Common Stats Project

[Case]

Let me suggest to include kill ratio (AA and AG) for individual pilots as well. :)

There is the A2A kill ratio on the A2A (the index) page, but I can (and should) add the same for the A2G page.

 

There is an inherent problem about what to do with A2A kills in bombers and A2G kills in fighters. Right now the A2A kills and losses in bombers are counted on the A2A page, whereas all the A2G stuff if counted on the A2G page. An option would be to really separate the A2A and A2G to have an A2A and A2G page for fighters, and an A2A and A2G page for bombers. Do you have any suggestions on this?

 

 

How often is it updated?

Right now it is updated when the 104th and 3Sqn server admins send me the latest files. I get those every few days and then download the 4c and 51st logs as well.

 

The 104th has not decided if they will join the experiment.

[Pilotasso]

Ok, got it. I needed to sort the navigation out to find kill ratios. S!

[Moa]

There is the A2A kill ratio on the A2A (the index) page, but I can (and should) add the same for the A2G page.

 

There is an inherent problem about what to do with A2A kills in bombers and A2G kills in fighters. Right now the A2A kills and losses in bombers are counted on the A2A page, whereas all the A2G stuff if counted on the A2G page. An option would be to really separate the A2A and A2G to have an A2A and A2G page for fighters, and an A2A and A2G page for bombers. Do you have any suggestions on this?

 

 

Right now it is updated when the 104th and 3Sqn server admins send me the latest files. I get those every few days and then download the 4c and 51st logs as well.

 

The 104th has not decided if they will join the experiment.

 

I found it worthwhile thinking in terms of air-to-air and air-to-surface (which is A2G and naval) for fighters/multirole and attack aircraft (the USAF does not call aircraft 'bombers' unless they're Big Iron like B-1, B-2, B-52).

[Grimes]

I think the current solution for categorizing A2A and A2G works well enough. I suppose it wouldn't be a terrible idea to separate the stats specifically by aircraft. By placing Migs, Flankers, F15s into the A2A category and A10, Su25A/T, and KA-50 into a strikers page. By doing so we would create a ton of flexibility for the scores. Possibly give more points to an A2G aircraft for achieving A2A kills, since its not what it is designed for and vice versa. On the flipside you do see alot of pilots flying Flankers and taking out ships, usually by taking advantage of game limitations (10m flying) and/or mission design limitations (ship AI weaponry and no fighters protecting carriers).

 

Stats are always a catch-22. You want it to be fair for all, but the player who has the most time in game will always have the best stats. You could make it a points per hour sort of deal then the more time you have invested the harder it is to improve, but you must be consistent in your scores otherwise the points go down easily.

 

Naturally that leaves out the biggest question of them all. Whats more important to players, stats or the mission?

 

 

ps, Case, I sent you a PM on our site for the latest logs.

[Moa]

I think the current solution for categorizing A2A and A2G works well enough. I suppose it wouldn't be a terrible idea to separate the stats specifically by aircraft. By placing Migs, Flankers, F15s into the A2A category and A10, Su25A/T, and KA-50 into a strikers page. By doing so we would create a ton of flexibility for the scores. Possibly give more points to an A2G aircraft for achieving A2A kills, since its not what it is designed for and vice versa. On the flipside you do see alot of pilots flying Flankers and taking out ships, usually by taking advantage of game limitations (10m flying) and/or mission design limitations (ship AI weaponry and no fighters protecting carriers).

 

Stats are always a catch-22. You want it to be fair for all, but the player who has the most time in game will always have the best stats. You could make it a points per hour sort of deal then the more time you have invested the harder it is to improve, but you must be consistent in your scores otherwise the points go down easily.

 

Naturally that leaves out the biggest question of them all. Whats more important to players, stats or the mission?

 

 

ps, Case, I sent you a PM on our site for the latest logs.

