How well can the Hubdub community predict the behaviour of the world markets next week?
Well done all. One point to the Hubdubbers. Null point to Walt Mossberg who told us you were all a bunch of faceless nobodies (direct quote).
Background:>
I asked the following 5 questions:
http://www.hubdub.com/e/Market/How_many_of_the_following_12_Indices_will_close_UP_on_Mon_Mar_3_3319/view
(See Settlement Rules for links to other 4)
For each question, we score it by subtracting the average odds of the wrong answers from the odds of the right answer. Because the odds add to 100 and there are 4 wrong answers out of 5 choices, you can derive that the score S = 5/4 *cOdds -25 (where cOdds is the final odds of the correct answer). If the cOdds is 0% then S = -25, if 20% (as if we all threw a five-sided dice to make a pick) then S=0, and if 100% then S=100.
Take the average of these 5 scores (between -25 and 100, as is the average). Use the odds from the graph and the formula for S above. What's this value going to be?
Settlement details:
Remaining 4 questions:
http://www.hubdub.com/e/Market/How_many_of_the_following_12_Indices_will_close_UP_on_Tue_Mar_4_3320/view
http://www.hubdub.com/e/Market/How_many_of_the_following_12_Indices_will_close_UP_on_Wed_Mar_5_3321/view
http://www.hubdub.com/e/Market/How_many_of_the_following_12_Indices_will_close_UP_on_Thurs_Mar_6_3322/view
http://www.hubdub.com/e/Market/How_many_of_the_following_12_Indices_will_close_UP_on_Fri_Mar_7_3323/view
When all 5 questions have been settled, the correctness score can be taken for each as follows: take the final odds "F" of the correct answer from the bar graph. Score "S" is given by S = (5/4)*F - 25
Finally take the average of these five scores. The scores and their average will each be in the range -25.00 to 100.00 (The scores and averages can be expressed EXACTLY to two decimal places.) The average is used to answer THIS question.
As judged by the HD editor/admin using above method.
Settled
|
|
Random or worse (Avg Score: -25.00 - 0.00) |
|
||
|
|
Just better than random (Avg Score: 0.01 - 15.00) |
|
||
|
|
Much better than random (Avg Score: 15.01 - 50.00) |
|
||
|
|
We must be collective geniuses (Avg Score: greater than 50.00) |
|
Suspend date: Fri 7th Mar 12pm PST Settlement date: Fri 7th Mar 2:32pm PSTPrediction cut-off: Predictions on this question after Fri 7th Mar 12pm PST have been voided because they were made after the question could be settled
Initial likelihoods: Random or worse (Avg Score: -25.00 - 0.00): 20%, Just better than random (Avg Score: 0.01 - 15.00): 30%, Much better than random (Avg Score: 15.01 - 50.00): 30%, We must be collective geniuses (Avg Score: greater than 50.00): 20%
Action history:
Day 1: Correct odds:80% Score:75
Day 2: Correct odds:80% Score:75
Day 3: Correct odds:53% Score:41.25
Day 4: Correct odds:74% Score:67.5
Day 5: Correct odds:51% Score:38.75
Sum of 5 scores: 297.5
Avg. Score:59.5 so "collective geniuses" (greater than 50) wins.
Suspend date: Fri 7th Mar 12pm PST Settlement date: Fri 7th Mar 2:32pm PSTPrediction cut-off: Predictions on this question after Fri 7th Mar 12pm PST have been voided because they were made after the question could be settled
more info...
Predictions (87)
87 predictions
Comments (16)
What is Hubdub?
Hubdub makes news more exciting by letting you stake virtual dollars on the outcomes of real running news stories.
Join now or learn moreRelated News
This news is selected automatically based on the question, its background, options and tags
Related tags
New in Technology » Hubdub
More questions by
59 settled, 5 voided




= )
I'm interested in feedback on this method of "scoring" a question. I think it might be useful to have a standard way to measure our performance as predictors, not just for derivative questions, but as a way of rating past performance.. Settled questions could be given a score, and events like the Olympics, Oscars, etc, could be assessed on average.
I like the formula here for its relative simplicity ;-) and the way it scores the expected value of random choosing at zero. But it is only symmetrical about zero for the 2-choice questions. Also, it doesn't really distinguish between "flip a coin" type questions and one's where there are clear favourite choices (and long shots).
If we could work toward a standard scoring system, then we'd have a standard approach for HubDerivatives(c), which would make questions like this one easier to understand, and simpler to settle.
Just don't start marketing and junk bonds ok. ;)
Having a standard scoring system - although it sounds like a good idea, imposing it is another matter
Inquiring minds want to know. If you don't respond in these comments in a day or two I might just be tempted to turn this into a question on Hubdub ... "Is InfernalMachine _really_ a professional poet? As reported by HD user InfernalMachine". Just kidding! (about the HD question that is)
Day 2: Correct odds:80% Score:75
Day 3: Correct odds:53% Score:41.25
Sum of Score over 4 days:258.75
Please log in or join to add a comment