Predictive Oscars Algorithm

[JEinOKCMember Joined: Jan 16th, 2016 Topics: 1 Posts: 5 View Predictions]

#214503

First, let me start by acknowledging this is my first post and it’s usually not proper ettiquite to jump right in on a new forum and start a topic. So please accept my apologies.

I am not a expert and am pretty new to the film prediction game. I have, however, jumped head-first into an attempt to do just that.  A little background – I’ve been ranking College Football for several years using a mathematical formula. In an attempt to branch out, I attempted last year to simulate the same process with the Best Picture award and the results were at least interesting, but it was mostly just friends and word of mouth.

This year, the website FiveThirtyEight.com asked for submissions of people who had creative ways to predict the Oscars. My model was selected as one of the 8 and I my story was published earlier today.

Why I’m here is that my model is a little different from others because it requires audience participation and a quick google search for groups that know their stuff showed that GoldDerby has a pretty intense group of people with this exact interest.

I’d be humbled if this is something anyone who reads this is interesting in helping with. There is no prize for me to win – only my pride is on the line.

The direct link to the ratings survey is: http://jameseng.land/Oscars

The link to my profile on FiveThirtyEight is: http://fivethirtyeight.com/features/can-you-fake-the-academy-to-predict-the-oscars/ 

I signed up a couple of weeks ago with the plan to ask for help, but decided it was unfair to try to talk you all into clicking a strange link of mine without any proof that it wasn’t some sort of ruse. So feel free to read all the details on FiveThirtyEight before jumping over

thanks!

-James

 

Reply

[PJ EdwardsParticipant Joined: Jun 2nd, 2011 Topics: 9 Posts: 2363 View Predictions]

#214505

I filled it out. But I think its an interesting idea but I strongly believe that you can’t use math to predict Oscars. It fails to pick up on perception and hype which is more important then simply counting critic and guild wins. Like it’s crazy that some of other formulas suggest that Room and The Martian are somehow front runners.,

ReplyCopy URL

[babypookParticipant Joined: Nov 4th, 2010 Topics: 108 Posts: 14465 View Predictions]

#214506

^Agreed.

You cant predict sports games this way either. Rankings are nice, until playoff time.

Very intriguing post! I’d like to check it out.

ReplyCopy URL

[MrGoodWoodParticipant Joined: Dec 14th, 2013 Topics: 4 Posts: 378 View Predictions]

#214507

That’s actually a very good idea. 
The thing is, the demographic of people who take your survey will be extremely different than the oscar voters. Still, it will be very interesting to see the results with this model. Good luck!

ReplyCopy URL

[JEinOKCMember Joined: Jan 16th, 2016 Topics: 1 Posts: 5 View Predictions]

#214508

Thanks for the replies so far.  You guys are right too about not relying solely on it to even predict football.  My one advantage I think is also my big assumption. This actually allows for human interaction instead of just mining tons of data, which I think can show some human trends.  I’m also running on the idea that if you’re seeing lots of the same films that are nominated,  you could at least fit close to the profile.  

You’d want to always use this as just assistance to your own gut when making a final decision,  but it is really fun to code for weeks,  hit a button and actually get an answer that makes a bit of sense 

ReplyCopy URL

[benuttyKeymaster Joined: Jul 3rd, 2012 Topics: 34 Posts: 580 View Predictions]

#214509

I don’t understand how this is any different than asking people to rank 8 films 1 through 8 and just automatically determining that whatever is ranked 1 won 7 of 7 matchups, 2 won 6 of 7 matchups, 3 won 5 of 7 matchups, where its losses are to the films ranked above it. Sounds like a complicated way of getting a result I could have answered in one question, which is the question the Academy already asks of its voters. 

ReplyCopy URL

[JEinOKCMember Joined: Jan 16th, 2016 Topics: 1 Posts: 5 View Predictions]

#214510

mrgoodwood – The funny thing is, I actually don’t think the demographics are too far off. Lets just say that fivethirtyeight has a “specific” type that reads and it’s got the same skew as what most say about the typical member of the academy… just a bit younger

benutty – I think you’re right that this could be modified exactly like what you said to make it simpler on folks who’ve seen a lot of the films. To be quite honest, this was a quick weekend project that I did a year ago that is being re-booted and there are areas where it can improve. That being said, the average voter on the site this year has only picked 18 games over 6 categories. Seeing it in the current format isn’t such a huge problem to someone like that who’s probably seen 2 or 3 of them, but I can imagine folks who frequent this site might find it a bit redundant

—–

Now that things are getting closer, I wanted to check back in and actually try my hand at setting predictions for the Oscars based off of what the current results are now that they’re getting closer. This looks like it can be a pretty fun way to keep up for other awards as well

ReplyCopy URL

[babypookParticipant Joined: Nov 4th, 2010 Topics: 108 Posts: 14465 View Predictions]

#214511

mrgoodwood – The funny thing is, I actually don’t think the demographics are too far off. Lets just say that fivethirtyeight has a “specific” type that reads and it’s got the same skew as what most say about the typical member of the academy… just a bit younger

benutty – I think you’re right that this could be modified exactly like what you said to make it simpler on folks who’ve seen a lot of the films. To be quite honest, this was a quick weekend project that I did a year ago that is being re-booted and there are areas where it can improve. That being said, the average voter on the site this year has only picked 18 games over 6 categories. Seeing it in the current format isn’t such a huge problem to someone like that who’s probably seen 2 or 3 of them, but I can imagine folks who frequent this site might find it a bit redundant

—–

Now that things are getting closer, I wanted to check back in and actually try my hand at setting predictions for the Oscars based off of what the current results are now that they’re getting closer. This looks like it can be a pretty fun way to keep up for other awards as well

I agree we’re not that different from the voting body, but I’d be tempted to change the “bit” to, A LOT.

ReplyCopy URL

[JEinOKCMember Joined: Jan 16th, 2016 Topics: 1 Posts: 5 View Predictions]

#214512

[quote=”TWJamesEngland”]

mrgoodwood – The funny thing is, I actually don’t think the demographics are too far off. Lets just say that fivethirtyeight has a “specific” type that reads and it’s got the same skew as what most say about the typical member of the academy… just a bit younger

benutty – I think you’re right that this could be modified exactly like what you said to make it simpler on folks who’ve seen a lot of the films. To be quite honest, this was a quick weekend project that I did a year ago that is being re-booted and there are areas where it can improve. That being said, the average voter on the site this year has only picked 18 games over 6 categories. Seeing it in the current format isn’t such a huge problem to someone like that who’s probably seen 2 or 3 of them, but I can imagine folks who frequent this site might find it a bit redundant

—–

Now that things are getting closer, I wanted to check back in and actually try my hand at setting predictions for the Oscars based off of what the current results are now that they’re getting closer. This looks like it can be a pretty fun way to keep up for other awards as well

I agree we’re not that different from the voting body, but I’d be tempted to change the “bit” to, A LOT.

[/quote]

and I’d be tempted to agree with you

ReplyCopy URL

[JEinOKCMember Joined: Jan 16th, 2016 Topics: 1 Posts: 5 View Predictions]

#214513

Turned out alright. Biggest miss was director, but this model was the only one in the feature to select Spotlight. The top priority for next year will be to get the sample of voters a little closer to the academy, but no complaints tonight. Thanks again for your interest and participation

ReplyCopy URL