Tactical Command
http://www.tacticalwargames.net/taccmd/

Battlestats
http://www.tacticalwargames.net/taccmd/viewtopic.php?f=43&t=4182		 Page 1 of 1
Author:  CyberShadow [ Mon Feb 20, 2006 2:07 am ]
Post subject:  Battlestats
This thread came up on the Tau board, and I am moving it here to open it to a wider audience.

dysartes:
Last I heard, the intention was for as many people as possible to report into B.Stats, especially with results from tournaments or other such events. Assuming this is the case, and I have no reason to doubt it,


Honda:
The point being that because Battlestats only reports (to my knowledge, I could be stepping out on a limb) win/loss and army (is there anything else?), that without controlled scenarios and conditions to drive the testing, it's very difficult to draw solid conclusions.

Extreme examples:

Game 1) SM vs Tau, Tau player takes 5 AX-1-0, the SM player doesn't take any AA or aircraft.

Game 2) SM vs. Orks, the SM player takes all scouts to try out a concept.

In both cases, the SM player loses.

What conclusions should be drawn from these examples to apply to the SM list?

Answer: None

Although, if one were trying to prove that Scouts are too expensive (i.e. you couldn't buy enough to swamp the Orks), then you do have some "statistical" proof to support your hypothesis.

CyberShadow:
I think that Honda has a good fix on my opinion of Battlestats. I do think that it is a very useful tool, but that this becomes more difficult with lists which are in such a state of flux. For example, of the Tau lists, which results deal with the version currently in the vault, and which with the current version here? It is impossible to tell, and the two lists are considerably different. I do think that Battlestats has a place in the development process, but more as an evaluation tool. I do think that there are some people with agendas, and that perhaps there are some results which are not true representations, but also that these are the minority and do not greatly affect the outcomes. Therefore, if Battlestats states that the Tau list is winning/losing more than its share of games after it has been 'written down in stone' then it will be examined as one source of the information.

The_Real_Chris:
Battlestats. Yes I enter the games I play (also PG kindly enters the tourney results as only an admin can do that many). I don't have much stock with it for anything more than watching trends. In lieu of decent info like what army list version, what size of battle etc to marry the results to I just look in when I enter a game to see how the percentages have changed. If they creep up odds on people have figured out how to ue a list, creep down and everyone has figured out how to beat it. Course this only works for races with a few games entered - for Tau with over a hundred its harder to see any significant change.

Certainly I think its good for Marines, Orks and Guard now they have been out so long and there odds seem to match mine.

Hena:
This brings up a real question. How hard would it to make a battlestats site. Which could include version numbers of lists and linking of batreps (and what else is needed)? Could someone host such a thing? I'm pretty sure that it would not be impossible... hard perhaps but...

dysartes:
Hena> It'd be doable, for sure - but it'd takequite a bit of effort to set up, and as CS points out, it'd keep growing as each new version of a list was added - plus the person running it would need to be kept up-tio-date with list releases.

Thinking on the verion factor, you could get around it (to an extent) by having a drop-down with the various version numbers in- these would then get stored as a variable. Of course, the version numbers would have to be driven by which army was selected, etc....

Then you've got to think about how you're displaying the data - just by adding version numbers in, it allows you to filter results to, for instance, show how AMTL v2.0 is performing, or how Tau lists since v4.0 have performed - which would certainly be useful compared to the current all-versions-in-one-table approach.

Adding a link to baterps would be fairly straightforward - you'd just need a field for people to enter the URL, including the http:// part.

Hmm....

*Goes away to think*

The_Real_Chris:
Plus you would have to consider how to deal with version two of AMTL playing version 6 and 7 of the Imperial Dairy Guard list - how to display results then? Auto ignore results against previous versions of a list or have the option to filter certain versions out?

dysartes:
It'd be doable to filter results to show only, say, results of games of AMTL 2.0 vs Tau 4.3.3 if desired - at a guess, I'd say you'd record the army type (ie, Iyanden, Steel Legion, Feral Orks) and their versions in one table, and use a key to refer to it in a main table, which'd be more efficient.

