Feedback Request: Starting a Movie Blog

Introduce yourself to the community or chat with other users about whatever is on your mind
CarsonWid
Posts: 50
3424 Ratings
Your TCI: na
Joined: Sun Sep 13, 2009 5:56 pm

Feedback Request: Starting a Movie Blog

Post by CarsonWid »

Hey guys,

So I decided to start up a blog recently to combine two of my passions: movies and data analytics

I've made a couple posts and was wondering if anyone could give some feedback/ideas on what works and doesn't work and maybe what could be added:

Link: http://movienumberz.tumblr.com/post/124 ... cide-squad

I'm trying to find reliable data that can be used to predict how good a movie is, but also trying to make it more visual and easy to digest.

Any feedback is appreciated,

Thanks!

CarsonWid
Posts: 50
3424 Ratings
Your TCI: na
Joined: Sun Sep 13, 2009 5:56 pm

Re: Feedback Request: Starting a Movie Blog

Post by CarsonWid »

If this request seems to spam-my let me know. I'll just remove the link and include description instead.

I don't want to give awareness of a blog that hasn't even started yet, just want to start a conversation about what factors could be used to predict whether a movie will be good or not.

ribcage
Posts: 195
4173 Ratings
Your TCI: na
Joined: Thu Aug 18, 2011 10:13 pm

Re: Feedback Request: Starting a Movie Blog

Post by ribcage »

i would say that it's fascinating, and i can't think of anything to add into your number crunching. i'll also say that i'm mathematically a total moron, so maybe someone else should come along and comment on your equations instead of me.

if this is for math/movie geeks it's certainly spot on(i think, again i'm a mathemidiot), but if you're going to go broader audience it might do to find a way to do a run down in layman terms for each article. not necessarily explaining your equations in every single article, you already posted a detailed explanation, but something to prevent it from being too dry for your average movie-watcher. i hate to mention a problem without suggesting solutions, but since i'm so terribly weak at math i'm afraid i can't elaborate on this.

continuing with the "dry" aspect, it could also use a little more color throughout. maybe a line graph to accompany the trend numbers? although with so many players that might begin to look cluttered. i do really like the "can they make this type of movie" box you made.

as a final note, i would say that actors/writers/directors who have worked together before certainly add something to a film, but does that make the reverse true that if they haven't it takes something away? is this something you researched before penalizing suicide squad for having cast/crew that haven't previously worked together?

in any case, i didn't find it to be spam at all and i do find it very interesting. would love to see it in it's finished form and follow it...so long as your numbers add up once the films come out! i assume you've applied your equations to previously released films, maybe that's something to add as examples in your explaining article until some of your predictions have come through.

ShogunRua
Posts: 3449
0 Ratings
Your TCI: na
Joined: Sun May 17, 2009 3:18 am

Re: Feedback Request: Starting a Movie Blog

Post by ShogunRua »

As a mathematician by education from Caltech, I do enjoy people using the subject to investigate all elements of our world.

However, your current method is very crude and inaccurate. I recall Ryan Kavanaugh stating in an interview that they do a Monte Carlo simulation with well over 50 different variables and greenlight a picture if it passes the 75 percent confidence interval for breaking even.

Okay, that's not very interesting from a mathematical perspective (I think a decently intelligent 16 year-old could code that up no problem), but it works decently enough for their aims.

I guess my question is; what do you want to get out of this blog? If you want to fine-tune the algorithm and do basic statistical comparisons, there are a wealth of jobs that allow you to do that full-time and collect a good paycheck. Do some kind of research and write a paper? I don't think there is anything there. Again, the math is rather boring/simple.

ShogunRua
Posts: 3449
0 Ratings
Your TCI: na
Joined: Sun May 17, 2009 3:18 am

Re: Feedback Request: Starting a Movie Blog

Post by ShogunRua »

As a general note, you're also massively, massively overrating the influence of actors. Most of them have no effect on a movie in the usual sense; they do a passable job in roles where they're not even expected to do anything difficult. (Particularly true in the superhero genre)

The best indication actors can give us isn't through their abilities so much as how discerning they are about scripts. For instance, with Daniel Day-Lewis, his movies aren't just as good because he is phenomenal, but because he only chooses those that have an outstanding script and director attached. He is still occasionally wrong, it's a strong indication of quality.

CarsonWid
Posts: 50
3424 Ratings
Your TCI: na
Joined: Sun Sep 13, 2009 5:56 pm

Re: Feedback Request: Starting a Movie Blog

Post by CarsonWid »

ShogunRua, you make some good points.

In regards to studios and their 50 variable equations, I think most of that has to do with the financial success of a movie and very little to do with the critical success. I had originally started this as a prediction of financial success but I didn't find it very interesting. I had various seasonal indices and factors in it but I wasn't getting good results and as mentioned before, wasn't really what I was trying to do.

