Stan` Posted January 28, 2024 Author Report Share Posted January 28, 2024 On 28/01/2024 at 1:35 PM, Feldfeld said: No separation Expand Okay. Do we count replays with no winners as draws ? https://github.com/mmai/glicko2js They say there that You should not update the ranking after each match. The typical use of glicko is to calculate the ratings after each tournament (ie collection of matches in a period of time). A player rating will evolve after a tournament has finished, but not during the tournament. So I'm not sure it would work well in our case Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 28, 2024 Report Share Posted January 28, 2024 (edited) On 28/01/2024 at 2:28 PM, Stan` said: Okay. Do we count replays with no winners as draws ? Expand Imo, no. There are no draws in 0 A.D., and many replays with no winners would just be replays where there is actually a winner in the field but the game was exited without a result. Could be disconnection too, anyway a draw is a definitive result and it would not represent what happened in the match. It would be better to just ignore these games in the calculation. Could consider in the future having a tool analyzing the replay and deciding if there actually was a winner. On 28/01/2024 at 2:28 PM, Stan` said: https://github.com/mmai/glicko2js They say there that You should not update the ranking after each match. The typical use of glicko is to calculate the ratings after each tournament (ie collection of matches in a period of time). A player rating will evolve after a tournament has finished, but not during the tournament. So I'm not sure it would work well in our case Expand I know lichess uses glicko and it is updated game by game. Here is what they say about it: https://lichess.org/faq#ratings https://lichess.org/page/rating-systems So the lichess boys are quite confident this is the better rating system. It's still difficult to compare as they also have a lot more games. As for what is written in the glicko2js github page, too bad they didn't get very specific about it, notably if the collection of games should involve the same players. Because when you think about it, provided the replay pallas scrapper runs at intervals, or if players post replays by dumps, then at each update you'd actually have a collection of replays to evaluate. Edit: nvm can't read. They did say 10-15 games per player. But it remains that lichess updates on the fly Edited January 28, 2024 by Feldfeld Quote Link to comment Share on other sites More sharing options...
Stan` Posted January 28, 2024 Author Report Share Posted January 28, 2024 Another issue we have is that unlike on lichess, one can upload replays in any order, so I suppose I should recompute the whole thing each time someone adds a new replay for all players (since a changing rating for someone would change everybody that ever played against them) Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 28, 2024 Report Share Posted January 28, 2024 On 28/01/2024 at 3:21 PM, Stan` said: Another issue we have is that unlike on lichess, one can upload replays in any order, so I suppose I should recompute the whole thing each time someone adds a new replay for all players (since a changing rating for someone would change everybody that ever played against them) Expand Yeah that's a bigger issue. A mitigation I see to avoid weird looking swings would be to recompute the whole thing at even slower intervals at which replays are scrapped. So there could be replays displayed not yet used in the rating calculation. So the problem would be "hidden" behind the big quantity of replays at each update. However, it happens that the multiplayer community plays much more TGs than they do 1v1s so not sure it would work well. Quote Link to comment Share on other sites More sharing options...
Stan` Posted January 28, 2024 Author Report Share Posted January 28, 2024 On 28/01/2024 at 3:27 PM, Feldfeld said: Yeah that's a bigger issue. A mitigation I see to avoid weird looking swings would be to recompute the whole thing at even slower intervals at which replays are scrapped. So there could be replays displayed not yet used in the rating calculation. So the problem would be "hidden" behind the big quantity of replays at each update. However, it happens that the multiplayer community plays much more TGs than they do 1v1s so not sure it would work well. Expand That could work I suppose. I can also add a big button for admins to recompute it. I'll try to implement it. Should we unlike lichess start at 1200 ? Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 28, 2024 Report Share Posted January 28, 2024 On 28/01/2024 at 3:30 PM, Stan` said: That could work I suppose. I can also add a big button for admins to recompute it. I'll try to implement it. Should we unlike lichess start at 1200 ? Expand Considering that experienced players are more likely to share their replays, I suspect that starting at 1500 would actually look more like ingame ratings. 1 Quote Link to comment Share on other sites More sharing options...
