! Right now each team only has six players to keep their total score fair. To do this I simply chose the players with the highest scores, so if your roster is incorrect, dm me the correction and I'll fix when I can!

Good luck on your matches this week, everyone!

bruh moment starting med is owenchungus, not just jazz

how the mighty have fallen