 

Well, I use 'ace points' to represent that for a2a. Killing anything flying with an attack aircraft is an ace point. Using a heater is an ace point. Using a cannon or R-60 gets you two ace points. Being in a fighter and getting killed by an attack aircraft gets you the comment "Owned". Self-kills get a "Duh!" remark. Teamkills are commented on and marked in red. The point is to reward excellence, highlight stupidity, and to liven up what would otherwise be an endless series of indistinguishable kills. Would be cool if the Common Stats did something similar.

 

ps. All your castles are in our possession. I suppose that pretty close to the notorious "your all your base are belong to us" - including the language marmalization. Is 3 Sqn considering branching into "Zero Wing" :)?

[RvEYoda]

Some people are landing on ROADS after Ship kills.........

 

This could be solved by specifying that they have to land within a certain radius from the center

of the airfield for example. Whoever writes the stats script could check this.

[Grimes]

Well, I use 'ace points' to represent that for a2a. Killing anything flying with an attack aircraft is an ace point. Using a heater is an ace point. Using a cannon or R-60 gets you two ace points. Being in a fighter and getting killed by an attack aircraft gets you the comment "Owned". Self-kills get a "Duh!" remark. Teamkills are commented on and marked in red. The point is to reward excellence, highlight stupidity, and to liven up what would otherwise be an endless series of indistinguishable kills. Would be cool if the Common Stats did something similar.

 

ps. All your castles are in our possession. I suppose that pretty close to the notorious "your all your base are belong to us" - including the language marmalization. Is 3 Sqn considering branching into "Zero Wing" :)?

 

Kudos on the Latin translation :thumbup: (i had stumbled on it)

 

But with regard to stats, how exactly do "Ace points" factor into the stats? I'd hate to say it, but specific weapons shouldn't have a weighted value relative to other weapons. A kill is a kill, it doesn't matter how one is achieved. Giving extra points to a weapon might just cause the players who really care about their stats to add R-60s to their payload, just in case they get in a close fight and want the extra point. Extra points for a Su-25T killing a fighter... why not, usually if that happens the fighter either screwed up or the striker was in the right place at the right time AND probably has some teammates in the area.

 

I was thinkin a while back of the concept of achievements in Lock-On... Now I realize some of you guys hate that sort of concept and view it as a childish along with most other mainstays of Xbox and Playstation. However, they do present a use with regard to statistics and an extra dimension to the "liven up game" as Moa put it.

 

Achievements are at their core are simple rules that use gameplay/stat information to award a player. Within LockOn they could be as simple as killing x number of specific units or having 20 hours in a specific aircraft. More complex achievements are possible that require a very specific sequence of events over multiple games. Some could be quite serious while others are more light hearted in purpose. From a pure statistics and group observation perspective, achievements are a quick and easy way to group the player base by distinguishable factors. Finding out that 32% of recorded players lack a pretty basic and simple to get achievement requires investigation into why so many players havn't got it yet.

 

Just an idea...

 

btw Yoda, if Flags get written to a log we can use triggers to decide if a player has landed at a safe base, then use the different logs and compare the information to score appropriately. One of the many uses to such a log. Bug the devs to include it :D

[RvEYoda]

There is another way.

Just do a LoGetWorldObjects serverside and check the position on the player that just landed.

This function can be enabled serverside but disabled for clients ( prevents cheating )

 

Concerning more details from the triggers exported, cant promise anything. I think there are some things

being worked on, but I'm not one of the best one to ask about the ME.

Edited February 24, 2010 by =RvE=Yoda

[Moa]

Kudos on the Latin translation :thumbup: (i had stumbled on it)

 

But with regard to stats, how exactly do "Ace points" factor into the stats? I'd hate to say it, but specific weapons shouldn't have a weighted value relative to other weapons. A kill is a kill, it doesn't matter how one is achieved. Giving extra points to a weapon might just cause the players who really care about their stats to add R-60s to their payload, just in case they get in a close fight and want the extra point. Extra points for a Su-25T killing a fighter... why not, usually if that happens the fighter either screwed up or the striker was in the right place at the right time AND probably has some teammates in the area.

