If you berserk and draw you get no bonus. Do you have to recalculate some things now?
Are provisional players included in this analysis? Maybe the same rating peak is due to both being 1500(?) and the 300 peak from having 1 win vs a new player? I think this is unlikely though for the 300 peak and it's more likely a measure that most people use to decide whether or not to berserk.
@oortcloud_o
It would be interesting to consider the effects of "Berserk only" accounts (smurfs) in this investigation, as I assume they account for a large portion of the games, particularly where the berserker is at +300 or lower rating difference with their opponents. We can see in the graph for the number of berserk games in total, that there is a jump at +300 rating. My initial hypothesis for the steep downfall at said rating difference would be that this is where we start seeing a larger number of "regular accounts" berserk their game. Since a large portion of the berserk wins with lower "rating advantage" might be from accounts that berserk most or all of their games, that section of the graph would be shifted upwards, closer to the normal win rate. This might not be the only reason for the anomaly, or perhaps even completely unrelated, but I think it is worth considering. Unfortunately, I suspect it would be very difficult to prune these accounts from the data set; We can only guess based on account names that include "berserk", and their percentage of berserk games.
It would be interesting to consider the effects of "Berserk only" accounts (smurfs) in this investigation, as I assume they account for a large portion of the games, particularly where the berserker is at +300 or lower rating difference with their opponents. We can see in the graph for the number of berserk games in total, that there is a jump at +300 rating. My initial hypothesis for the steep downfall at said rating difference would be that this is where we start seeing a larger number of "regular accounts" berserk their game. Since a large portion of the berserk wins with lower "rating advantage" might be from accounts that berserk most or all of their games, that section of the graph would be shifted upwards, closer to the normal win rate. This might not be the only reason for the anomaly, or perhaps even completely unrelated, but I think it is worth considering. Unfortunately, I suspect it would be very difficult to prune these accounts from the data set; We can only guess based on account names that include "berserk", and their percentage of berserk games.
@Cedur216 said in #2:
>
Thank you for letting me know! It will affect the dataset but not by much since a draw is a low-likely result of the berserk games
>
Thank you for letting me know! It will affect the dataset but not by much since a draw is a low-likely result of the berserk games
Clearly, you should join lichess.org/team/berserkers and go berserk every game.
Well there is "Smart Berserk" browser extension that assists a player to help them make a logical choice, to berserk or not. The berserk lights up when it thinks it's appropriate to do it. Seems like only 111 users have it.
chrome.google.com/webstore/detail/smart-berserk-lichess/gebadcpcikhmmddgcknoiikdikojbphc
Since it's a form of assistance to a player, should it be permitted on lichess or not?
chrome.google.com/webstore/detail/smart-berserk-lichess/gebadcpcikhmmddgcknoiikdikojbphc
Since it's a form of assistance to a player, should it be permitted on lichess or not?
Exellent blog
Have you seen that graphs on github?
github.com/jmviz/when-to-berserk-on-lichess
I did not check the math or anything, but its provided if anybody wants to, and if it was under the same rules, but I have reasonable success following that guideline and there is some quite interesting stuff.
For example in 3+0 you should always(!!) berserk with white, if your opponent berserked too and you do not have a streak yet. And even with one win you should 9/10 times berserk in the same situation.
Or the stats, that you should always(!!) berserk against much stronger players, even on a streak!
github.com/jmviz/when-to-berserk-on-lichess
I did not check the math or anything, but its provided if anybody wants to, and if it was under the same rules, but I have reasonable success following that guideline and there is some quite interesting stuff.
For example in 3+0 you should always(!!) berserk with white, if your opponent berserked too and you do not have a streak yet. And even with one win you should 9/10 times berserk in the same situation.
Or the stats, that you should always(!!) berserk against much stronger players, even on a streak!
@Cedur216 said in #2:
> If you berserk and draw you get no bonus. Do you have to recalculate some things now?
errr... no
If you go berserk and win you get three points
If you go berserk and draw you get two points
If you go berserk and lose you get nothing
> If you berserk and draw you get no bonus. Do you have to recalculate some things now?
errr... no
If you go berserk and win you get three points
If you go berserk and draw you get two points
If you go berserk and lose you get nothing