MMR Data

[daresound]

interested to see if a cohesive lower rating team can defeat a high rated low cooperation team...
Kim, stay off my freq.

[cubone]

cubone

If you wanted to incorporate MMR ratings into your data, I've made it available from the API:

https://trenchwars.org/api/mmr/ratings/{seasonID}/dd
https://trenchwars.org/api/mmr/ratings/{seasonID{/jd
https://trenchwars.org/api/mmr/ratings/{seasonID}/bd/{shipType}

https://trenchwars.org/api/mmr/ratings/1/jd
https://trenchwars.org/api/mmr/ratings/1/bd/3

Nice. Thank you! I’m assuming this is the current values but not historical values?

[cubone]

Match/player data up to 16/06/2023: https://drive.google.com/file/d/1IZU...ew?usp=sharing

Thanks again!

[Hookecho]

cubone

If you wanted to incorporate MMR ratings into your data, I've made it available from the API:

https://trenchwars.org/api/mmr/ratings/{seasonID}/dd
https://trenchwars.org/api/mmr/ratings/{seasonID{/jd
https://trenchwars.org/api/mmr/ratings/{seasonID}/bd/{shipType}

https://trenchwars.org/api/mmr/ratings/1/jd
https://trenchwars.org/api/mmr/ratings/1/bd/3

[BIET]

Match/player data up to 16/06/2023: https://drive.google.com/file/d/1IZU...ew?usp=sharing

[cubone]

BIET Just doing my usual follow-up to see if you can post updated MMR files. Thank you!

[cubone]

Turban beat me to the punch a bit by making his own sheet so I decided I would take the data I've gathered thus far and add it to Google Sheets to make it a little more user friendly and digestible.

Data has been filtered only for people with at least 30 games in TWJD3.

The 'Stats' tab is summarized data from TWJD3 games with my attempt at an overall rating/score on the far right. Similar to Turban, I looked at KPG/KPM (modified to remove strays and TKs), average range of kills, win percentage, KOs per game, and average W/L margin.

The 'Player Win Pct' represents each player combination (minimum of 5 games together) and their winning percentage as a team.

The 'Player Kill Pct' represents the % of each player's total kills from every other player. The way to read this one is find your name in column A and then read across to see what % of your total kills were from each player. This isn't the most useful as-is since I need to adjust for % of games each player has faced each other. Right now, Bombed might represent 5% of my total kills, but I might have faced him in 25% of my total games. Ideally I will adjust this in the future to look at expected kills based on # of games vs actual kills.

Things I would like to add at some point:

• Based on overall player score/rating, look at the average score of each player's kills (do they hunt noobies or kill similarly skilled players?)
• Go through game-by-game stats and calculate the number of 2fers/3fers/4fers/5fers??? for each player
• Other suggestions people might have

https://docs.google.com/spreadsheets...gid=1764461569

[cubone]

Thanks for crunching the stats. I have 98 games, though, not 60 - is some data missing?

The data is from December 2 through the day Biet posted it so missing the beginning months of MMR and the past 2 weeks or so

[MHz]

Updated with latest MMR data. A few ideas I'm working on whenever I'm bored:

• W/L pct for player combinations
• Looking at which players each player kills the most (does a player hunt noobies, do they kill mostly good players, etc)

Thanks for crunching the stats. I have 98 games, though, not 60 - is some data missing?

[beam]

not surprised at all to see pawner the goat as my highest win % teammate

[cubone]

Okay, meant to work on this for a bit but never got around to it. It's done now. This shows the win percentage for all combinations of players that have at least 30 games played. In order to keep it a bit cleaner, I also required each combination of players to have at least 5 games played together in order to show the percentage.

The way to read it is the y-axis is you and x-axis is each person you've teamed with. Diagonal shows each players overall winning percentage to give you a gauge of how they perform below/above their average based on who they are teaming (or stuck) with.

[cubone]

Okay newest project from Cubone Data Services. This shows the % of each players kills for every player with at least 30 games played. Obviously players like Bombed and Greed are green for a lot of players because they have the most games played so naturally they would account for a lot of kills for other players.

[Rasaq]

Just look at the respect I have for my teammates. I have the lowest TK percentage. If there was a way to see how many times I was tked and tk ko'd I'd be number 1 in those categories. My teammates have zero care.

[cubone]

Updated with latest MMR data. A few ideas I'm working on whenever I'm bored:

• W/L pct for player combinations
• Looking at which players each player kills the most (does a player hunt noobies, do they kill mostly good players, etc)

[Jessup]

Worst player ever. On and off the game. 1000s of people can confirm.

Must you always? That is quite the claim meanie. Come play and prove you are better Mr. 6.99 Average kills or are you retired? I'm confused all around by your input.