Jump to content


Statistics suggestion: win/loss-specific stats, average position on team

statistics personal rating pr position place winrate

  • Please log in to reply
11 replies to this topic

Poll: Additional Statistics Poll (11 members have cast votes)

Should these stats be added to the WoT API?

  1. Separate stats for wins and losses would be good. (2 votes [14.29%] - View)

    Percentage of vote: 14.29%

  2. Average position on team would be a good stat to have. (2 votes [14.29%] - View)

    Percentage of vote: 14.29%

  3. Both of these stats would be useful. (2 votes [14.29%] - View)

    Percentage of vote: 14.29%

  4. Neither of these stats would be useful. (8 votes [57.14%] - View)

    Percentage of vote: 57.14%

Vote Guests cannot vote Hide poll

Jaspo #1 Posted Jun 26 2021 - 01:36

    First lieutenant

  • Players
  • 39550 battles
  • 639
  • [SLUGZ] SLUGZ
  • Member since:
    03-12-2015

I've not messed with modding or anything but I was looking through the API statistics variables to see what was available.

Why? Well, stats are useful to clans trying to size up an applicant's skill and playstyle, and while they can get a decent idea by looking at the info that is currently available, and the various ratings built from that info, I noticed a particular possible trend in my play that isn't reflected by the available set of statistics, but would potentially be useful information, since I hypothesize it would give a good indication of how aggressive or defensive a player is. What trend? I noticed that I seem to be doing as much or more damage in losses than in wins, and also that my average position on my team seemed to be higher in losses than in wins. My hypothesis concludes from this observation that I'm perhaps being overly cautious, or unsuccessfully aggressive, or both. However without actual statistics showing that I am in fact doing personally better in losses than in wins, I really can't conclude anything. I realize my personal statistical quirks don't necessarily justify the existence of stats tailored to track them, but nevertheless I think the following statistical additions would be potentially beneficial to everyone. TL;DR the last paragraph if you don't care for statistical theory.

 

my proposed additions:

statistics.[mode].battle_avg_xp_win                 average experience in wins

statistics.[mode].battle_avg_xp_loss                average experience in losses

statistics.[mode].battle_avg_damage_win        average damage in wins

statistics.[mode].battle_avg_damage_loss       average damage in losses

(could do likewise with the other average statistics as well; separate for wins and losses)

statistics.[mode].battle_avg_position_xp          average place on your team by experience

statistics.[mode].battle_avg_position_xp_win   average place on your team by experience in wins 

statistics.[mode].battle_avg_position_xp_loss  average place on your team by experience in losses

(could do likewise with position by other statistics as well - damage, etc; total, wins, losses)

 

 

Here's the paragraph I said you would do well to skip if you're not into this sort of thing; it's about further reasons these additions would be potentially beneficial. The game's current official and unofficial rating systems are all heavily influenced by winrate; most use player experience as a main element of their calculations, and base player experience is of course by its calculations much higher in wins than in losses, regardless of player performance. Now, I'm not saying winrate is not an important metric, it is, but we've clearly established it as a metric and I'm asking if metrics that are not biased by winrate would be equally and additionally useful. My consideration is that the more factors that can influence a given metric, the less potential it has to be a "clean" metric. To illustrate in this case, we will compare the average experience of a set of, to make the illustration clearer, players who are of identical skill level, over a number of battles. If we use WoT's average experience for this, we will see a greater variance in results than if we compare the average experience of the players in wins only, or in losses only, due to the inequity of experience awarded in wins versus losses, combined with the semi-random chance of winning versus losing. Even when comparing identical players by equitable experience (only comparing wins or only losses) there will of course also be variance due to each battle containing many other semi-random chances affecting the outcome, but the variance will be smaller than if the win/loss factor is included in addition to this, making it a "cleaner" stat without the artificial alteration to the experience value from winning vs losing. In case this still isn't clear, allow me to further illustrate. Every statistical plot has outliers. If we take the outlier with the fewest wins and the outlier with the most wins from our data set, even if they have performed individually identically in all their battles in the sample, the player with the most wins will have much better average experience than the player with the fewest, and will appear to be "better" due to circumstances largely outside their control - so the artificiality of difference between experience in a win and experience in a loss makes the overall experience rating a bit of a dirty statistic, since they get combined into a single value despite being unequal metrics. So basically what I'm after by separating win and loss statistics is a clean and unbiased personal experience rating. And as I mentioned previously, this would also show if a given player performs better on average in wins than in losses or vice versa, which would likely indicate various things about their playstyle. Now, as for the position on team statistics, these are useful because sometimes you play the battle, and sometimes the battle plays you - in a roflstomp win or a roflstomp loss, a player's individual performance, and earned experience, will often be considerably different than in a close battle. These differences are important statistical information, of course, but they are already handled by the existing metrics. Comparing average position on the team gets away from experience values altogether and simply shows how well each player performs relative to their team, independent of the outcome or nature of the battle. Once again, this produces a simple, clean, and useful statistic that isn't influenced by a bunch of other unrelated factors, and quite useful for determining player skill, particularly when viewed in conjunction with all the other statistics and ratings (average tier: 2.3...hmm...)