I have a limit on what data is available to me, so as much as I'd like to add more variables I also have to consider that this is just a for-fun after work project of mine. I've thought of things such as release date, budget, production company, etc. but none of these really held a strong correlation to the critical success of the movie.

I'd love to have a job running analytics of this nature but alas I do not have it, nor do I have the tools or databases to run this at a super effective level.

My objective for this blog is to show data and analytics at a high level in a simplified way that people who don't like math can digest it. The objective is not to 100% accurately predict a movie's score, because I think that is impossible. Rather, I would like to give a reasonable expectation as to how good a movie will be instead of us giving in to the marketing hype machines that we always get.

It should obviously focus more on Hollywood blockbusters than smaller films. I could make a prediction that The Revenant will be awesome, but I don't think it makes for the best blog (although I would like to sprinkle those types of movies throughout).

So, final point/question: What easily-accessible variables do you think would be most relevant when predicting a movie's critical success?

mattorama12
Posts: 887
3086 Ratings
Your TCI: na
Joined: Mon Feb 13, 2012 3:05 am

Re: Feedback Request: Starting a Movie Blog

Post by mattorama12 »

I'm a fan of the idea and have some comments for you:

First,
Criticker (Somewhat snobby movie fans who wish they were critics but aren’t)


How dare you!

Next,
I take up to 20 of an individual’s past movies and create a simple linear equation (y = mx + b) to describe those scores and forecast what the 21st movie in that equation would be. I do not take into consideration how near or far apart each of these movies were made.


Is x the number of film here? To me, I think you're overestimating "trend" here. For example, an actor whose first movie was a 1, 2nd was a 2, and 3rd was a 3. Do we really think his fourth is going to be a 4? Isn't it more likely that his fourth would be a 2, as that's his average? Maybe some sort of weighted average with more recent films counting for more would be useful here. That way you could account for trends without over-representing a trend. [Edit, I didn't think this would be much use practically, as I assumed the "trend" would be relatively minimal. Looking again at the posts you have, it seems that "trend" is almost completely negligible anyway as most appear to be less than +/- 2 out of a scale of 100]

But who knows without testing it? And that brings me to my next point.

I'm assuming you did this already, but maybe not. Why not show some of the results from tests done on movies that are already out? Of course there's no way for me to know whether you pre-selected ones that fit your equations well, but it'd be interesting to see. Here's an idea--take the 15th highest grossing film from 2012, 2013, and 2014 and run an analysis on them. See how close your prediction is to the actual result. Ideally, you could even get all the data in a spreadsheet (the more movies the better), and tweak your weighting and coefficients to see if you can better match the data.

Of course you're never going to have this be super predictable, both because of the lack of information available to you and the complexity of predicting such things. However, I think that might get you closer to it than you are now.

A final though, the design of the site is not very pretty. It's very, very basic and could use some sprucing up. I'm not really the guy to talk to for help there, though.

paulofilmo
Posts: 2586
2428 Ratings
Your TCI: na
Joined: Mon Apr 07, 2008 3:40 pm

Re: Feedback Request: Starting a Movie Blog

Post by paulofilmo »

I guess if a website told me Lubezki, Iñárritu, Jack Fisk, Caprio, Tom Hardy were doing stuff together, I would pay attention (rather than via the whim of watching a trailer).

But then I have no commercial value

Image

Also: Cool!

crack on son

paulofilmo
Posts: 2586
2428 Ratings
Your TCI: na
Joined: Mon Apr 07, 2008 3:40 pm

Re: Feedback Request: Starting a Movie Blog

Post by paulofilmo »

paulofilmo wrote:Lubezki, Iñárritu, Jack Fisk . . . were doing stuff together


And by association, you could recommend films by Malick, Cuarón, PT Anderson. Films w/music from Jonny Greenwood. Films edited by Dylan Tichenor, , ever more tenuously.

Image

(click to enlarge)

Butterick’s Practical Typography
Last edited by paulofilmo on Sun Jul 19, 2015 11:12 am, edited 2 times in total.

CarsonWid
Posts: 50
3424 Ratings
Your TCI: na
Joined: Sun Sep 13, 2009 5:56 pm

Re: Feedback Request: Starting a Movie Blog

Post by CarsonWid »

In regards to testing, I took a sample of 2013 and 2014 movies to figure out which positions most correlated to the success or non-success of their movies. The director and writer more often contributed to success, and actors were for less note-worthy.

I'm thinking of adding cinematographers, production designers, music, film editing, art designers, and others. The information is easily accessible on IMDb. Visual effects, sound, and some other have many people involved so I think that would be tough to run the numbers on.

Of course more testing will need to take place to see if these people actually factor in that much.

Post Reply