If you've been following the posts around this subforum recently you'll have realised that The Open Tactic League is a poor indication of how good a tactic is. It uses an outdated system which involves limiting a tactic to ten runs which is far too few a number of runs to really evaluate how good a tactic is. I intend to change that. Welcome to The Alternative Tactic League.
How It Works
I'll be using the benchmark .exe and save game that Tapani manipulated. Each tactic will be run 100 times.
How To Enter
Post any tactic you would like testing below. I'll create a spreadsheet similar to the Open Tactic League Spreadsheet so you can see how many goals are scored and conceded. (NOTE: I will only be testing tactics submitted to this thread, I'm not going out of my way to collate and run tactics which are quite simply outdated).
Context (by Tapani)
Current Table (Sorted by Points Average)
Last edited by Jesus; 21-04-16 at 09:06 PM.
Would it be a good idea to test the tactic with three different teams? de Graafschap, one good team (PSV?) and one really bad (Excelsior?). Just use the same benchmark. And of course as many runs as possible.
Last edited by Jesus; 02-03-16 at 08:55 PM.
can u test this for me pls ? it gave me 28w 4d 2d with scored 90 and conceaded 20, points 88 with one test i did cause of a slow pc.
Last edited by minusmf; 01-08-16 at 01:08 PM.
Should be ok ! Thank You for givin motivation me about my tactic in future
I've been working on a few 3-4-3 formations. I am not happy with my CBs and my main FC so I want to fix them up a bit.
That said can you test these please?
3-4-3 Mark Best (Seems to do well enough on the De Gra test)
3-4-3 Mark Best edit 2
I have more in the wood but 3 is loads to ask to test for now.
Tactic Name Formation Centralised? For Against Points Games Won Drawn Lost Goals Average Conceded Points Points Average 3-4-3 Mark Best_edit_2 343 No 5591 2438 5829 2600 1786 471 343 55.91 24.38 57.13 - 59.45 58.29 3-4-3 Mark Best 343 No 5551 2469 5774 2600 1756 506 338 55.51 24.69 56.47 - 59.01 57.74 1ObiWonKonobi 343 No 4984 2740 5235 2600 1572 519 509 49.84 27.40 50.94 - 53.76 52.35
Last edited by Jesus; 09-03-16 at 12:05 AM.
Maybe add instructions on how someone else might be able to test tactics as well. We know from OTL and the like that people's enthusiasm can come and go.Originally Posted by Jesus
Do you have any better suggestions on how to mitigate the shortcomings of the OTL?
In case this is going to take off, lets try to do it right.
First, 100 runs is not always enough. Actually, it is usually not enough. It is not me being snobby, it is unfortunate reality (try running the same tactic 100 times, but under different names -- you'll see some difference!).
Also running enough runs for every tactic is too much work.
Could we have something in between? Maybe increase the burden of proof for higher positions?
Example: A tactic claiming to be in, say top 10 after 100 runs, would need 200 runs to prove their position.
Similarily, a tactic claiming to be in top 5 after 200 runs, need 400 runs to prove it was not just luck.
Top 3 would need whopping 800 runs, and to be the champion, you need to beat the old champ in 1600 runs.
(Actually starting the doubling at top 20 is probably a better idea, but possibly too much work).
CMTacTool displays the 95% confidence interval for the average score (as low - high). This means that the true average of the tactic is between those number for 19 tactics of 20.
Over the past week I've developed my own batch files which runs the tactic 100 times and then I have to take the benchresult.txt file and use a spreadsheet I designed to get the data in the right order.
When I asked how you run in parallel you only gave me a snippet of code to work with and I have had to do some work to even get to this point (and bear in mind I don't have a background in programming, I've literally had to Google a solution and test my own batch file which, although it has been fun, there is a more efficient way of doing it).