Code center > Darwinbots3
DB3 Leagues
spike43884:
I've been mulling over the elo thing. What about just using multiple factors across the entire simulation and scoring them, instead of relative to their opponents, relative to their performance in the simulation...Do it for multiple battles, maybe even average out the score and then rank them?
Peter:
--- Quote from: Numsgil on April 12, 2015, 04:39:12 PM ---Looks like it places them all around the same skill? I think that's what we'd want if so. It's hard to get a sense of the relative scales. Can you add in a strategy that has a 80% chance to win against any of RPS? I'd like to see if it's noticeably higher than all the others.
--- End quote ---
Added.
Well the Rocks are placed higher than Scissor, Scissor higher than Paper. But the difference isn't a lot.
id ConservativeRating Mean StandardDeviation win% wins\draws\losses\matches\type
0 22.0095591377625 25.2836566772087 1.09136584648208 16.66667% 7 26 7 42 Scissor
1 21.7821811891753 24.8081119506593 1.00864358716132 23.07692% 12 23 11 52 Scissor
2 19.2005201463236 22.5465603332125 1.11534672896296 31.37255% 16 4 28 51 Paper
3 23.1122029152767 26.3987975437787 1.095531542834 53.84615% 28 5 13 52 Rock
4 22.1717569038829 25.5748763424698 1.13437314619566 21.95122% 9 23 4 41 Scissor
5 21.9605295488926 25.3588763504315 1.13278226717963 21.95122% 9 18 8 41 Scissor
6 21.1505654852868 24.269753597743 1.03972937081875 16.66667% 8 24 12 48 Scissor
7 18.2254824660445 21.8923336948571 1.22228374293755 22.72727% 10 4 26 44 Paper
8 21.4013812078553 24.7129946001344 1.1038711307597 20.93023% 9 20 9 43 Scissor
9 22.9477399984015 26.4990987016865 1.18378623442832 51.11111% 23 5 13 45 Rock
10 23.6013905336762 26.8589475968437 1.08585235438918 73.33334% 33 0 12 45 Percent80
After a million random matches the upper 2 sigma of Scissor is still higher than the lower 2 sigma of Rock.
id ConservativeRating Mean StandardDeviation win% wins\draws\losses\matches\type
0 22.4396128116937 24.6603104463431 0.740232544883134 20.1069% 33217 82356 33136 165202 Scissor
1 22.3980369678499 24.6155831330575 0.739182055069193 19.99733% 32991 82504 32976 164977 Scissor
2 18.6285481147599 21.0068128521428 0.792754912460967 19.95815% 33000 16463 99321 165346 Paper
3 24.7068243962618 27.0176793425063 0.770284982081475 59.98304% 99038 16415 33058 165110 Rock
4 22.8829721831836 25.1137388028982 0.743588873238181 20.0285% 33030 82475 32812 164915 Scissor
5 23.0007569918914 25.2302209591881 0.743154655765566 19.97823% 33042 82536 33128 165390 Scissor
6 22.8817029092318 25.1019168680217 0.740071319596633 20.11042% 33330 82598 33106 165735 Scissor
7 20.4789112234694 22.7848573440528 0.768648706861115 19.92037% 32920 16463 99293 165258 Paper
8 23.2252037617677 25.4660024444581 0.746932894230129 19.98256% 33004 82413 33157 165164 Scissor
9 25.138645723613 27.4710506811453 0.777468319177446 60.21605% 99277 16415 32862 164868 Rock
10 24.5903878471413 26.8158904155278 0.741834189462153 79.99807% 132500 0 33129 165629 Percent80
Edit: it's wrong
Numsgil:
I find it interesting that after a million matches the one that wins 80% of the time is actually a bit lower than the second rock player. I guess the system finds its occasional losses to "weak" bots confusing. Also the flat win rate treats a draw the same as a loss, which is probably over penalizing the rock strategy.
Interesting results, I'll need to mull it over.
Peter:
For reference the results of player 0>1>2 with a million games.
