Global score

[JC (naval supremacist)]

I had a talk with elexis about it . Do a sum of eco and military is the best option, knowing that you can hit high simply by making

hundreds and thousands of farms ?

If one divide the military score by the eco score, one have something which represent approximately the efficiency in using of ressources . Right ?

I propose to define a score as the division of the

(total amount of health killed (units + buildings) - total amount of health lost (units + buildings) )

by

the total amount of ressources invested (in units and buildings)

This score would reflect strictely the results and no more the ressources engaged to win. And so, producing hundreds of units - making them walking , idling, working ressources or even building 100 towers will have no effect on this score.

Edited July 8, 2015 by JC (naval supremacist)

[zzippy]

Why care about the score at all?

The only relevant thing is who pushes the resign button ...

Many wars were lost despite the better eco/military score in history.

[JC (naval supremacist)]

Many (but still the minority) wars were lost despite the better eco/military score in history.

you are right .. and its precisely the matter of this topic

-------------

But agree, its like playing Mario Bross, just finish the game.... but this score will be probably correlated with the final result (at least, try to, the aim of the score is to show how good you are in achieving the goal)

Moreover, after a team match, it can show which players play best even among the team which lost. You can even report this efficiency score for a player on the average efficiency score of the whole team -->

the score of a good player will be even more good if his allies were bad and did nothing.

----------

By now, i notice that my proposition is not perfect as a player can send his inner soldier units to go kill some isolated units. If no building or units are build (no invest) , as a result, the ratio will be

a division by 0 ... --> so, to avoid this, the invest denominator should add the cost of the CC + the 9 starting units .

Also, if a player resign, all the remaining health (health of buildings and units) should be charged as a lost in the score to avoid that tower-defensive players artificialy grow the numerator of the score

by resign just before ennemy start to destroy his buildings

So,

If a active-defensive player resist and dont resign, he will score high by killing lot of ennemies attacks with no lost. His high score will prove that he has been usefull for the team.

If a player thinks he can only skirmish and rush with some horses with no invest in growing, he will be quicky crushed and his kills will be lower than the lost

If a player is totally passive and wait behind his towers that all allies die, he will be finally sieged by all ennemies at the end with catapults, rams and upgraded champs ->

he will succeed to kill no more than 3 or 5 champs with his castle, but will lost all the health of his buildings and units in garnison once as his buildings will crawl one after another.

----------

i think the subject of the score can be very interessting and deserve the remarks and suggestion of all contributors

Edited July 8, 2015 by JC (naval supremacist)

[JC (naval supremacist)]

Nobody wants to join the debate ?

[FeXoR]

IMO "scores" are always arbitrary because not well defined or at least arbitrarily defined.

For the end of game summary screens I'd stick to things that actually are clear in-game properties and no arbitrarily defined "scores".

I agree with your calculations of the "efficiency score" ( though it's just the military efficiency score).

However, it doesn't help to be highly efficient in using the resources you have if your opponent is half as efficient as you are (in your sore) but managed to gather 3 times the resources (has 3 times the "resource effectiveness" - not efficiency, or you compare it to the time the players had which would be the same for both players). In that case he'd still win. And that's the point of the game.

In the end you can simply see which player had the over all better effectiveness by the fact he won the game.

[JC (naval supremacist)]

Sorry, i think the name of my topic is bad. This cannot be a global score but , indeed, an military efficiency score. You are right.

The kill/death ratio is a definied score and not a clear in-game proprety . Despite this, everyone refer direct to that score to make kindly fun of each other ( especially me on naval maps )

That said, my proposition is just to redefine a ratio (considering the total health of the units/buildings killed or lost) but also adding an invest dimension in the score.

( total amount of health killed (units + defensive buildings) - total amount of health lost (units + defensive buildings) )

/

the total amount of ressources invested (units + defensive buildings + military upgrades)

As you can see, in this score i dont consider what your units did during their time (gathering useless extra food or wood, wasting time idling somewhere on the map or in a CC) but only the way

the player made the best choices in his decisions to weaken his ennemies. For exemple, the size of his armies in an attack operation,

the choice of units developped in each attack, the decision to build a castle rather than attack nearby wood-working females

(counter-exemple : a fool who send expensive low-range cav to kill a full-garnision-castle will have poor score ; build in a distant useless tower will grow the invest denominator of the ratio but with no kill results!)

