34 replies to this topic

[Nichts]

We know that there is a direct and easily provable relationship between damage done and credits received. As you increase the amount of damage you do, you increase the amount of credits you receive. It's a very simply, straightforward relationship.

What's been harder to pin down is exactly how exp is gained. The trouble with exp is that there is no direct relationship between it and credits earned, and therefore no direct relationship between it and damage done.

When you compare games with similar damage output, you often get very different exp results. They might, in fact, resemble the following hypothetical sample of games (run against exactly the same team, to remove that variable):

Game 1: 6000 credits, 600 exp
Game 2: 6000 credits, 500 exp
Game 3: 12000 credits, 1000 exp

(In each case, the result was a victory for your team.)

In Games 1 and 2, you did the exact same amount of damage (we'll hold it to damage, since that's the biggest contributor - you'll see why in a second,) and yet walked away with different amounts of exp.

In Game 3, you did twice the amount of damage... but your exp didn't come close to doubling compared to Game 1.

How is this possible?

I've got a theory on how this could be possible, and it's based on four observations:

1. When I am in my AMX 12t in a Tier 8/9 battle, and my team wipes out the other team but I do no/little damage, I walk away with very little exp.
2. When I am in my AMX 12t in a Tier 8/9 battle, and my team wipes out the other team and I contribute significantly to that (by, for example, taking out a Ferdi by myself,) I receive a very high amount of exp.
3. When I am in my AMX 12t in a Tier 8/9 battle, and my team gets wiped out by the other team but I do no/little damage, I still walk away with very little exp.
4. When I am in my AMX 12t in a Tier 8/9 battle, and my team gets wiped out by the other team even though I contribute significantly to the damage done by my team (by, for example, taking out a Ferdi by myself,) I don't receive a lot of exp.

The exp formula theory: Exp is awarded based on the percent damage done by your team to the opposing team's total health pool, multiplied by the percent of damage you did as a an individual to the enemy team's total health pool.

The math goes something like this:

Assume the following:

- 2 credits per 1 hp of damage done
- 1 exp point per 5 hp of damage done (0.2 exp per 1 hp)
- the enemy has a 15000 hp pool for your team to damage

Game 1 - In this game, your team kills every tank on the other team (ie. 15000/15000 possible damage points dealt), and you are responsible for 3000 hp of that damage.

Credits awarded to player = 6000 credits

Team Exp Pool (TEP) = 0.2*(15000/15000)*15000 = 3000 xp
Percent of TEP awarded to player = (3000/15000)*3000 xp = 600 xp

Game 2 - In this game, your team does not kill every tank on the other team and only does 12500 of the possible 15000 damage points it could have dealt. You are responsible for 3000 hp of that damage.

Credits awarded to player = 6000 credits

Team Exp Pool (TEP) = 0.2*(12500/15000)*15000 = 2500 xp
Percent of TEP awarded to player = (3000/15000)*2500 xp = 500 xp

Game 3 - In this game, your team does not kill every tank on the other team and only does 12500 of the possible 15000 damage points it could have dealt. You are responsible for 6000 hp of that damage.

Credits awarded to player = 12000 credits

Team Exp Pool (TEP) = 0.2*(12500/15000)*15000 = 2500 xp
Percent of TEP awarded to player = (6000/15000)*2500 xp = 1000 xp

Among the things this formula would explain are the following:

- Why low tier battles give far less exp than high tier battles (the HP pool is much smaller, and battles are done between fairly evenly matched tanks which all often have similar damage output)
- Why low tier tanks make more exp in high tier battles (the proportion of damage they do is minimal, but the HP pool is so large compared to low-tier battles that it negates the lack of damage done by the low tier tank)
- Why platooning seems to have such a large impact on players. Because exp is awarded based on team performance, guaranteeing a larger portion of that team's performance by bringing in two other high-performance teammates will mean that you will have larger TEPs from which to pull exp points more consistently. The opposite should be true for low-performance platoon mates - they should bring down the TEP more often, resulting in fewer exp points which you can be awarded.
- Why exp is more difficult to come by in arty. Even in good games, half of the damage they deal is being awarded to their spotters. Thus, their share of the TEP is proportionally smaller than the damage they have actually dealt.

And I'm sure there's many more.

So, what do you guys think?

[Lert]

What is this I don't even ...

+1, my good friend. +1.