Stan` Posted January 28, 2024 Author Report Share Posted January 28, 2024 Regarding determining if a player has lost or not, would the percentage Building Lost + Units lost work? Lowest Percentage wins? (This would be done only for 1v1 and if there is no state set to won) Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 28, 2024 Report Share Posted January 28, 2024 On 28/01/2024 at 3:49 PM, Stan` said: Regarding determining if a player has lost or not, would the percentage Building Lost + Units lost work? Lowest Percentage wins? (This would be done only for 1v1 and if there is no state set to won) Expand Imo, first of all, for whichever criterion that is used there needs to be a threshold to say that the game is decided. Because it can be that some replays with no winner were truly undecided at that time (e.g. disconnection). So the algorithm would still decide to discard those games in the rating calculation. Then, as for the actual criterion, imo the biggest tell is if there was a population swing close to the end of the game, or a swing a few minutes before that has never been recovered. If a player has noticeably bigger population after that event then he can be considered winner. If there was a big fight at the end of the replay but no one on top in the population statistic then economy and military score can also be taken into consideration. Quote Link to comment Share on other sites More sharing options...
borg- Posted January 28, 2024 Report Share Posted January 28, 2024 @Feldfeldback to play Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 28, 2024 Report Share Posted January 28, 2024 Not motivated for now Quote Link to comment Share on other sites More sharing options...
borg- Posted January 28, 2024 Report Share Posted January 28, 2024 On 28/01/2024 at 6:05 PM, Feldfeld said: Not motivated for now Expand Playing other game? Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 28, 2024 Report Share Posted January 28, 2024 On 28/01/2024 at 8:37 PM, borg- said: Playing other game? Expand None, or very casually at best Quote Link to comment Share on other sites More sharing options...
Stan` Posted January 28, 2024 Author Report Share Posted January 28, 2024 On 28/01/2024 at 3:27 PM, Feldfeld said: Yeah that's a bigger issue. A mitigation I see to avoid weird looking swings would be to recompute the whole thing at even slower intervals at which replays are scrapped. So there could be replays displayed not yet used in the rating calculation. So the problem would be "hidden" behind the big quantity of replays at each update. However, it happens that the multiplayer community plays much more TGs than they do 1v1s so not sure it would work well. Expand Found this little thing that might solve the issue (it's also what lichess uses), but I need to convert the rust code to TS https://github.com/gpluscb/instant-glicko-2/tree/f34fd27aba57c21b0554a6ccc2b6a26559a0f5e7 1 Quote Link to comment Share on other sites More sharing options...
Stan` Posted January 30, 2024 Author Report Share Posted January 30, 2024 Pushed some updates, you now have the glicko rating on the player profile. Also added the game rating info I have to that graph. I didn't exclude 1v1 from local ratings as for now those are two different things. People with the contributor badge can upload replays. I haven't set up the scrapper yet, not sure if there I should create a specific thread for it and whether it's a good idea. 3 Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 31, 2024 Report Share Posted January 31, 2024 Nice! The glicko data looks inconsistent. It tells me my rating is 1545 while the graph puts me above 1600 Now may be time to invite the 0addicts to dump their replay. Many players should have hundreds of teamgames, though they would all be in now outdated version of community mod. (that would amount to GBs of disk size) Is there a duplicate detection for replays? Ideally it would keep also the longest version available (there could be a somewhat frequent situation where a player spectated a 1v1 then left in the middle of the game, and posted the replay first) Guess this problem would occur only if players post replays by dumps instead of hand picked 1 by 1. I'd recommend posting the link to the website at the top of the opening post of the thread for visibility. Quote Link to comment Share on other sites More sharing options...
Stan` Posted January 31, 2024 Author Report Share Posted January 31, 2024 On 31/01/2024 at 6:44 AM, Feldfeld said: Is there a duplicate detection for replays? Ideally it would keep also the longest version available (there could be a somewhat frequent situation where a player spectated a 1v1 then left in the middle of the game, and posted the replay first) Expand Only the first one remains. The replay matchid is used to avoid duplicate replays. On 31/01/2024 at 6:44 AM, Feldfeld said: The glicko data looks inconsistent. It tells me my rating is 1545 while the graph puts me above 1600 Expand Ha I messed up the sorting should be fixed now. Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 31, 2024 Report Share Posted January 31, 2024 On 31/01/2024 at 8:35 AM, Stan` said: Ha I messed up the sorting should be fixed now. Expand Now the graph indicates I'm dropping rating for each match, not sure that's accurate Quote Link to comment Share on other sites More sharing options...