This score will not judge who won and who lost . For sure not. But will be usefull in a multiplayer team-match to show each player if he was right in his military decisions.

This score can be used also by 0AD dev team as a usefull statistic to rebalance each version of the game. For exemple, to detect if a too much phase3 rush leads to victory or not. This score should tend to

be correlated with the final result of the game : victory or defeat.

____________________

That said, in a scenario of pure cooperation in a team where PLAYER A decide to focus on military and PLAYER B on a pure strong eco (100% females and high pop expension), its natural that PLAYER B should have a

retribution in a new score for the risk taken (focus on ECO with no effective defence). For sure, this new score has to consider the fact that PLAYER B effectively support PLAYER A !

So your concept of "resource effectiveness" has to be developped in a new score showing an ECO efficiency ...

exemples: decision to invest in fruit picking, corrals , fishing, females on fields rather than slow soldiers ; decision to build a new storage near wood for an enough working time ; decision to focus on long-term grow in peace zone

by gathering stone to build barracks , market exchange efficiency ...

i propose something like

( total amount of ressources gathered (workers + traders + killed ennemy traders) + total amount of ressources given to others (allies + market exchange) - total ressources given by allies )

/

the total amount of ressources invested (units + armless buildings + eco upgrades + traders + market exchange )

Unfortunately, gathering like a silly-sausage only useless food will be the same than gathering equally each usefull ressources ; like, giving unnecessary tons of food to allies will grow the score in the same way than giving precious metal ;

like exchange expensive metal for cheap food on the market will be always efficient on the score ...

--> If one consider that ressources worth something only if you spend it, one can consider the score should be something like :

( total amount of ressources spend (in everthing) + total amount of ressources given to and spend by allies + market exchange(received) - total ressources given by allies and spended by you )

/

( the total amount of ressources invested (workers + armless buildings + eco upgrades + traders + market exchange(given) )

What is the dynamic of this score ?

--> This score will retribute a player who are not scared in growth investing and be efficient in his eco choice: In short term, he will lower his score by investing but will be able, after a time, benefit the results and will be in position to spend more ressources. These ressources have to be spend to affect the numerator : either the player decide to invest in more growth , either to give it to allies .. the first choice is more effective on the score in long-term, but the second can be far more saviour for the team ! For sure, these ressources have to be used by the ally .. and so, only valuable ressources (i.e not tons of food) would affect the numerator. The fact that the Ally has to spend the given ressources will avoid the scenario where a little smart player give all his ressources at the end at a game to his almost dead allies and grow artificially his score.

The best ECO-score player will be the most rocket efficient pop-grower and will the one who immediately after pop max gives usefull ressources to players who are able to spend those precious ressources in the right time. For sure, bidding only on ECO is risky and a rush can be desastrours on this eco-score on long term.

__________________

So, we have 2 propositions on 2 efficiency scores : a military one and a eco one. The victorious may have 2 good score.

[Karamel]

Just my few cents. I don't have the mind to dig into the math right now (it's late by there).

I use the score to see where I did wrong. Trying to understand a bit what made me win (or more often lose). I don't really use the global score, but more the details. If a resource is far below (harvested or spent), I must focus more on building and less on fighting/raiding (or more efficiently). If I had half the number of military units, I should build a stronger army and less females. If trading is far better I know I have underlooked it etc. If I had more military and lost, I should ask how to play

I don't really care about how much, but it's just a tool for me to try to understand what append. Effeciency could be fun to subcompare at a glance, but I don't think I would rely upon it (overwhelming with stronger eco even if not optimised is still perfectly viable).

[mickmca]

I'm still trying to figure out what some of the score elements refer to. Recently I resigned on an early game and my 'feminization' score was 100%. Now I know that at least half of my population was male, so whaaa? I assume that the "vegetarian" score is a sign of some positive thing, because everybody who outscores me has a significantly higher "vegetarian" score. And yet as I examine the screen, it doesn't appear that the "high veg" folks have that many more farms than I do. The "barter" scores don't seem to have anything to do with markets or traders. I always run at least 3 traders and my barter score is always 0. In a recent game, I had 5 of my 10 siege rams convert to the enemy control; I presume that makes them "lost" rather than killed; but when "lost" and "killed" are generally within a few percent of the same number, that's not very clear. Can someone point me to a discussion of the score data?

