[AyunuS]

I've been getting tired of playing with people where I'm always guessing if they're actually using systems or not and so it's been pretty tough doing nearly as well as I'd like. But with robots it's so much easier since they actually follow a system. Anyway, I'm still a junior, but I'm aiming to be one of the best players at this game someday, and right now, I feel like I'm already doing extremely well at a table of 3 robots, but I'd like to know what's the best anyone can do so I can see how close I am to it. Here's an image of my current best of a little more than 2.6 IMPs/board on average.

http://img109.images...1/superimps.png

[PhilKing]

Measuring success by numbers is an excercise in futility if you are trying to improve. You could play poorly and win 100 imps. In the short and medium term, bridge players will experience fluctuating levels of success independent of how well they are playing.

If you really want to improve by studying your robot sessions, you could do the following:

1. Save your sessions in a pbn or lin file and open them in a strong playing program. I would suggest Jack5, since it will allow you to review the hands as played. If that is too expensive (you might be able to pick up an earlier version more cheaply), you can review them in the BBO hand viewer, which you can select as your default program for opening lin and pbn files.

2. Go through them meticulously using the GIB DD function (if using BBO replayer) to see if you could have done better at any stage. In Jack5 this is much easier and quicker.

[billw55]

Depends. Do I get to pick the humans at the other tables?

[AyunuS]

Yes, I know it's not that good of a way to measure success but it seems better than the other ones I have. Are any good playing programs free? I already looked over each hand as I played them, and figure maybe I should have had one more trick on the 3♣ bid hand, but I think I did the best I could on all the rest.

And you do not pick the humans at the other tables. They are just supposedly about random. But most of them don't seem all that good as on some of the hands, like where I got all those imps from 6♦ I noticed I was the only one that showed a splinter when I thought it was an obvious bid. I figure if you play around 20+ boards or more it should approximately even out but I know it is still luck-based in some ways. But I have stayed at one table with 3 robots long enough to get my IMPs into the 300s before so I don't think it's just me being lucky.

[billw55]

Yeah, on average I guess it must mean something. Just watch out for weird results, like board 22 in the image you posted. -140 on an uncontested partscore, +9 IMPs? Presumably at the other tables EW are bidding game and making it, but this seems unrelated to your own actions. Possibly board 19 also, -680 for +13, did you do something to stop EW bidding slam?

I wonder, when playing with three robots, are scores compared with tables of four humans? Or only robot tables?

[PhilKing]

AyunuS, on 2013-March-07, 17:48, said:

I figure if you play around 20+ boards or more it should approximately even out but I know it is still luck-based in some ways.

This is how to eliminate variance - play at least twenty boards.

[winkle]

For a challenge try sitting east/west.

[barmar]

billw55, on 2013-March-08, 07:13, said:

I wonder, when playing with three robots, are scores compared with tables of four humans? Or only robot tables?

I think they're always compared with other tables with robots, but not necessarily three robots. So you might be compared against an all-human partnership beating up on the robots. Another thing is that in the MBC, you can choose your seat. If you just click on "Take me to a table with 3 robots" you get put in South, so that's where most people play, but you can move around if you want. As a result, you could be compared against robots, not other humans.

But when I looked at a recent set of hands I played with 3 robots in the MBC, most of the other tables were also 3 robots, with humans in South. So it's mostly a fair comparison against similar tables.

In robot tourneys there's none of this variation: all tables have humans South, robots in the other 3 seats.

[barmar]

PhilKing, on 2013-March-08, 07:22, said:

This is how to eliminate variance - play at least twenty boards.

Which is why it sucks that robot tourneys are only 12 boards. We have the 18 board tourneys, but they don't get much attendance, so no one has even contemplated 24 or 26, like a session at a real club game or tournament. People just don't seem to be willing to commit to a significant length of time for online bridge.

[MickyB]

barmar, on 2013-March-08, 09:44, said:

Which is why it sucks that robot tourneys are only 12 boards. We have the 18 board tourneys, but they don't get much attendance, so no one has even contemplated 24 or 26, like a session at a real club game or tournament. People just don't seem to be willing to commit to a significant length of time for online bridge.

Maybe more people would play in longer tournaments if they were advertised as being "officially variance-free".

[JLOGIC]

There is far less variance in a 12 board robot duplicate than a 24 board human tournament, it's not even close

[Leo LaSota]

barmar, on 2013-March-08, 09:44, said:

Which is why it sucks that robot tourneys are only 12 boards. We have the 18 board tourneys, but they don't get much attendance, so no one has even contemplated 24 or 26, like a session at a real club game or tournament. People just don't seem to be willing to commit to a significant length of time for online bridge.

The 18 board tournaments seemed to be doing OK when you offered an 18 bd MP game :02 after the hour AND an 18 bd IMP game :32 after the hour during prime hours. When you eliminated the variety of 18 bd games offered and went back to only offering 18 bd MP tournaments, the level of participation dropped significantly which is no surprise. I am baffled as to why you elected to eliminate the 18 bd IMP tournaments. I do not believe that you gave them enough time before you eliminated them entirely.

[AyunuS]

For those that know stats, the variance of distributions of just the sum of each of their variances. Thus 24-board has provably higher variance than 12 board, and yet it still seems like 24-board would get closer to averaging out and I'd think for sure the better players would be more likely to win on a 24-board than on a 12-board.

Also on board 19 I did nothing to stop them from bidding slam. It was just good luck. On the partscore at the end, a bunch of tables bid and made game so I got a bunch of points.

[PhilKing]

It seems like you've got it cracked.