Stan` Posted January 31, 2024 Author Report Share Posted January 31, 2024 On 31/01/2024 at 5:20 PM, Feldfeld said: Now the graph indicates I'm dropping rating for each match, not sure that's accurate Expand I'm not sure but I think it's normal due to the very low amount of matches. At the beginning you have a deviation of 350 which probably boosts your rating at each each match, now your deviation is closer to 45. But at least the dates are in order now Another thing is that there aren't many matches. The code responsible for this is there https://github.com/StanleySweet/replay-pallas-api/blob/main/src/instant-glicko-2/Glicko2Manager.ts#L64 1 Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 31, 2024 Report Share Posted January 31, 2024 On 31/01/2024 at 5:32 PM, Stan` said: I'm not sure but I think it's normal due to the very low amount of matches. At the beginning you have a deviation of 350 which probably boosts your rating at each each match, now your deviation is closer to 45. But at least the dates are in order now Another thing is that there aren't many matches. The code responsible for this is there https://github.com/StanleySweet/replay-pallas-api/blob/main/src/instant-glicko-2/Glicko2Manager.ts#L64 Expand What I mean is that the graph indicates I lost every game beside the first, at least that's the understanding I get from this. But since most replays on the website are from the tournament I'd expect it to be mostly victories, and victories would bring the graph up Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted January 31, 2024 Report Share Posted January 31, 2024 I checked the graph and it seems like what is plotted is (1500 + PlayerDeviation) instead of PlayerRating This is why the graph gets closer and closer to 1500 after each match. Maybe after giving a replay to the glicko code it doesn't take the current player rating, but instead the base value of 1500? Quote Link to comment Share on other sites More sharing options...
Stan` Posted January 31, 2024 Author Report Share Posted January 31, 2024 On 31/01/2024 at 5:54 PM, Feldfeld said: I checked the graph and it seems like what is plotted is (1500 + PlayerDeviation) instead of PlayerRating This is why the graph gets closer and closer to 1500 after each match. Maybe after giving a replay to the glicko code it doesn't take the current player rating, but instead the base value of 1500? Expand Found the guilty line https://github.com/StanleySweet/replay-pallas-api/commit/e085022fc93e7a6578a82d18c7a3df21970a2d5e Explains the weird link between deviation and rating. I hope my friend's @Stockfish's mouse is okay, they are clicking at least twice as much as any one @borg- has a terrible rating because he lost the three matches on the site (His rating is 900? the '?' being for provisional) 1 1 Quote Link to comment Share on other sites More sharing options...
sarcoma Posted February 2, 2024 Report Share Posted February 2, 2024 On 31/01/2024 at 9:54 PM, Stan` said: @borg- has a terrible rating because he lost the three matches on the site (His rating is 900? the '?' being for provisional) Expand borg is aqua level Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted February 2, 2024 Report Share Posted February 2, 2024 Tried to upload a replay of my first game back, it didn't seem responsive. I had no feedback on clicking the upload button. Quote Link to comment Share on other sites More sharing options...
Stan` Posted February 3, 2024 Author Report Share Posted February 3, 2024 On 02/02/2024 at 8:12 PM, Feldfeld said: Tried to upload a replay of my first game back, it didn't seem responsive. I had no feedback on clicking the upload button. Expand Yeah I need to work on interactivi (also for a phone version)ty. Usually when the replays upload if it works, one or multiple replay lines will appear below the upload block for replays that were uploaded successfully. Quote Link to comment Share on other sites More sharing options...
Feldfeld Posted February 3, 2024 Report Share Posted February 3, 2024 On 03/02/2024 at 1:29 AM, Stan` said: Yeah I need to work on interactivi (also for a phone version)ty. Usually when the replays upload if it works, one or multiple replay lines will appear below the upload block for replays that were uploaded successfully. Expand I guess I couldn't upload the replay since nothing happened. I'm on Librewolf browser Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.