HOLLYWOOD STOCK EXCHANGE
THE ENTERTAINMENT MARKET
THE ENTERTAINMENT MARKET
Some initial thoughts...
Theatre counts are a much smaller factor than I'd have thought for opening weekend, but I guess the predictors take theatre #s into account when making their predictions.
There is a lot of missing data for RS, and the back filled predictions - well, let's just say that at first glance they are a lot more accurate than I'd expect RS's predictions to be. And as it looks like we won't be getting RS predictions again, I suggest removing RS from your model. (I've seriously thought about deleting them from the SS but, history)
Relatedly, it would be interesting to run a regression using only two or three predictors and see if you can achieve similar accuracy. I'd suggest trying MTC, LAT and EW
one thing to be aware of with the data is that it's not always objective. Most data sources that rely on tracking - MTC, LAT, THR, Variety etc will give rough estimates, like "high 20s"or "mid-teens" or "above 30", rather than a reliable number. So in entering the predictions I've had to exercise some judgment by picking a mid-point or translating a phrase into a number.
Here are the results containing my regression model's predictions on BO, using HSXSanity Data from pre-release forecasts. Feb 25, 23:08
Wow. What happens when smart dudes get together and share #s is awesome. Thanks elchan and Mr Moore! {nm} Feb 26, 00:09
WOWZER - A Tip O' The Hat. Thanks for all of your Contributions. {nm} Feb 26, 04:03
Some initial thoughts... Feb 26, 04:34
Actually, the extrapolated RS data is not all that accurate. Mean error rate for RS is 0.45 for, compared to ,,, Feb 26, 07:18
Here are the result of a few runs: Feb 26, 08:09
Nate Silver's missing out on all the fun :) {nm} Feb 26, 08:56
