CloudHype Posted June 16, 2007 Share Posted June 16, 2007 (edited) Mystre would like to thank Jeff and John for a smooth running first night. For the first time since season 4, no complaint forms where used on opening day! :P Here's to a smooth season! Edited June 16, 2007 by CloudHype Quote Link to comment Share on other sites More sharing options...
JohnD Posted June 16, 2007 Share Posted June 16, 2007 I wonder what the "unknown" factor is in the generated scores. If I just went with straight generated scores you'd have thought all we did was have two sets and play whole notes (mind you they would have been beautiful and in tuned whole notes nontheless). Oh trust me, I'll find out. ;) :OT: does anyone here also dislikes bittorret?? The reason for Boston's score is that the initial scores were generated from the average of the corps' FIRST FIVE scores of 2006. Given that Boston didn't enter the tour until sometime after the other corps, their average was higher than you would expect. You'll see that settle-down once we start having consistent live scores. The method being used to generate scores is not a big secret .... its just impossible to predict the results. All 'generated scores' are based on the corps' last score recorded (real or generated) - multiplied by a random number within a range of 95 - 105%. We will tighten this range throughout the season to reduce the variance from the previous score. The wider range at the beginning of the season better simulates the variances in judging, the number of changes being made to show designs, and other factors. As the season progresses, there is often more consistency / less variation ... so the range is tightened to the point of 99%-101%. We believe this element of randomness provides enough 'uncertainty' to make the game interesting, as there is NO way to predict the final outcome. So - unless you're able to guess the random numbers that Jeff's computer generates (and re-generates EVERY time he makes an update to the sheet) ... you won't be able to calculate or predict the generated scores - other than to know that they'll fall into that range. Obviously, 'live' scores are un-touched. So now you know - the rest of the story. ($1 to Paul Harvey ..... gooood-day) Good luck, everyone! -john Quote Link to comment Share on other sites More sharing options...
JohnD Posted June 16, 2007 Share Posted June 16, 2007 Mystre would like to thank Jeff and John for a smooth running first night. For the first time since season 4, no complaint forms where used on opening day! :PHere's to a smooth season! Thank you for the kind words ... Like Matt last season, Jeff has done an incredile job in learning the new system and processes. We don't have all of the automations I had hoped to have in-place yet, but the few new things that we've actually been able to get together should make things a LOT easier to execute and reduce the human-factor for error significantly. The process of transcribing all of those caption changes was a yeoman's task - which our new Caption Selection Editor is making easier ... hopefully for our members as-well as our administrator. Reducing your "complaint forms" to ZERO is my goal. B) -john Quote Link to comment Share on other sites More sharing options...
sburstall Posted June 16, 2007 Share Posted June 16, 2007 (edited) The reason for Boston's score is that the initial scores were generated from the average of the corps' FIRST FIVE scores of 2006. Given that Boston didn't enter the tour until sometime after the other corps, their average was higher than you would expect. You'll see that settle-down once we start having consistent live scores.The method being used to generate scores is not a big secret .... its just impossible to predict the results. All 'generated scores' are based on the corps' last score recorded (real or generated) - multiplied by a random number within a range of 95 - 105%. We will tighten this range throughout the season to reduce the variance from the previous score. The wider range at the beginning of the season better simulates the variances in judging, the number of changes being made to show designs, and other factors. As the season progresses, there is often more consistency / less variation ... so the range is tightened to the point of 99%-101%. We believe this element of randomness provides enough 'uncertainty' to make the game interesting, as there is NO way to predict the final outcome. So - unless you're able to guess the random numbers that Jeff's computer generates (and re-generates EVERY time he makes an update to the sheet) ... you won't be able to calculate or predict the generated scores - other than to know that they'll fall into that range. Obviously, 'live' scores are un-touched. So now you know - the rest of the story. ($1 to Paul Harvey ..... gooood-day) Good luck, everyone! -john hehehe...remember this is me we're talking about. ;) Edited June 16, 2007 by sburstall Quote Link to comment Share on other sites More sharing options...
CloudHype Posted June 16, 2007 Share Posted June 16, 2007 Reducing your "complaint forms" to ZERO is my goal. B) -john I'll be a happy director :) Thanks for all the work. Quote Link to comment Share on other sites More sharing options...
StarOrg Posted June 16, 2007 Share Posted June 16, 2007 Not a bad night for Star of Jupiter in JAX. Star of Hawaii starting off where they left off last season. Quote Link to comment Share on other sites More sharing options...
