[mgoetze]

Does anyone want to share their EV playing 55% Robot Rebate tournaments? I would be interested if there is anyone breaking even or better on these. If so, how well should one be doing in the 25 cent robot duplicates before moving up to the rebate tournaments?

[Rain]

If you're interested in making the most BB$ and you're a very good player, dups are less profitable and much slower than robot rewards. Do dups for the challenge, robot reward for ROI...

[mgoetze]

Rain, on 2011-May-30, 10:28, said:

If you're interested in making the most BB$ and you're a very good player, dups are less profitable and much slower than robot rewards. Do dups for the challenge, robot reward for ROI...

I'm not a very good player, so I prefer something closer to Real Bridge for useful practice. Incidentally, I would welcome random hand Robodups... But while I prefer dups, that doesn't mean I wouldn't like to break even if possible.

[JLOGIC]

I'm pretty sure I'd be up but I played a huge majority of mine when it was just getting your money back for 55 % +. They are definitely beatable imo, especially if you adjust later and don't play to win but play to cash(I am usually playing to win because I am trying to beat cherdano or clee when I play them, which lowers my ROI for sure lol).

[mgoetze]

To answer my own question, the field in the 55% Robot Rebate tournaments seems, if anything, even weaker than the field in the 25 cent Robot Duplicates. I have played 5 of them now and cashed in every single one.

Interesting factoid: since it costs 37.5 cents to play the same amount of boards in robot duplicates, you only need to cash 5 times out of 12 for robot rebate to be cheaper per board than robot duplicate.

[mgoetze]

In 59 roborebates, I have averaged 54.81% with a standard deviation of 7.82%, which gives an EV of -26 cents. In 11 robodups, I averaged 54.12% with a standard deviation of 9.38%.

I guess there's kind of an upper limit on how well you can do in these things. I've seen some pretty bad auctions accidentally end up in the right contract, beating the field when they got a different defense. The robots just do too much random stuff.

Anyway, anyone averaging at least 53.5% should consider playing robot rebate, as you'll pay less per board (if your standard deviation is similar to mine).

[JLOGIC]

I was trying to set up a robodupe prop bet/challenge similar to the robot race one but it was tough to come up with a spot. Would be interesting to see average and std deviation for that, they have a lot less variance than the robot races so I imagine 100 would be a reasonable sample size for them.

[semeai]

mgoetze, on 2011-August-22, 17:01, said:

In 59 roborebates, I have averaged 54.81% with a standard deviation of 7.82%, which gives an EV of -26 cents. In 11 robodups, I averaged 54.12% with a standard deviation of 9.38%.

Some good data. Both sets work out to a standard deviation of 27% for a single board.

Quote

I guess there's kind of an upper limit on how well you can do in these things. I've seen some pretty bad auctions accidentally end up in the right contract, beating the field when they got a different defense. The robots just do too much random stuff.

I've seen top players with long term averages in the 58 or 59 range.

Quote

Anyway, anyone averaging at least 53.5% should consider playing robot rebate, as you'll pay less per board (if your standard deviation is similar to mine).

Neat.

JLOGIC, on 2011-August-23, 19:00, said:

I was trying to set up a robodupe prop bet/challenge similar to the robot race one but it was tough to come up with a spot. Would be interesting to see average and std deviation for that, they have a lot less variance than the robot races so I imagine 100 would be a reasonable sample size for them.

Over 100 roborebates or robodups I compute that with standard deviations like mgoetze, you'd need to beat the other guy by around 2% to achieve statistical significance ("95% confidence") that you're the better robodup player. Of course if you both play the same events there'd be a more sensitive test for significance. Statistical significance is probably not what you're going for, but I thought I'd throw that out there.

This would be neat to see. Maybe you should have side bets on whether you can break 60%?

[JLOGIC]

semeai, on 2011-August-23, 23:22, said:

This would be neat to see. Maybe you should have side bets on whether you can break 60%?

If someone would give me even money on this I'd think it was christmas

[semeai]

JLOGIC, on 2011-August-24, 03:20, said:

If someone would give me even money on this I'd think it was christmas

Oh dear. I guess the top players I've seen averages for mostly play the ACBL robot game, which maybe has a better field. Also, I suppose you'd be more motivated, plus none of the players whose averages I've seen are Bermuda Bowl representatives. Still, a 60% running average would be impressive. What were you thinking, 63%, 65%?

[hrothgar]

mgoetze, on 2011-August-22, 17:01, said:

I guess there's kind of an upper limit on how well you can do in these things. I've seen some pretty bad auctions accidentally end up in the right contract, beating the field when they got a different defense. The robots just do too much random stuff.

It's possible that the awful auction influenced the defense...

[Phil]

I looked at some weekly averages for some good players I know. The range was 56 to 63. All had at least 15-20 games in.