Some mathematical ridiculosities (yes, I just made that word up in case you were wondering):
Hey - this is an edit - the previous post was NOT posted when I started writing this, thus, displaying the obvious superiority of my model.
Since I am a firm believer in doing proper DD, I have taken the time to analyze the FY 97 estimates submited by participants in this thread. Please note that this study has not been peer reviewed and, thus, should be regarded with the same disdain as we show towards all other bogus data on this thread (the numbers are real though, I didn't make this crap up).
I performed a regression analysis on the quarterly estimates using three x-variables attempting to determine the one y-variable, namely the quarterly earnings prediction. The x-variables were as follows:
1st x variable: date of submission of estimate
2nd x variable: time of submission of estimate
3rd x variable: a bogus number between from 1 to 3 that categorizes the time of the estimate based on the following: 8am-5pm = 1, 5pm-12midnight = 2, 12 midnight-8am = 3. I know we don't all live in the same time zone, so if you have a problem with my methodologies then go do your own damn study.
Basically what this predicts is that anyone posting after this will most likely have an estimate of $1.00 or $1.01for FY97.
Some conclusions:
1.The later the date you make your guess the lower your estimate will be.
2.If you post during business hours you'll estimate lower than you do if you post after hours and if you're up between midnight and eight in the morning you'll estimate even higher. (Of course we've only had one post after midnight and before 8 am so put even less weight in this than the other stuff)
3.Within the three categories above, the later you post the higher you'll estimate. So someone at 4:59pm will estimate higher than someone at 8:01am even though they're in the same group (8-5).
Well, I hope all of our guesses are lower than the actuals - good luck everyone,
Gary
P.S. for those of you who are math nuts here's some info about the regression for you:
Constant: 2.721866
St Err of Y - est: .015182
R Squared: .187927 (so yes, this model is inefficient)
Observations: 21
Degrees of Freedom: 17
X - coefficients & standards of error:
1st: -7.6E-05, .000227
2nd: .10061, .097928
3rd: .064637, .045101
Footnote: No I did not regress each variable individually against the Y to determine inclusion in the model. I threw it all together and out this came. Nor was there any error correction methodology used. |
