BGonline.org Forums

Nine 5k results (report) - that why people need to check SI

Posted By: eXtreme Gammon
Date: Monday, 17 January 2011, at 6:56 p.m.

In Response To: Nine 5k results (report) - that why people need to check SI (Stick)

Nack Yup. Your 25k result was [D S9]. In order to achieve the 46k extended result you got, the last 20k would had to have been [S D7]. In other words, it is as if you did two separate and independent rollouts with results of

[D S9] <25 [S D7] <20 The rollouts could just have easily occurred in the other order. It is conceivable that depending upon which rollout happens to occur first, the roller is "satisfied" and doesn't bother extending.

The 25K shows a 0.007 CI, so a 20K would have been at 0.008 CI. So neither individual RO had reach stat sig (if you defined stat sig at 99%). and the combine RO did not either.

Let's remember that the CI given are at 95%, this means 1 time out of 20 it will be outside the interval (which was not the case for the 20K but is for the 25K)

For RO such as the one for opening I agree with Nack that we should extend them so the confidence of the difference (which is typicaly 1.41 bigger than of the move) is less than a small amount. now if you look at the combine results (46K) the difference's confidence is 0.007! To reach something like 0.002 you'll need 12 times more games! so it's not very practical.

It is very difficult to proove or disproove that the CI are under (or over) evaluated. I did the test on one position : http://www.bgonline.org/forums/webbbs_config.pl?frames;read=81963 and did not see any problem, but that's ONE position, so it does not proove much.

When i got the time, i can try to do the same on an opening move such as that one, i probably will not make the analyze bust just publish the Raw data (game per game results)

Messages In This Thread

• Opening 52 at 4 away 4 away
  Stick -- Saturday, 15 January 2011, at 9:00 p.m.
  • Opening 52 at 4 away 4 away
    neilkaz -- Saturday, 15 January 2011, at 9:37 p.m.
  • Stat sig
    Timothy Chow -- Sunday, 16 January 2011, at 2:35 p.m.
  • Opening 52 at 4 away 4 away - extended
    Stick -- Monday, 17 January 2011, at 9:55 a.m.
    • Nine 5k results (report)
      Nack Ballard -- Monday, 17 January 2011, at 3:20 p.m.
      • Nine 5k results (report) - that why people need to check SI
        eXtreme Gammon -- Monday, 17 January 2011, at 6:00 p.m.
        • Nine 5k results (report) - that why people need to check SI
          Stick -- Monday, 17 January 2011, at 6:15 p.m.
          • Nine 5k results (report) - that why people need to check SI
            eXtreme Gammon -- Monday, 17 January 2011, at 6:56 p.m.
            • Nine 5k results (report) - that why people need to check SI
              Nack Ballard -- Monday, 17 January 2011, at 10:59 p.m.
            • Nine 5k results (report) - that why people need to check SI
              Neil Robins -- Tuesday, 18 January 2011, at 9:34 p.m.
              • Nine 5k results (report) - that why people need to check SI
                eXtreme Gammon -- Wednesday, 19 January 2011, at 4:47 a.m.
                • Nine 5k results (report) - that why people need to check SI
                  Timothy Chow -- Wednesday, 19 January 2011, at 2:33 p.m.
              • Nine 5k results (report) - that why people need to check SI
                Daniel Murphy -- Wednesday, 19 January 2011, at 9:10 p.m.
        • You mean CI *NM*
          neilkaz -- Monday, 17 January 2011, at 6:41 p.m.
          • You mean CI - I did indeed *NM*
            eXtreme Gammon -- Monday, 17 January 2011, at 6:56 p.m.
        • Nine 5k results (report) - that why people need to check SI
          Timothy Chow -- Tuesday, 18 January 2011, at 5:30 p.m.
      • Nine 5k results (report)
        Daniel Murphy -- Monday, 17 January 2011, at 9:43 p.m.