CloudHype Posted June 16, 2007 Share Posted June 16, 2007 The reason for Boston's score is that the initial scores were generated from the average of the corps' FIRST FIVE scores of 2006. Given that Boston didn't enter the tour until sometime after the other corps, their average was higher than you would expect. You'll see that settle-down once we start having consistent live scores.The method being used to generate scores is not a big secret .... its just impossible to predict the results. All 'generated scores' are based on the corps' last score recorded (real or generated) - multiplied by a random number within a range of 95 - 105%. We will tighten this range throughout the season to reduce the variance from the previous score. The wider range at the beginning of the season better simulates the variances in judging, the number of changes being made to show designs, and other factors. As the season progresses, there is often more consistency / less variation ... so the range is tightened to the point of 99%-101%. We believe this element of randomness provides enough 'uncertainty' to make the game interesting, as there is NO way to predict the final outcome. So - unless you're able to guess the random numbers that Jeff's computer generates (and re-generates EVERY time he makes an update to the sheet) ... you won't be able to calculate or predict the generated scores - other than to know that they'll fall into that range. Obviously, 'live' scores are un-touched. So now you know - the rest of the story. ($1 to Paul Harvey ..... gooood-day) Good luck, everyone! -john This is similer to Dr. Jeffs system but not exactly. Last season we almost saw proof that with the "perfect" captions, you might not win. It was obvious that Matt and I had the same idea for our Finals set (we've both new the highest "raw" score). Rather then try to beat him in with picks and risk a possible .5 drop in score if I missed (and 4th place), I decided to send in a set that had a "chance" (ment I changed my plan for semis) to win, if Matt's generated captions droped and mine stayed the course... I win. Didn't happen though but I got a very second insteed of 4th :) Moral of the story for you rookies... You don't have to have the "best" or "perfect" set to have a chance to win it all, you just need to be REALLY close. Quote Link to comment Share on other sites More sharing options...
sburstall Posted June 16, 2007 Share Posted June 16, 2007 (edited) This is similer to Dr. Jeffs system but not exactly. Not really. I have the code from Dr. Jeff and there was nothing in it that had a random in it (I may be able to program in PERL but I could read it). It was strictly a line fit with the data. He even stated this in the FAQ. The randomness the John put into this program seems to be in similar to the weighed factor I used for the generated scores. Edited June 16, 2007 by sburstall Quote Link to comment Share on other sites More sharing options...
StarOrg Posted June 16, 2007 Share Posted June 16, 2007 Live season and Off season at totally different ballgames. Everyone has a great chance of making Finals and doing well in Live season. Just stay involved and with luck the scores swing your way. That what makes the Live seasons so much fun. Quote Link to comment Share on other sites More sharing options...
JohnD Posted June 16, 2007 Share Posted June 16, 2007 This is similer to Dr. Jeffs system but not exactly. Last season we almost saw proof that with the "perfect" captions, you might not win. It was obvious that Matt and I had the same idea for our Finals set (we've both new the highest "raw" score). Rather then try to beat him in with picks and risk a possible .5 drop in score if I missed (and 4th place), I decided to send in a set that had a "chance" (ment I changed my plan for semis) to win, if Matt's generated captions droped and mine stayed the course... I win. Didn't happen though but I got a very second insteed of 4th :)Moral of the story for you rookies... You don't have to have the "best" or "perfect" set to have a chance to win it all, you just need to be REALLY close. Not really. I have the code from Dr. Jeff and there was nothing in it that had a random in it (I may be able to program in PERL but I could read it). It was strictly a line fit with the data. He even stated this in the FAQ. The randomness the John put into this program seems to be in similar to the weighed factor I used for the generated scores. I reviewed Dr. Jeff's code and had his permission to use it - although, I wound-up doing things a bit differently. It appeared to me that, especially in the 'off-season', with a good understanding of the linear function and access to the historic data, a person with enough time and desire could calculate the 'perfect' combination of captions. Further, I had some feedback that the math involved was a bit confusing to some that were trying to validate their scores. Using a random number method (within a well-defined range of variance) solves a number of problems (pun not intended). It makes things a bit easier to understand - "the value is going to fall somewhere in this range, but other than that - you can't predict or calculate what its going to be" ..... and it removes the ability for a 'numbers-shark' (no disrespect meant to the statistical gurus in the crowd) to calculate the winning combination. There will always be the slightest variance between scores that include a 'generated' value .... so even the 'best' caption selection set has the possibility of not winning. Our cost caps will demand that generated scores (for those corps not performing on Finals night) are a part of every DCPI corps score. In the end, its not all that different from the real field of competition. Within the top X corps ... on any given night - anyone can win. Hopefully, it all makes for an exciting and entertaining game for everyone. -john Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.