 

 



Engineer777 #2 Posted Jun 26 2021 - 02:27

    Captain

  • Players
  • 26500 battles
  • 1,165
  • [KEA] KEA
  • Member since:
    01-03-2013
I like how I can choose "A" and "B" and "A and B" and "neither A nor B" simultaneously in your poll.

Jaspo #3 Posted Jun 26 2021 - 03:52

    First lieutenant

  • Players
  • 39550 battles
  • 639
  • [SLUGZ] SLUGZ
  • Member since:
    03-12-2015

View PostEngineer777, on Jun 25 2021 - 19:27, said:

I like how I can choose "A" and "B" and "A and B" and "neither A nor B" simultaneously in your poll.

Fixed it, thanks.



Avalon304 #4 Posted Jun 26 2021 - 06:55

    Major

  • Players
  • 26935 battles
  • 12,828
  • [SNPAI] SNPAI
  • Member since:
    09-04-2012
Neither of of the 'in a loss' stats would be particularly useful, they wouldnt show anything other than how well you can farm in a losing situation... which isnt particularly useful for clan evaluation. The only stat which might provide some usefulness is average team position in a win, and even then with the glut of other available average stats (all of which generally correlate with win rate pretty well) it wouldnt be all the useful in the long run either.

Engineer777 #5 Posted Jun 26 2021 - 14:22

    Captain

  • Players
  • 26500 battles
  • 1,165
  • [KEA] KEA
  • Member since:
    01-03-2013

View PostAvalon304, on Jun 26 2021 - 01:55, said:

Neither of of the 'in a loss' stats would be particularly useful, they wouldnt show anything other than how well you can farm in a losing situation... which isnt particularly useful for clan evaluation. The only stat which might provide some usefulness is average team position in a win, and even then with the glut of other available average stats (all of which generally correlate with win rate pretty well) it wouldnt be all the useful in the long run either.

 

Some of us aren't interested in competitive clans, but still find stats interesting. For example, I can predict Battle Pass progression more accurately if I have stats for frequency of each ranking position among the team.



dunniteowl #6 Posted Jun 26 2021 - 16:26

    Major

  • Players
  • 38168 battles
  • 12,983
  • Member since:
    09-01-2014

Not sure I'd be interested too much in any of the choices, however, having them available MIGHT allow players to do a better 'self-assessment' of their play.

 

One thing I think WOULD be useful is a 'number of shots fired' average.  This would allow folks to determine quickly if someone is missing a lot of shots or just not taking them, relative to their overall damage.

 

Again, not sure how interested personally *I* would be in the polled information, though I would not say 'no' to it, either.  More information to analyze for the purpose of attempting to improve a player's performance, in my view, would always be welcome.

 

 

GL, HF & HSYBF!
OvO



valeman12345 #7 Posted Jun 26 2021 - 18:29

    First lieutenant

  • Players
  • 14610 battles
  • 609
  • [BCGC] BCGC
  • Member since:
    05-19-2012
Neither
There should not be any player statistics displayed in games (include win rate) and barred any mods that display such information in game and weak tanks like XVM heavily used by "high win rate users" pro players. Make them guess more and change the way they play.

Edited by valeman12345, Jun 26 2021 - 18:31.


Engineer777 #8 Posted Jun 26 2021 - 18:51

    Captain

  • Players
  • 26500 battles
  • 1,165
  • [KEA] KEA
  • Member since:
    01-03-2013

View Postdunniteowl, on Jun 26 2021 - 11:26, said:

This would allow folks to determine quickly if someone is missing a lot of shots or just not taking them, relative to their overall damage.

 

It'd be nice if overall hit rate was a weighted average based on alpha or caliber, with SPGs excluded from the calculation. SPGs could have their own stats for overall rate of direct hits. With the current system, if a player drives something like a Pz I C, it will disproportionately drop their hitrate.



dunniteowl #9 Posted Jun 26 2021 - 19:07

    Major

  • Players
  • 38168 battles
  • 12,983
  • Member since:
    09-01-2014

View PostEngineer777, on Jun 26 2021 - 11:51, said:

 

It's be nice if overall hit rate was a weighted average based on alpha or caliber, with SPGs excluded from the calculation. SPGs could have their own stats for overall rate of direct hits. With the current system, if a player drives something like a Pz I C, it will disproportionately drop their hitrate.

