Jump to content
shutupshake

Potential Workaround for Spotting Damage in API-based Rating System

Recommended Posts

I was thinking about how the API doesn't report spotting damage and how we might work around it. Does the API report experience received?

 

If it does, couldn't we roughly back out spotting damage since we know the formula? e.g. divide the total by 1.5 if it was a win and subtract out damage done and multiplying it by 2? I know this would be pretty rough due to the many little bonuses added for various actions, but would it make sense to try?

Link to post
Share on other sites

Premium/platoon advantage in experience is a specific reason why it's not a good idea at this point, and the Courageous Resistance feature also messes it up a bit.

Link to post
Share on other sites

So how come they haven't added it in?  Surely it can't be that difficult to just update the API...

It's a low priority, AKA doesn't improve the game directly, AKA doesn't make them more money.

Link to post
Share on other sites

I was thinking about how the API doesn't report spotting damage and how we might work around it. Does the API report experience received?

 

Yes, but not base experience. All the experience values currently present in the API have unknown amounts of premium account bonus added in, which makes them useless for performance metrics. Base experience was added internally at the same time as assisted damage.

 

As far as I can tell, the API programmers have been trying to fix the tanks/stats lost battles bug(s) for several weeks. If they ever manage to sort that out then I'll try bugging them about adding the 8.8 stats.

Link to post
Share on other sites

Yes, but not base experience. All the experience values currently present in the API have unknown amounts of premium account bonus added in, which makes them useless for performance metrics. Base experience was added internally at the same time as assisted damage.

 

Well, I guess that settles it. If it was base experience, we may have had a chance. But total experience would make it impossible. Thanks for the input. Glad to hear they are working on it.

Also, hi Planetary.

Link to post
Share on other sites

     Damage upon spotting and damage upon tracking are both in the API since about 8.6. The reason it's not used in the WN8 calculation is because:

 

It's not. Please don't make shit up.

 

The 8.8+ values are present in the dossier files and probably available to in-game mods, but they're not in the API. We don't even have reliable per-tank values for the older parameters, only games played and wins. Feel free to complain to WG if you care.

 

 

API reference for account/info:

https://na.wargaming.net/developers/api_reference/wot/account/info/

 

API reference for account/tanks:

https://na.wargaming.net/developers/api_reference/wot/account/tanks/

 

API reference for tanks/stats (bugged):

https://na.wargaming.net/developers/api_reference/wot/tanks/stats/

 

 

VBE can potentially use the new values because it's dossier-based, but that makes it much less flexible.

Link to post
Share on other sites

It's not. Please don't make shit up.

 

The 8.8+ values are present in the dossier files and probably available to in-game mods, but they're not in the API. We don't even have reliable per-tank values for the older parameters, only games played and wins. Feel free to complain to WG if you care.

 

 

API reference for account/info:

https://na.wargaming.net/developers/api_reference/wot/account/info/

 

API reference for account/tanks:

https://na.wargaming.net/developers/api_reference/wot/account/tanks/

 

API reference for tanks/stats (bugged):

https://na.wargaming.net/developers/api_reference/wot/tanks/stats/

 

 

VBE can potentially use the new values because it's dossier-based, but that makes it much less flexible.

 

Thanks for the info. I found this thread: http://forum.worldoftanks.com/index.php?/topic/366156-official-request-for-the-inclusivity-of-spotting-damageassisted-damage-in-the-api/

and showing my support.

Link to post
Share on other sites

By the way, with a tanks/stats system it isn't a problem that assist damage (or base xp) is only available since 8.8. You'd just use the number of 8.8+ battles to interpolate between the true rAssist and a combination of the other rStats. It's a one-line adjustment.

 

It wouldn't work well with an account/info + account/tanks system because you couldn't determine which tanks had gained the assistance damage and so you couldn't calculate the correct expected value. Hence fixing tanks/stats is a prerequisite.

Link to post
Share on other sites

WG API has been updated today, according to the post in the developer corner on the EU forum from Hedeone, to add damage on spotting to the vehicle stats (as well as other new features not relevant to stats calcs).

Link to post
Share on other sites

WG API has been updated today, according to the post in the developer corner on the EU forum from Hedeone, to add damage on spotting to the vehicle stats (as well as other new features not relevant to stats calcs).

Holy shit! Does it have historic data, too?

Link to post
Share on other sites

Holy shit! Does it have historic data, too?

 

Only since ~8.8, or whenever it was that they started collecting the data.

 

Currently the new parameters are only in account/info not tanks/stats, and the "battles since 8.8" parameter appears to be missing, so it's not actually useful yet.

 

New parameters in account/info:

 

avg_damage_blocked

direct_hits_received

explosion_hits

piercings_received

piercings

tanking_factor

avg_damage_assisted

avg_damage_assisted_radio

base_xp

no_damage_direct_hits_received

explosion_hits_received

avg_damage_assisted_track

 

The avg_damage_assisted* values are actually totals, not averages. Tanking factor was zero in the accounts I checked.

Link to post
Share on other sites

avg_damage_assisted is now, according to the update from Hedeone, an average value. Looking at my own API data for the EU server I see this:

 

avg_damage_assisted_track: 163.61

avg_damage_blocked: 516.99

avg_damage_assisted: 755.46

avg_damaged_assisted_radio: 591.8

 

so those do appear to be averages as per the update, rather than totals. I also have a tanking_factor value of 0.85. And the base_xp value will be useful to so that bonus XP can be accounted for and level the playing field for those, like myself, who only play premium when it's a freebie. I'm also at a disadvantage as never having fired a gold shell, so I'm somewhat stuffed on the damage related calcs anyway.

 

As RichardNixon points out, without a "number of battles since 8.8" value they're only partly useful, but at least they're averages rather than totals so can be used in a stats calculation.

 

However, it's only account totals so far, the data has not been added to the vehicle stats API results so there are no per tank values yet, but are listed in the forum post as being worked on.

Link to post
Share on other sites

Redo the validation/eureqa step with overall assist variable.

You have avg spots per tanks and there is a correlation between spots and spotting dmg. Spotting dmg will probably replace spots when per vehicle spotting dmg is available but in meantime one could impute per vehicle spotting dmg by using overall avg dmg and per tank spots.

Link to post
Share on other sites

Redo the validation/eureqa step with overall assist variable.

You have avg spots per tanks and there is a correlation between spots and spotting dmg. Spotting dmg will probably replace spots when per vehicle spotting dmg is available but in meantime one could impute per vehicle spotting dmg by using overall avg dmg and per tank spots.

 

I don't see the point. If we only have the proxies, then the most accurate multiple regression will use the proxies as direct inputs. If you substitute the proxies afterwards then you're just adding noise.

 

If WG eventually add assist damage in usable form (not hopeful, given that they refuse to add bc88), then the formula will need additional steps anyway.

 

 

BTW, the current formula is pretty bad anyway, due to lack of platoon filtering and population weighting, and bad zero points.

Link to post
Share on other sites

I don't see the point. If we only have the proxies, then the most accurate multiple regression will use the proxies as direct inputs. If you substitute the proxies afterwards then you're just adding noise.

 

If WG eventually add assist damage in usable form (not hopeful, given that they refuse to add bc88), then the formula will need additional steps anyway.

 

 

BTW, the current formula is pretty bad anyway, due to lack of platoon filtering and population weighting, and bad zero points.

 

In the real research world when we get new data we check to see if the model is still the best. When assist spotting damage is added, it is going to drastically change the estimates.

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...