 

I was thinkin a while back of the concept of achievements in Lock-On... Now I realize some of you guys hate that sort of concept and view it as a childish along with most other mainstays of Xbox and Playstation. However, they do present a use with regard to statistics and an extra dimension to the "liven up game" as Moa put it.

 

Achievements are at their core are simple rules that use gameplay/stat information to award a player. Within LockOn they could be as simple as killing x number of specific units or having 20 hours in a specific aircraft. More complex achievements are possible that require a very specific sequence of events over multiple games. Some could be quite serious while others are more light hearted in purpose. From a pure statistics and group observation perspective, achievements are a quick and easy way to group the player base by distinguishable factors. Finding out that 32% of recorded players lack a pretty basic and simple to get achievement requires investigation into why so many players havn't got it yet.

 

Just an idea...

 

btw Yoda, if Flags get written to a log we can use triggers to decide if a player has landed at a safe base, then use the different logs and compare the information to score appropriately. One of the many uses to such a log. Bug the devs to include it :D

 

Yes. The excellent site for Battlefield 2, http://www.bf2s.com was my inspiration.

 

Yoda, that is an excellent idea. I have a real-time TacView data parser but I have found it to be erroneous sometimes so never integrated it into my stats system. Are the direct export functions reliable or are they sometimes wrong (is the fauklt due to the exported data or TacView's interpretation of it)?

[Grimes]

Ribbons, Badges, and Medals would be a more fitting term than achievements.

[RvEYoda]

Yes. The excellent site for Battlefield 2, http://www.bf2s.com was my inspiration.

 

Yoda, that is an excellent idea. I have a real-time TacView data parser but I have found it to be erroneous sometimes so never integrated it into my stats system. Are the direct export functions reliable or are they sometimes wrong (is the fauklt due to the exported data or TacView's interpretation of it)?

 

I don't think the global lua exports will provide the wrong data, unless the game plays back

what happened incorrectly ( as seen in corrupt game .trk files ).

If the game shows a player at LAT/LONG position A, then the lua export will also show A,

at least this has happened in all my tests... ( and also what my radar scripts have always used )

 

Maybe tacview makes some assumptions that it stores in the tacview files, causing some

of your data to be incorrect?

More precisely, what errors are you seeing?

 

Personally the scoring system that would interest me the most would be judging the performance

of objective completions. For example some planes are assigned to escort, some sweep, some

CAS etc. If they follow certain "patterns" or hit certain objectives, then they should get some form

of achievement points ( ace points or whatever ). Then the total score should be based on

( ace points / time ) * F(points Available). F here since the mission can be empty, no players present

etc, there must be something available to kill.

Edited February 24, 2010 by =RvE=Yoda

[Pilotasso]

I don't think the global lua exports will provide the wrong data, unless the game plays back

what happened incorrectly ( as seen in corrupt game .trk files ).

If the game shows a player at LAT/LONG position A, then the lua export will also show A,

at least this has happened in all my tests... ( and also what my radar scripts have always used )

 

Maybe tacview makes some assumptions that it stores in the tacview files, causing some

of your data to be incorrect?

More precisely, what errors are you seeing?

 

Personally the scoring system that would interest me the most would be judging the performance

of objective completions. For example some planes are assigned to escort, some sweep, some

CAS etc. If they follow certain "patterns" or hit certain objectives, then they should get some form

of achievement points ( ace points or whatever ). Then the total score should be based on

( ace points / time ) * F(points Available). F here since the mission can be empty, no players present

etc, there must be something available to kill.

 

Wrong data can be obtained after playing back a track to record it on tacview. Not when the data is extrated in real time. There is a difference.

 

When tracks are played back I have watched numerous inconsitencies like pople getting killed when they survived and vice versa, reverting human players into AI control causing a cascade effect of inconsistencies, such as increasing numbero of AI players, and cloned players (one human and another or multiple AI clones).

 

Ths doenst happen when data is extracted in real time beacause it wont be taken by the AI if something anomalous hapens, warping, someone using wrong loadouts, whatever etc etc etc.

