League General Discussion Thread

gamerman01

This is the thread for all League General Discussion

The ELO rankings, where you can see many different statistics and even enter a player’s name to see complete history, are here:

Click the link
Bookmark it

ELO RANKINGS SPREADSHEET

Gargantua

Responding to your last comments in the previous thread GMan.

Quote from: Gamerman01 on January 06, 2018, 08:24:40 pm
Adam’s example was a great one - the example of a sub trying to clear a destroyer that would lead to a coup, and whether it succeeds or fails.
Probably an even better example is an Italian attempt to clear a territory for Germany that would allow Germany to sack Moscow with a very high IPC swing. Any automated attempt to quantify the luck of this battle would fail miserably. It could be a very small battle with a very small TUV swing, but actual massive game consequences.

Not to downplay your efforts - there would still be a lot of value to a TUV swing calc - would just need caveats to go with it is all we’re saying smiley

What I think people don’t always realize is that there are many types of luck in Axis and Allies. In your example, Russia not realizing or seeing the Italian can-opener (a situation that’s probably happened to all of us) is it’s own kind of luck, luck that can be entirely different from dice luck over a pro-longed period. Perhaps we could clarify this as establishing “dicing” vs “bad luck”.

What if the Italian Can-opener is 4 aircraft vs 1 infantry that was left in place by Russia. Russia could survive two rounds and get two hits, and it could be considered amazing luck in the battle, but still have their capital sacked because they didn’t see the opener.

We could also get into exponential differentials. IE, in a game say your opponent is +20 hits and you are -20 hits. That’s 20 hits worth of units you should have on the board producing for you on average, and 20 units worth of hits they shouldn’t, and this is compounded round by round for each turn those units are alive. From that point we start moving into equations with impossible to determine variables, like how many units the enemy has left in range etc. It’s impossible to calculate or quantify so we don’t.

Back to the hit-handicap ratio; it’s just there to establish what happens to you each time you go to roll each dice. How you did vs what would be the expected average. As it’s implemented it will start to change how we “perceive” luck/dicing, and the types of luck out there, and end alot of the debates about whether people were diced overall in a game or just in some few battles.

gamerman01

Good luck ending the debates :lol:
I think a battle by battle game review, by an objective experienced human player is the closest we can ever get to figuring out who was luckier and by how much. Like Adam, I’m just very leery of something that purports to settle the question of who got luckier, when any automated system will have substantial weaknesses because only a human can weight each battle by its significance to the game.

gamerman01

http://4.bp.blogspot.com/-aUu81M15YSk/UNiGLEnNRSI/AAAAAAAAAeI/8UGrWYv0bco/s1600/term.jpg

oysteilo

I think the point is to have a note at the end of the battle report indicating + x ipc in tuv or - y ipc in tuv, where x and y is the difference from the expected outcome. Now, if you sum up the + and - throughout the game and end up with say + 120 ipc it indicates that the person overall has had dice luck over average. I have tried in some situations to do this math by hand and it is tedious. Obviously this number must be used with caution and I expect most games will give very small differences. I really don’t see the harm, it is just a number that is interesting to know I think. I would welcome this.

oysteilo

If I remember correctly I did this manually in a game I felt I had very poor dice. My calculations ended up with - 60 ish ipc, meaning I lost units worth 60 ipc more than I was supposed to throughout the game. Note, this was for axis combined. I believe this is hardly important over more than 10 rounds. Luck is such a subjective term and if overall dice luck was important to win games I am surprised that the same people are top of the league year after year. Again I don’t see the harm in adding this and my prediction is, everyone is equally lucky over time. I think it will “prove” to people claiming bad luck that there luck is pretty good after all……

simon33

The other limitation of a hit differential calculator is a variation in the first round of a four round battle means some rolls don’t even happen, even though they would if the dice came out even.

Anyway, it seems that it has been included in the latest pre-release of Triple-A. Go nuts!

https://github.com/triplea-game/triplea/releases

Adam514

@oysteilo:

If I remember correctly I did this manually in a game I felt I had very poor dice. My calculations ended up with - 60 ish ipc, meaning I lost units worth 60 ipc more than I was supposed to throughout the game. Note, this was for axis combined. I believe this is hardly important over more than 10 rounds. Luck is such a subjective term and if overall dice luck was important to win games I am surprised that the same people are top of the league year after year. Again I don’t see the harm in adding this and my prediction is, everyone is equally lucky over time. I think it will “prove” to people claiming bad luck that there luck is pretty good after all……

Dice becomes important when the skill difference is small, or if you’re playing Allies in vanilla G40 :lol:.

oysteilo

@simon33:

The other limitation of a hit differential calculator is a variation in the first round of a four round battle means some rolls don’t even happen, even though they would if the dice came out even.

Anyway, it seems that it has been included in the latest pre-release of Triple-A. Go nuts!

https://github.com/triplea-game/triplea/releases

cool with the expected hit calculation, but that isnt exactely rocket science. For this to be of any real value the actual TUV difference based on the hit value difference should be calculated to I think

simon33

I like it. You can easily calculate the expected hits but this means you don’t have to do that leg work.

redrum

Hit differential is just 1 way of trying to have some sort of measure of “luck”. Its far from perfect but better than nothing IMO which was why I decided to add it. Each dice roll and battle have different strategic importance which is close to impossible to capture in any automated way.