Even if it proves not to be true, just how true it rings combined with the amount of thought and work you must've put into this.

[Nichts]

Lert, on Jan 18 2012 - 18:41, said:

What is this I don't even ...

+1, my good friend. +1.

Even if it proves not to be true, just how true it rings combined with the amount of thought and work you must've put into this.

What I'm expecting to happen is for people to come up with scenarios which challenge the thinking I used, and in the process it's either proven, disproven, or refined.

I think it's a decent starting point, however.

[JohnGaltCobraCommander]

Well if you want to get some people to together and screenshot matches and try to get estimates on all the variables il help assist.

[pilotx234x]

is there a reason why Devs want to keep experience / credit formula secret?

[Lert]

I'll be willing to provide some screenshots of some of my battles if wanted.

[mattauto]

Dude nice work! Thanks for putting the time into this!  

[mattauto]

pilotx234x, on Jan 18 2012 - 18:49, said:

is there a reason why Devs want to keep experience / credit formula secret?

  WHAT?!? THE DEVS KEEP SOMETHING SECRET FROM US!?! I frankly don't believe it!  

[Moonrider]

I don't think it's quite that simple. There's a few things that will generate more XP for the same amount of damage . . .

• Critical damage to enemy tanks. This means directly damaging an enemy tank's module/crew. The difference in tier-levels between the tank is taken into account. The higher the tier-level of the tank you damage, the more experience you'll get.
• Bonus experience for restarting the capture counter by damaging an enemy tank which is capturing your base. The amount is not very large, less than 100 exp.
• Close Combat -- shooting in some radius (up to about 200m) from the enemy and being in the radius of enemy fire. You don't get much (less than 100 exp), but it still influences your experience gain.
• Survival Bonus -- while there is a bonus for a tank's survival, it is very small.
• The more your team damages the enemy, the more experience each player on your team gets. Only damage to a killed enemy is taken in account. Coefficient is not that big, but it's still experience

Granted none of these by themselves is very large, but even small amounts can add up to significant differences. Add in the non-damage XP earners, and I can easily see an unlikely scenario in which all 15 tanks on a winning team do the exact same.amount damage and all have a different amount of XP earned, depending on where they were, how they earned the damage points, and what non-damage factors came in to play.

[ForcestormX]

The wiki lists some small modifiers in addition to the above, but it also agrees that the single largest factors in xp calculations are:

Total damage by the team
Damage you yourself deal

Where damage calculated in the team calculation is ONLY for destroyed tanks (explaining the abysmal xp gains in a 14>2 roflstomp) and not counting the xp bonus for base capping. There's also the -50% for spotted shooting (which annoys me to no end, can't you just give the spotter +50% BONUS 'damage' for the xp equation...?)

Honestly, at least in general terms, I think your formula is really, really close to the true formula. It seems to fit the what I've seen and what's been begrudgingly given to the wiki team for public release. Obviously, it could be completely different (it just doesn't seem a proper equation without an integral or a log term in it ), but as a starting point this looks pretty good.

WG has liked to keep the equations hidden, but with a large number of participants to gather data, I'm sure we could extrapolate a pretty good estimation. Although in all honesty I'd like to see it, but then I'm sure they think we'd figure out how to exploit it if we knew the equation...

[Nichts]

Lert, on Jan 18 2012 - 18:51, said:

I'll be willing to provide some screenshots of some of my battles if wanted.

Maybe we could test this by setting a Team HP pool goal, and posting after game screenshots (for total team damage) as well as our final damage count for each.

Off the top of my head, I'm not entirely sure where to put the goal line so that it's reasonable and provides good, workable data.

Moonrider, on Jan 18 2012 - 18:59, said:

I don't think it's quite that simple. There's a few things that will generate more XP for the same amount of damage . . .

• Critical damage to enemy tanks. This means directly damaging an enemy tank's module/crew. The difference in tier-levels between the tank is taken into account. The higher the tier-level of the tank you damage, the more experience you'll get.
• Bonus experience for restarting the capture counter by damaging an enemy tank which is capturing your base. The amount is not very large, less than 100 exp.
• Close Combat -- shooting in some radius (up to about 200m) from the enemy and being in the radius of enemy fire. You don't get much (less than 100 exp), but it still influences your experience gain.
• Survival Bonus -- while there is a bonus for a tank's survival, it is very small.
• The more your team damages the enemy, the more experience each player on your team gets. Only damage to a killed enemy is taken in account. Coefficient is not that big, but it's still experience

Granted none of these by themselves is very large, but even small amounts can add up to significant differences. Add in the non-damage XP earners, and I can easily see an unlikely scenario in which all 15 tanks on a winning team do the exact same.amount damage and all have a different amount of XP earned, depending on where they were, how they earned the damage points, and what non-damage factors came in to play.