*Goes to dig up his lecture notes on database design/MySQL, and have a think about how to set it up*

The question becomes to do with how you want the data displayed, and whether you want to be able to display individual records. Also, preventing duplication of results (ie, both players reporting the same game) should be doable, with a little care.

CyberShadow:
Actually, you shouldnt even need a table of army versions. A simple field in the battle record for each forces version number would do. You could then compile a report with criteria of the army version number.

The problem may be the labelling system that is used. Unless you allowed the user to manually type in the army list version number (which is subject to error) you would need to be aware of every version number and update the option list as and when required.
Author:  nealhunt [ Mon Feb 20, 2006 3:58 pm ]
Post subject:  Battlestats
I posted my opinion on the other thread, but just to recap:

It's useful.  It has the same inherent flaws as any batrep, namely that you cannot count on it that the people doing the entering did a decent, honest job of playing.

For an army in development, because things are changing, the best thing to look at are trends.  If a change is followed by a sudden increase or decrease in overall win %, it's mostly likely due to the change.
Author:  dysartes [ Tue Feb 21, 2006 10:25 am ]
Post subject:  Battlestats
It is useful as it stands, but I think it could b made *more* useful. More information is always a good thing, and just seeing win/loss % isn't all we could get out of the info in that database - for instance, we could get some idea of what % of the time a given force achieves each GT victory condition in GT scenario games fairly easily - all the data we'd need to filter it down to that is in the db already, we just can't get at it.

With more views on the data, we'd be able to look at other areas too - a few ideas of the top of my head, based on the data that should already be being stored:

- % of wins/losses at different game sizes (provide views for the most common, ie 2k, 2500, 2700, 3k, 3.5k, 4k, 4.5k, 5k, plus one for "Other")
- % of games finishing on T3, T4 or a tiebreaker
- % of wins in T3, T4 or a tiebreaker
- Ability to view individual game information (a table with the info that's submitted displayed, with a row per game)

Now, that's 5 areas where we could get more information (including the victory condition one) from the data we store at the moment, which I'm certain would be of use to AC's at a minimum, let alone anyone else. I'm sure there's other views that others could suggest that would help, too.

Adding the ability to filter the results by the version of an army list would also enable us to see how different versions perform, and the trends from there might help indicate if a particular version is over- or under-powered.

And, if nothing else, it'd be an interesting programming challenge :) - not to mention giving me something to think about at work :p
Author:  nealhunt [ Tue Feb 21, 2006 3:13 pm ]
Post subject:  Battlestats
PG has the ability to do all of that and has been happy to do so in the past, especially when something looked squirrelly.
Author:  dysartes [ Wed Feb 22, 2006 6:55 pm ]
Post subject:  Battlestats
Really? That's the first time that I've heard anyone mention that it's possible to get more than just win/loss results as functionality from Battlestats.

I still think it could be better done than it is at present, though.
Author:  nealhunt [ Wed Feb 22, 2006 7:22 pm ]
Post subject:  Battlestats
PG set it up and maintains it.  He has access to all the information that has been entered - army v army, points, GT or not, which objectives, tiebreaker points, etc..

At one point, the Eldar win % spiked up to 70+% and on investigating, PG found that 19 entries had been by one person and they were almost entirely Eldar wins.  He emailed the guy to find out what was up and it turned out that it was a new group of EA players and the Eldar were romping everyone, but th emargins of victory were gradually slimming.  The player attributed most of the disparity to simple inexperience.
Author:  netepic [ Thu Feb 23, 2006 9:27 am ]
Post subject:  Battlestats
Its been done before, it is a relatively simple task to write a battlestats script that allows you to drill down to the individual results and include a field for a battle report.

The only problem is minimising the administrative procedures. I'd have no problem writing the site, just updating the version numbers would be a chore i'd rather avoid. I would never have time to keep them up to date. Hats off to PG for the work he does for the site :).
Page 1 of 1 All times are UTC [ DST ]
Powered by phpBB® Forum Software © phpBB Group
http://www.phpbb.com/