I have considered adding a sort of TUV differential as well (actual TUV outcome of all battles - expected TUV outcome of all battles) but one problem with this is that its no longer a purely mathematical statistic as it takes into account casualty selection. Example being if you attack a stack of units that has say bombers then the battle calculator is optimized to try for best win percentage not best TUV swing where as a player may choose to save bombers longer than the battle calc chooses. This can cause somewhat subjective choices in casualty selection and the battle calc may not select the best which skews the TUV differential. This is even more drastic for maps that are more complex than Global such as Total World War as casualty selection is much less automatic and requires more thought.

gamerman01

Thanks for the post redrum!

redrum

Np. You can also thank Garg who made the request to add it :-D

All ideas and feature requests for improving TripleA are welcome and that can be done by creating a thread here: https://forums.triplea-game.org/category/42/feature-requests-ideas.

And as always, I encourage players to try out the plethora of custom TripleA maps. I have to imagine most of you league players need some variety by now :-D

Adam514

@redrum:

Hit differential is just 1 way of trying to have some sort of measure of “luck”. Its far from perfect but better than nothing IMO which was why I decided to add it. Each dice roll and battle have different strategic importance which is close to impossible to capture in any automated way.

I have considered adding a sort of TUV differential as well (actual TUV outcome of all battles - expected TUV outcome of all battles) but one problem with this is that its no longer a purely mathematical statistic as it takes into account casualty selection. Example being if you attack a stack of units that has say bombers then the battle calculator is optimized to try for best win percentage not best TUV swing where as a player may choose to save bombers longer than the battle calc chooses. This can cause somewhat subjective choices in casualty selection and the battle calc may not select the best which skews the TUV differential. This is even more drastic for maps that are more complex than Global such as Total World War as casualty selection is much less automatic and requires more thought.

Casualty selection difference is such a small difference on the scale of a match. Only a handful of battles require different casualty selection, and of those only a few changes the TUV swing significantly.

redrum

@Adam514 - For Global and other standard A&A maps, I’d probably agree with you as the OOL tends to be fairly standard and doesn’t deviate that much. There are other TripleA maps where it does due to the complexity of the unit sets. That being said since the standard A&A maps are pretty popular on TripleA, it probably would be worth adding TUV differential as well.

oysteilo

@redrum:

@Adam514 - For Global and other standard A&A maps, I’d probably agree with you as the OOL tends to be fairly standard and doesn’t deviate that much. There are other TripleA maps where it does due to the complexity of the unit sets. That being said since the standard A&A maps are pretty popular on TripleA, it probably would be worth adding TUV differential as well.

I think this is a good decision! Also if it meets massive protests from other map useres it is always possible to remove it later? It will also be interessting to see if it adds any real value or not when it comes to analyzing the games. That I do not know. I just dont see the harm in trying it out!

simon33

As a software developer, I wouldn’t be too keen on the idea of removing it later if some users don’t like it. Why do the work if you are thinking that way? I don’t see the harm to other map users. They can just skip over the info if they don’t find it useful.

Gargantua

Adam514 - asked for this game to be reviewed

This one was up and down, your statistic will say that Allies had the luck advantage, while a few key battles went in favor of the Axis:
https://www.axisandallies.org/forums/index.php?topic=40627.120

I just ran this games UK10 through my parser. Axis is -4 Allies -3 so luck, despite being a bit under par for both sides is almost equal between each other! The handicap in this game is +1 allies, exactly what you said it would be! :P

Great overall guess Adam! :)

Gargantua

If the trend continues, this game will be an epic judge of skill.

Two masters, playing their preferred nations, with near even luck when the rock drops; who knew such a thing was possible at A&A.org! :)

Adam514

@Gargantua:

Adam514 - asked for this game to be reviewed

This one was up and down, your statistic will say that Allies had the luck advantage, while a few key battles went in favor of the Axis:
https://www.axisandallies.org/forums/index.php?topic=40627.120

I just ran this games UK10 through my parser. Axis is -4 Allies -3 so luck, despite being a bit under par for both sides is almost equal between each other! The handicap in this game is +1 allies, exactly what you said it would be! :P

Great overall guess Adam! :)

Was it a guess though? :wink:

League General Discussion Thread

Featured Topics

T-shirts, Hats, and More

Suggested Topics

L26 PTV GeneralDisarray (Axis+10) vs AD Allies)

BM4 fasthard(X) vs Oskar2KX(L+25) - alternate bidding

L26 PTV axis-dominion (+6) vs pacifier

cwglee axis vs dawggg allies +40 g40 oob

L26 1940 Global OOB Dawgoneit (X) vs. Martin (L+15)

PTV 25 playoff Ghostglider vs 666 (axis +14)

L25 G40 OOB cwglee51 (L+55) v Jacob16 (X)

L24 OOB GeneralDisarray (X) vs AndrewAAGamer (L+60)

20

18.0k

40.8k

1.8m