I realize that and the limitations it poses. However, if we could identify a trend with this method, I think it would be pretty strong proof that this is something along the lines of what is happening.

[tritium4ever]

I dunno how you can do a post on XP formulas and ignore two of the biggest factors in determining how much XP you get for a given point of damage:

1) Bonus for hitting higher tier tanks
2) Sharing 50% of XP for damaging targets spotted by other players

The first factor is a significant one, because it has a noticeable impact on XP rewards. Doing damage to a tank that's higher tier than you gives you more XP than doing the same amount of damage to a same tier tank, and both will give more XP than doing that amount of damage to a lower tier tank. So saying you did 3000 damage in and of itself tells you almost nothing about how much XP you should expect to get...did you do that 3000 damage in a Maus against tier 8 targets, or did you do 3000 damage in the French tier 7 light against a tier 9 target? The latter would give far more XP, despite the amount of damage being the same.

The second factor is also a huge contributor. If you're shooting a target that you don't have visibility on (i.e. someone else is lighting them up for you), you're sharing 50% of the XP with the spotter. Losing 50% XP is a huge deal, and it's why arty tends to get low XP rewards despite doing lots of damage and hitting higher tier targets (this can also affect long range snipers and those who use the "invisible tank" trick of staying more than 15m behind a bush so that they're not spotted when they shoot).

These two factors explain the highly variable nature of XP rewards. Without knowing exactly what targets you hit (relative to your own tier) and how the hits were made (with you spotting, or with another player spotting), it is impossible to make any real correlations between damage done and XP rewards other than to say that more damage = more XP.

[Lhos]

Moonrider, on Jan 18 2012 - 18:59, said:

• Critical damage to enemy tanks. This means directly damaging an enemy tank's module/crew. The difference in tier-levels between the tank is taken into account. The higher the tier-level of the tank you damage, the more experience you'll get.

I'm pretty sure that's BS, despite being in the wiki. Plenty of people have mentioned dealing nothing but crits (20-30 of them) and getting almost no exp. Crits are either worthless or very nearly so.

Also, I'm not sure if anyone's mentioned the 50% credits/exp win bonus you get. If you and your team do well, you're getting 1/3rd less exp if you lose than if you'd won. (Win = 150%, lose = 100%)

ForcestormX, on Jan 18 2012 - 19:04, said:

Where damage calculated in the team calculation is ONLY for destroyed tanks (explaining the abysmal xp gains in a 14>2 roflstomp) and not counting the xp bonus for base capping. There's also the -50% for spotted shooting (which annoys me to no end, can't you just give the spotter +50% BONUS damage...?)

Wait what. I really hope you wrote that improperly, because a 50% damage bonus to someone spotting an enemy tank would be horrible. People already complain about invisible tanks. TDs would be nuke launchers.

[Unhelequined]

In my experience the close combat bonus for being close to enemies and taking some (or being near their) fire seems actually significant as well. Despite what is says on the wiki, I have found that dealing approximately the same amount of damage to similar enemies will yield noticeably different XP gains between doing so unseen from cover or in the enemy's face. This is purely from anecdote and experience however. Also, more noticeable on a win than a loss, so the bonus is likely added before multipliers.

[Nichts]

tritium4ever, on Jan 18 2012 - 19:08, said:

I dunno how you can do a post on XP formulas and ignore two of the biggest factors in determining how much XP you get for a given point of damage:

1) Bonus for hitting higher tier tanks
2) Sharing 50% of XP for damaging targets spotted by other players

The first factor is a significant one, because it has a noticeable impact on XP rewards. Doing damage to a tank that's higher tier than you gives you more XP than doing the same amount of damage to a same tier tank, and both will give more XP than doing that amount of damage to a lower tier tank. So saying you did 3000 damage in and of itself tells you almost nothing about how much XP you should expect to get...did you do that 3000 damage in a Maus against tier 8 targets, or did you do 3000 damage in the French tier 7 light against a tier 9 target? The latter would give far more XP, despite the amount of damage being the same.