Edited February 24, 2010 by Pilotasso

[Raimir]

Wrong data can be obtained after playing back a track to record it on tacview. Not when the data is extrated in real time. There is a difference.

 

When tracks are played back I have watched numerous inconsitencies like pople getting killed when they survived and vice versa, reverting human players into AI control causing a cascade effect of inconsistencies, such as increasing numbero of AI players, and cloned players (one human and another or multiple AI clones).

 

Ths doenst happen when data is extracted in real time beacause it wont be taken by the AI if something anomalous hapens, warping, someone using wrong loadouts, whatever etc etc etc.

 

I can confirm inconsistencies in a played track. We had one were the Su-25T was shot down in the real-time mission by a missle, but in a track the missile doesn't hit the plane and fly little bit behind him and not cause any damage. But anyway plane falls down, but the hit is not in the log.

 

And i can add one more when we start server and clients connect before i open a mission and run it, than in the mp_log.txt is wrong start time of the client (always 00:00:00 not a mission start time) than the fly time stats is wrong.

[Case]

Inconsistencies in the replaying of a track is not a problem for a stats page. Moa mentioned Tacview having some problems but this may be due to the fact that it has to guess some events. As Yoda mentioned, these problems may be solved by a dedicated script interpreting the exported events instead of the events recorded in Tacview.

 

I think, however, that for now this is beyond my scope for the stats project.

[RvEYoda]

Wrong data can be obtained after playing back a track to record it on tacview. Not when the data is extrated in real time. There is a difference.

 

When tracks are played back I have watched numerous inconsitencies like pople getting killed when they survived and vice versa, reverting human players into AI control causing a cascade effect of inconsistencies, such as increasing numbero of AI players, and cloned players (one human and another or multiple AI clones).

 

Ths doenst happen when data is extracted in real time beacause it wont be taken by the AI if something anomalous hapens, warping, someone using wrong loadouts, whatever etc etc etc.

 

That's what I said, or at least tried to say.

I'm asking if there are any more errors on top of this, inherent to the lua export

used by tacview, not the game's track recording ( which does it wrong )

[Case]

The only thing flawed with the system is the points. To see what I'm talking about look here http://www.51st.org/stats/overall.php?pid=09all

Can you elaborate your point?

 

What I see is that the top three pilots each have approximately the same flight hours and each has the approximately the same points. Only one of them scored the points completely from A2A kills, where as the other two have half the points from A2A kills and half from A2G kills. They also have half the number of A2A kills compared to the pilot with only A2A kills. To me this shows that we made the right choice in weighing an A2A kill to A2G kills.

 

However, if you have suggestions to improve it, please let us know!

[Boberro]

Dunno if it was posted but for me personally the best system is just kd ratio, without any funny points system ;]

[Case]

Dunno if it was posted but for me personally the best system is just kd ratio, without any funny points system ;]

We have something like that on the 51st stats page: http://www.51st.org/stats/aces.php?pid=09all.

 

I will try to implement it on the common stats page as well.

[GGTharos]

Case, if you could provide different sorting options it would be great ... also a 'points per hour' ratio would be interesting from a statistical point of view, but it would definitely have to be category specific (a2a/a2g) rather than overall.

[Fudd]

There is another way.

Just do a LoGetWorldObjects serverside and check the position on the player that just landed.

This function can be enabled serverside but disabled for clients ( prevents cheating )

 

Concerning more details from the triggers exported, cant promise anything. I think there are some things

being worked on, but I'm not one of the best one to ask about the ME.

 

This is easily acheived via the ME with the trigger system already in place for DCS.

 

Continuous Condition / Unit Altitude Lower Than and Unit within Zone / Unit Explode

 

Set a zone to encompass all areas outside the airfield (focusing on where you think offenders will be) and then set the altitude to your choosing (ie 100m, 10m, etc)

 

You can keep people from 10m flying and enforce a hard deck...

