Fraggle_Reloaded_c playing cat and mouse with me?

See https://www.littlegolem.net/jsp/game/game.jsp?gid=2472114
It seems to me that bots have greater advantage when playing higher point games. See my game in this 19-points game.
After trailing 2-6, I came back to 6-6, and then only after trailing 6-14 I won 3 points again. The current point will no doubt be for Fraggle_Reloaded_c
So possibly I played weaker and/or was less lucky, but I get the feeling that the bot plays a bit weaker when it leads with a number of points.
In a (long time ago) match against RoRoRo the bot, somewhere in the middle I leaded with 6 or 7 points, and then maybe I played more careless or the bot played stronger, but I lost anyway without much chances.

What are your experiences?

Anyway, In the current tournament, Fraggle_Reloaded_c will win without a doubt seeing the current standings.

it would be odd to program / train a bot that played differently depending on the match score. Doing so makes sense for e.g. Backgammon, where the individual games can have payoffs different than 0/1 (Gammons and Backgammons) and optimal play depends on the score, but in EWN you always just want to choose a next position to maximize win prob for that game.

I think the sampling is too low to tell, but that seems dubious.
Maybe you are thinking about typical MCTS play where the bots with an advantage do not care to play best move but just any move allowing to win ?

I think it is more difficult/ less statistically probable to win in a game with higher points, so good job to keep par with Fraggle_Reloaded !

-yes, there is the phenomenon e.g. in Go where the AlphaZero-style bots can start playing worse when they have a point lead. But when you are playing them in a best-of-7 match, they won't play better or worse based on match score, because (by design, and very reasonably) they always ignore the match score, all their training revolves around single-games. And I would assume the same of EWN bots here, since the designer presumably understands this. As opposed to Backgammon, where the leading bot (Extreme Gammon 2) takes match score into account when playing, and again for good reason (Gammons/Backgammons).

-yes, it is harder to win a best-of-7 match against a superhuman bot, than to win a best-of-5 (say).

add3993, I'm quite close to Go scene - not sure what you meant exactly, but to be more precise. Bots in Go that are on superhuman level do not play worse when they have lead, they play more "safely" and reduce variance by giving up point lead. In Go there is no benefit of winning game by high margin, so they maximize the function of winning the game, not maximize the function of winning by highest possible margin. This is huge difference, as generally also human players in Go play that way (but of course, for bots it's more extreme). I don't think that applies to EwN, as EwN is basically sum of independent games, you never increase your match-winning chances by giving up single games within this match.