The second factor is also a huge contributor. If you're shooting a target that you don't have visibility on (i.e. someone else is lighting them up for you), you're sharing 50% of the XP with the spotter. Losing 50% XP is a huge deal, and it's why arty tends to get low XP rewards despite doing lots of damage and hitting higher tier targets (this can also affect long range snipers and those who use the "invisible tank" trick of staying more than 15m behind a bush so that they're not spotted when they shoot).

These two factors explain the highly variable nature of XP rewards. Without knowing exactly what targets you hit (relative to your own tier) and how the hits were made (with you spotting, or with another player spotting), it is impossible to make any real correlations between damage done and XP rewards other than to say that more damage = more XP.

A couple thoughts:

1. We know that this is correct for damage vs credits. We do not know whether or not this is true for damage vs exp, nor if it even has the same impact on exp as it does on credits.

2. Before we can understand the relationship between exp and various forms of damage, we must first understand the relationship between exp and one form of damage. Whichever form of damage this is is irrelevant to the calculation, since the only thing that will change is a multiplier. In my examples I assume that all damage is done under the same conditions.

3. However, it is very important to remove as many variables from an attempt to gather data, and all the variables you listed are variables which could skew results. That's why we'd need a very narrow definition of what type of data we are gathering, right down to type and tier of tank used, and types and tiers of tanks faced. Using data obtained from an M4 Sherman, a T29, and AMX 13 90 would probably skew the result significantly.

That would establish a baseline of damage data (ex. Tier 5 medium shooting Tier 7 heavies where the TEP is 7500 hp with no capping or defending,) from which data could be plotted and correlations (or lack thereof) drawn.

[tritium4ever]

Nichts, on Jan 18 2012 - 19:50, said:

3. However, it is very important to remove as many variables from an attempt to gather data, and all the variables you listed are variables which could skew results. That's why we'd need a very narrow definition of what type of data we are gathering, right down to type and tier of tank used, and types and tiers of tanks faced. Using data obtained from an M4 Sherman, a T29, and AMX 13 90 would probably skew the result significantly.

That would establish a baseline of damage data (ex. Tier 5 medium shooting Tier 7 heavies where the TEP is 7500 hp with no capping or defending,) from which data could be plotted and correlations (or lack thereof) drawn.

Right, you'd need to do rigid testing to isolate variables. The problem is, that's difficult to do with any repeatability in pub matches, and training battles don't award any XP (would be nice if training battles showed you the XP and credits you would've earned if it were a real battle).

I suppose one could set up a tank company battle for testing purposes, assuming that you had people willing to take losses on their stats to do the necessary testing.

[Nichts]

tritium4ever, on Jan 18 2012 - 20:04, said:

Right, you'd need to do rigid testing to isolate variables. The problem is, that's difficult to do with any repeatability in pub matches, and training battles don't award any XP (would be nice if training battles showed you the XP and credits you would've earned if it were a real battle).

I suppose one could set up a tank company battle for testing purposes, assuming that you had people willing to take losses on their stats to do the necessary testing.

Get the minimum number of people together in Absolute company battles, and form a minimum of two teams. You should have usable data pretty quickly that way, especially if everyone has a T10.

[VonKrieg]

There is one battle that I have had that threw any notion I had of the exp system out the window:

Was in my Mutant Six on Cliff (starting north), I rolled to the west to defend as I noticed no-one else was heading there. As I parked in the default defensive position I spotted an IS-3 (the first enemy spotted) and fired, My shot bounced, the IS-3 returned fire on the move and missed, then he swung his tank to the east to take cover under the cliff. As I reloaded I popped out a little further than I should of and was one hit by an S-51 (not an ammo rack so I'm assuming it was a 203mm AP round) and the IS-3 disappeared without taking a hit. All of this happened while the rest of my team was in transit and no-one else was engaged or had even spotted a target.

I earned over 1,100 exp for this match  

[C_Menz]

pilotx234x, on Jan 18 2012 - 18:49, said:

is there a reason why Devs want to keep experience / credit formula secret?

Because it really doesn't matter? Just play the game and you will be rewarded if you do a good job.

Also it would lead to people nitpicking every little detail of what they post, and then post rage threads if something is slightly off.  It would also push people further away from driving scout tanks and other niche tanks since the formula is already heavily dependent on damage done.

[bailey4500]

Math....hurts....brain....