[X-man]

A pilot that only flies strikers would not even get close to making that list even if he had 5,000 kills. But like Grimes said stats are a catch-22, there is always going to be some flaw/mistake/etc...How would it be rated if the military assigned points to targets? maybe it would be just like how the 51st has it set-up...I don't know.

 

Yes he would. The point of the overall stats are to have A2G pilots compete with A2A pilots, where a typical A2G sortie will give the same as a typical A2A sortie. Right now it looks quite balanced to me. Look at this example, which is quite representative for the 51st Stats:

 

• <51>Teknetinium has 112 Hrs and 6270p on A2A alone
  
• 3Sqn_Cali has 109 Hrs and 10374p on A2G alone
  

 

It seems to me that A2G pilots does NOT have an tougher time, and even might have an easier time racking up points per hour... Ofc, all this is dependent that you finish your sorties and not get crashed, as this will make your point go down.

Edited February 24, 2010 by X-man

[RvEYoda]

This is easily acheived via the ME with the trigger system already in place for DCS.

 

Continuous Condition / Unit Altitude Lower Than and Unit within Zone / Unit Explode

 

Set a zone to encompass all areas outside the airfield (focusing on where you think offenders will be) and then set the altitude to your choosing (ie 100m, 10m, etc)

 

You can keep people from 10m flying and enforce a hard deck...

 

I dont know if I would want to prevent people from flying low.

How would this be exported btw ?

[Frostie]

One pilot has over 2200 a2g and another has over 1700 a2g

 

The top 3 look like this, with some only a2g guys

 

a2a a2g points

#1 322/2 2250/2 17,298 <----both

#2 649/14 3 15,195 <----mainly a2a

#3 272/4 1719/6 14,275 <----both

#13 79/2 831/2 4,560 <----mainly a2g

#15 2 862/1 3,668 <----a2g

#23 2/2 1,663/14 3,072 <----a2g

 

 

A pilot that only flies strikers would not even get close to making that list even if he had 5,000 kills. But like Grimes said stats are a catch-22, there is always going to be some flaw/mistake/etc...How would it be rated if the military assigned points to targets? maybe it would be just like how the 51st has it set-up...I don't know.

:doh:

By forgetting to look over flight hours, crashes etc. you've gone and forgotten some serious reasons as to why pilots are in their respectable positions, im more than pretty sure that 2nd place guy deserves to be where he is.

Not to mention that ground pounding in an empty server is like shooting fish in a barrel.

 

......a2a ......a2g.... points

#1 322/2... 2250/2 ..17,298 <----both...............267 hours + 40 crashes

#2 649/14.... 3 .......15,195 <----mainly a2a.......223 hours + 13 crashes

#3 272/4... 1719/6 ..14,275 <----both...............231 hours + 50 crashes

..

#13 79/2... 831/2 .....4,560 <----mainly a2g........81 hours +44 crashes

..

#15 2 .......862/1..... 3,668 <----a2g................109 hours + 43 crashes

..

 

Im sure if the 13th place guy put in 200 +hours he would be well into the 5 figure score.

Edited February 24, 2010 by Frostie

[Case]

But like Grimes said stats are a catch-22, there is always going to be some flaw/mistake/etc...

Indeed, it is very much that situation. In the 51st we aimed to design a scoring system that would allow the skills of fighters to be compared to that of strikers. You could indeed argue that A2A kills are overrated compared to A2G kills, given that no true 'striker only pilot' made it into the top 10 of the overall stats. However, it should be reminded that it is indeed like shooting fish in a barrel, as there is no human element working against you. in fact, the strikers can even learn the missions in rotation and improve their tactics to minimize their risk and maximize their kills. With that in mind I would say the scoring system is very reasonable.

[Case]

Case, if you could provide different sorting options it would be great ...

This has been on my 'todo' list for quite some time, but I have not gotten around to implement it yet. Anyone with basic MySQL and PHP knowledge is welcome to try and implement it. I can prove the scripts and a dump of the database.