 

I like that addition.  I would also like to see Shells Fired by Type vs Damage by Type.  If this were possible/done, then you could determine the 'effectiveness' of a player's gun/shell choices based on number of shots fired by type and overall damage values as assigned (HE does damage in most cases even when no penetration occurs, for example, and that's true of Derp Howitzer Guns on tanks or SPG howitzers).

 

I'd love to be able to look at a whole slew of data in that regard, because it would help you determine the play style of some folks.  I mean, if it's an auto loader or machine type gun (like the small 12mm and 20mm multi-round cannons, for example) it would be nice to see an overall average of hits/pens/damage ratios depending on shell type being used vs gun type in use.

 

And @

Block Quote by valeman12345

 Neither
There should not be any player statistics displayed in games (include win rate) and barred any mods that display such information in game and weak tanks like XVM heavily used by "high win rate users" pro players. Make them guess more and change the way they play.

 

Way to completely miss the point, sir. 

 

I am pretty sure English is not your primary language, judging by the way you misuse it, so I definitely can appreciate that you might not fully understand what's being posted.  That said, this is NOT about using this as an XVM IN-GAME set of data.  It's about having access to this information (like on your Profile Page or out of game on a site, like WoTLabs or TanksGG, for example) to assess a player's overall skills and abilities, especially as regards someone deciding whether or not they might want that person to join a Clan.

 

I think most folks using XVM color codes or stats In-game are making a huge error in judgment (information overload and/or inability to properly utilize the information they ARE seeing) and are actually distracting themselves from playing their best.  Of course, some folks will be able to use that data and make some very good assessments of their team mates, their opponents and the 'level of challenge' they might be up against.  These folks are rare in the extreme if Life Experience is any indicator of how folks ignore, misuse or completely fail to understand what's right in front of them on a day-to-day basis at any rate) and the vast majority of folks using XVM are literally just adding one more layer of something they don't properly understand or use to make determination of 'game play quality of experience.'

 

That's just my own interesting point of view, however and isn't meant to be an indictment or negative judgment of my fellow gamers.  I play 100% mod free and when I play, it's a Green Team and a Red Team on the field whereby I believe MY Green Mates have to show me they are good or bad and I presume that my Red Team Opponents are ALL way better than my team and they have to show me I'm wrong before I change my mind.

 

That's just to help me remain focused on doing my best, because if I think my opponents are all good players and my team mates all need to be carried at the start, I am very hard to disappoint and very easy to be pleasantly surprised on both counts.

 

 

GL, HF & HSYBF!
OvO



Avalon304 #10 Posted Jun 26 2021 - 21:24

    Major

  • Players
  • 26935 battles
  • 12,828
  • [SNPAI] SNPAI
  • Member since:
    09-04-2012

View PostEngineer777, on Jun 26 2021 - 06:22, said:

 

Some of us aren't interested in competitive clans, but still find stats interesting. For example, I can predict Battle Pass progression more accurately if I have stats for frequency of each ranking position among the team.

 

Im glad for you... but the OP's post literally opens up with clan evaluation as the reason for adding these stats... 

 

 

 



Jaspo #11 Posted Jun 27 2021 - 02:13

    First lieutenant

  • Players
  • 39550 battles
  • 639
  • [SLUGZ] SLUGZ
  • Member since:
    03-12-2015

View PostAvalon304, on Jun 26 2021 - 14:24, said:

Im glad for you... but the OP's post literally opens up with clan evaluation as the reason for adding these stats... 

As a reason; I also mentioned personal assessment, perhaps my wording wasn't the best though...stats are interesting and useful to look at, no matter what you choose to use them for.



valeman12345 #12 Posted Jun 27 2021 - 23:09

    First lieutenant

  • Players
  • 14610 battles
  • 609
  • [BCGC] BCGC
  • Member since:
    05-19-2012

 

Public service records stats it just for show and only seem be concern by certain individuals apparently in WoT..

Yes service record stats can expand but at same time its best to really have a privacy settings to hide specific stats (win rates, tank winrates, pr rating, damage..etc) except for "Battles played". Prevent public from seeing it even prevent third party sites and game mods from seeing it. Make them guess more before game starts and less focus on toxcity trash talking whatever your a pro or avg.  This really needed.


Edited by valeman12345, Jun 27 2021 - 23:22.






Also tagged with statistics, personal rating, pr, position, place, winrate

1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users