[sanderd17]

• That femenization score is weird. Don't know what happened there
• Vegetarian score is indeed the amount of food from fields and berries vs the food from animals or trade. Perhaps your opponents have their fields earlier, or research some techs to gather quicker?
• There's also a difference between barter and trade. Trade is what you do with traders between markets, and trade generates resources. Barter on the other hand, is selling one resource for another in the market (the left pane when you select the market). Bartering costs some resources, as you won't get 100% return normally (only when there's a certain resource in abundance, but then you don't need to get it via bartering either). So the bartering score is the return you got (if you bartered 100 wood for 80 metal, you will have a barter efficiency of 80%).
• Capturing of buildings and siege doesn't yet appear in the stats. But "lost" is what you lost, "killed" is what units of the enemy you killed.

[causative]

It would be more useful to have "resource value destroyed" stats instead of "number of units/buildings destroyed."  "Resource value destroyed" would multiply units/buildings destroyed by the cost of those units/buildings.  That way, killing 50 women would not count the same as killing 50 champs.  Starcraft II uses this system.  (To get a little fancier - adjust resource weights according to how valuable that resource is, so that killing units that cost metal gains you relatively more score).

Kill/death ratio could then be replaced by a more useful "combat efficiency" score - the ratio of the resources of the enemy that you destroyed, to the resources of your own that were destroyed by the enemy.  This would help you figure out whether a certain strategy is cost effective or not.

Total score should then be:

1. the total resources you spent
2. plus the value of enemy units or buildings you destroyed
3. plus twice the value of units or buildings captured from the enemy (capturing counts twice because it both takes from the enemy and gives to yourself)
4. minus the value of your own units/buildings that you deleted.  (If you resign, that counts as if you deleted everything you had remaining)

To see the logic of this, consider a team game such as 4v4.  The "strength" of either team at any point in the game can be approximated by the total value of all the units and buildings owned by that team, which will go up and down over the course of the game.  The four proposed factors for total score count your total contributions to the difference of (your team strength) - (enemy team strength) over the course of the game.

Those 4 factors for total score have the nice property that the winning team is guaranteed to have a greater total score than the losing team - indeed, even if the game ends prematurely, the team with a higher score ended with a higher total value of units and buildings.

Also - in all 4 factors, sheep should not be counted.  (If you did want to count them, producing a sheep would fall into factor #1, and butchering it would fall into factor #4, canceling itself out).

[jonbaer]

I just did a patch to remove the sheep from the count but in doing so pretty much questioned every other score along the way, pretty much finding this topic on the forums at least confirms some of my thinking :-)  What I think should be pulled apart is what constitutes a simple "stat" (fem%, etc)  vs. what is used for a "score".  There are a few other items I would like to see, I think AoE-like chart summaries, timelines, etc.  I haven't quite figured out how this would all be possible but one thing I thought of is if nearly all score counting can come from just the replay files themselves?  (is that possible?)

[Imarok]

There is some information in the corresponding ticket: #3403

[JC (naval supremacist)]

Hi guys,

 

Just coming back with the subject for 0ad23.  

 

Introduction :   Do you notice in a team game (for exemple) that some players are very good to play with absolute no considerations for their alliy (allies) and play for them selves , resulting in behaviours such as  : building defences just 10 meters away from the battle you are fighting in,  the very same battle you cross the whole map to get in, in order to save him. As a result, he often let you fight 2 vs 1 and total ignores the dynamic of numbers in a fight (lanchester law) and the fact that those towers will be totaly useless if you fail to defend him.

Results : While you exhaust your whole eco and lose your whole army, that same ally can finish the 2 weakened ennemies, kill all their workers, destroy buildings with rams  .. and.. get the highest military at the end of the game ; and all the glory.

Observation : the current score system doesnt take into account the time ; in other words : 100 kills at min 5  = 100 kills at min 20.

Simple idea :  The graph tab shows the evolution of the score.  Why dont we take the "area under the curve" (AUC) (the integral in math)  as new score  (a new tab in the stats)?  Early success in the game will have an score impact all along the game.  This should be easy to implement as it could be computed at each event (+1 kill) just like it is for the graph.

Derivated idea : If you divide the military AUC of a player  by the  AUC of the whole team / number of players in the team , the you can have a good indicator of performance of that player as his military activity is compared to the average AUC of the team  (on the military aspect , but could be done for the eco as well)