1. The difference is quite big in comparison with RPC + random.
2. Fights with big difference in rating cause NaN errors. Games with NaN errors didn't count in the final results. You can see that in matches played, top and down players got less games registrations. NaN erros may be fixed somewhere in the settings, but at least in this implementation it doesn't like huge power differences. This can also be the reason for the flattened win rates, as a big portion of games got skipped. Might be fixed with different settings, I picked the default rating settings.
id ConservativeRating Mean StandardDeviation win% wins\draws\losses\matches
0 157.364160859048 175.093666523749 5.90983522156717 100% 93357 0 0 93357
1 122.187090794375 137.148117096133 4.98700876725273 68.71486% 94351 0 42957 137308
2 89.1558840881495 103.337093844302 4.72706991871741 52.63669% 95334 0 85783 181117
3 57.9310423376475 71.3621156175018 4.47702442661809 51.34298% 95290 0 90305 185595
4 27.50057942478 40.4213063289044 4.30690896804144 50.3262% 94266 0 93044 187310
5 -3.32350541905488 9.68240334301635 4.33530292069041 49.87444% 93547 0 94018 187565
6 -34.637104314333 -21.3219560268339 4.4383827624997 48.72458% 90999 0 95763 186762
7 -67.5653127637627 -53.3291196269003 4.74539771228745 47.14106% 85512 0 95884 181396
8 -102.079416500229 -87.1381894801877 4.98040900668033 31.22995% 43028 0 94750 137778
9 -142.80346240494 -125.061759330556 5.91390102479457 0% 0 0 93180 93180
Peter:
Uh, there was a mistake with creating wins from the 80% win player. Half of his wins were registered as draws. :redface:
Correct stats.
RPC/random
id ConservativeRating Mean StandardDeviation win% wins\draws\losses\matches\type
0 21.2025014802368 24.5072604715912 1.10158633045147 19.04762% 8 27 7 42 Scissor
1 21.1703789746567 24.2605458085435 1.03005561129561 22% 11 24 15 50 Scissor
2 19.3873917419835 22.8479451170323 1.15351779168295 38.77551% 19 3 27 49 Paper
3 23.2809913533546 26.6242733109851 1.11442731921019 58% 29 10 11 50 Rock
4 22.1327373676122 25.5369245974301 1.13472907660596 26.31579% 10 25 3 38 Scissor
5 20.8980341485925 24.3963410032841 1.16610228489721 23.07692% 9 21 9 39 Scissor
6 21.6943165763599 24.9387952612258 1.0814928949553 34.78261% 16 23 7 46 Scissor
7 17.9036343796341 22.1480159446247 1.41479385499687 40% 14 3 18 35 Paper
8 20.7780899350452 24.0535665005142 1.09182552182302 26.08696% 12 24 10 46 Scissor
9 22.3978387144379 25.7676782802645 1.12327985527554 49.01961% 25 10 16 51 Rock
10 26.8815568846738 31.1951332533136 1.43785878954662 76.19048% 32 0 10 42 Percent80
RPC/random after a million games
id ConservativeRating Mean StandardDeviation win% wins\draws\losses\matches\type
0 21.8673281444733 24.1097779335181 0.747483263014924 27.83857% 45947 82858 36243 165048 Scissor
1 22.0131180929267 24.2333995808674 0.740093829313591 27.82333% 45948 82799 36395 165142 Scissor
2 17.8442112961662 20.2239537778002 0.793247493877987 27.95814% 46141 16516 102379 165036 Paper
3 24.4196078416116 26.7526319905644 0.777674716317609 68.05129% 112237 16466 36227 164930 Rock
4 21.8505862259689 24.0823444478475 0.743919407292882 28.11717% 46485 82358 36483 165326 Scissor
5 21.9716403885154 24.2016352220766 0.743331611187086 28.10022% 46489 82414 36537 165440 Scissor
6 21.7202167758399 23.9643481135206 0.748043779226886 28.10896% 46479 82355 36519 165353 Scissor
7 19.2396494056951 21.5913457018039 0.783898765369623 27.98824% 46360 16516 102765 165641 Paper
8 21.3340998118027 23.5937401326201 0.753213440272473 28.09844% 46422 82480 36310 165212 Scissor
9 24.2867711234825 26.617695694018 0.776974856845171 68.22602% 112824 16466 36078 165368 Rock
10 26.9820148326741 29.4765897709184 0.831524979414755 80.06544% 132638 0 33024 165662 Percent80
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version