In partnership with CBSSports.com
Online Now 2401
Online now 1801 Record: 18710 (2/25/2012)
We aren't just committed to college football; we're early enrolling in it.
Where the madness isn't just in March.
You have no favorite boards.
The most viewed topics.
The most replied to topics.
The most up-voted topics.
The most down-voted topics.
The most up-voted posters.
The most down-voted posters.
The most followed posters.
Predicting within a week of a commitment could present issues, since there's plenty of times when recruits let the analysts know beforehand. I think everyone should have to stick with their first call.
Many people would disagree that predicting an outcome before it happens is of "marginal value." We will get the protocols worked out, but the primary purpose of this product is to tell/predict outcomes. Hopefully, the prediction will occur well in advance and not hours before. We are looking into a lockdown protocol.
Even in a case like Shaun Hamilton where your claim holds some water it is working. Currently in an article on AL.com, two of the three state reporters are picking Hamilton to Auburn. While the Crystal Ball is 100% Alabama. I agree that many came in late in the process, but the CB does provide clarification prior to an announcement. It is doing its job. We will create transparency and a formula that only rewards predictions for the analyst with an acceptable prior prediction. From my perspective, Crystal Ball and Analyst Rankings/Value are two different products.
Again we will setup a point system based on advance warnings. Currently posting the date and soon a time stamp. We will be incredibly transparent. It will be very clear who is predicting in advance and who comes in later in the process. Please give us some time to work it all out.
This post was edited by Lucky 12 months ago
247Sports and BOL updates: http://twitter.com/sbterry247
think it would be cool to list all of the predictions for each school by percentage, not just those above x%.
I actually love the way it worked with SDH. Since everyone basically changed to bama it told the users basically where he is going. Made it obvious when we got a commit from Davis this morning. When you guys implement some kind of rating for the analyst thats when you need to put a timer on the prediction. Almost like those trivia games at the local bars that have four answers and if you pick the correct answer when the question first shows it's the max amount of point. Over time the point mas trickles down until one answer remains.
Other than putting it into team thing, I'm not sure what else you guys could do. I think it is a great thing for this site and really really enjoy it. I'm sure most other users loved it too.
Kind of agree with this.
History of predictions for a player would be interesting, maybe a chart like the history of a player's rating.
Also, I think someone else mentioned this with different wording, but using confidence points/percentage might be a good add.
100% agree. Noah was talking about that this morning. Definitely plan on adding a timeline of the predictions.
Noah and I are going to meet this afternoon about adding some enhancements to make it better. These are all very helpful.
The Crystal Ball is not perfect and never will be, but we can get it close.
The list will in all likelihood do this. We rarely cut things off.
Follow me at: http://twitter.com/NoahStanley247 or go behind the scenes of 247Sports at http://noahstanley.com/
The ability to click on an analyst and see all of their predictions. Awesome feature though.
This post was edited by PSUfaninOnt 12 months ago
Pretty sure we have it. Click on the analysts name on the Crystal Ball and or board posts (then click user predictions if your go this route).
Let me know if that doesn't work for you.
I know this isn't about the crystal ball but thought about a feature that I would think would be pretty easy for you guys to implement. You guys have a list of all the official visits for a team like http://auburn.247sports.com/Season/2014-Football/Visits
Most writers already put in when a prospect has an unofficial visit in the timeline. Is there anyway so when a writer does that it can have a list of unofficial s too instead of having to looke at the timelines of individual recruits?
A decommit percentage would be nice.
D'onofrio is just a pawn. This is Al Golden's team and Al's defense. Al's the clown. Dorito's just the floppy shoes.
Good idea. Will figure a better way to present that data.
Good idea, but tough to pull off.
In all honesty, I think it is just a tidy way to make people talk more and increase the site hits- which I guess is the whole point. I don't think it really adds anything of significant value to what people who are paying members of sites already know. Actually, this is my biggest beef---these predictions are free to anyone, but they are based on inside knowledge of your staff apparently. Information that you are asking all of us to pay for on individual sites. Of course, you can say that all the details aren't laid out in the "crystal ball", but my reply would be "the main piece of info that everyone wants to know, and is paying for, is." So, if you are going to tell me where highly sought after recruits are likely to end up for free, why should I pay for it?
Our sites offer a balance of free and pay information. We have been upfront about our model from the start. The VIP subscription is based primarily on the inside information provided in detail by the site experts (team specific insiders) through articles and interaction on the message board communities. The majority of the network content is free to attract users to the network and the team sites.
Also to your point .. we are not hiding from the fact that we are a business. Our goal is to turn a profit through subscriptions and advertising from traffic. However, we have proven to reinvest and invest in our product at a scale that ESPN, Rivals and Scout can not even come close to matching.
Theses additions are just the beginning of our offering. It is up to you guys to decide if you like our service enough to visit our site or enough to subscribe. But, make not mistake, we are not trying to pull anything over on anyone. We are here to be the best and have zero plans of slowing down. Whether you pay or not is up to you, we just hope that you feel we are providing you more value than any other service that you feel the need to support 247Sports.
Hope this ramble helps answer your questions. There's no end around or trickery here, just trying to give you guys more value for less money. We appreciate everyone's support -- no matter the level.
Saw this tweet from Noah and thought I'd throw some thoughts out there
@NoahStanley247: "Weighing the fact that National guys will predict 100s, while others will be in the 10s #CrystalBallProblems"
1) The one obvious way to do this is to use percentages. Missing a few will still affect local guys more than national, but it's simple and relatively effective. Still should be enough local recruits in most areas to predict so the percentages will even out.
2) Another way that would take more development and testing, is to give each analyst an "area." So national guys will have the entire nation as their area, but local guys will have a specific geographic region. Then, as a simple example, you could give each analyst 1 point for each correct prediction inside their area (which would be every recruit for national guys), but then 5 points for each correct prediction outside of their area. That way, it encourages local analysts to reach outside their area and gain more points. Allows the local guys a way of "catching up" to the national guys in terms of points (of course, you'd have to test to see which point totals are best for local vs. non-local predictions).
3) You could split up the rankings between national and local guys. Although that's not nearly as much fun as ranking everyone in one big pool.
4) This one is really just a brainstorm, not sure I see much coming from it. However, you could use Gaussian factors, similar to what you use for team rankings. This time, order the number of players predicted to specific teams and multiply those by Gaussian factors. Hard to explain, but I'll give an example. We'll say that Sean Fitz predicted 10 correctly to Penn State and 5 correctly to West Virginia. Put those in order, so 10, 5. Then multiple the top one by a Gaussian factor (10*1), and the second one by a Gaussian factor (5*0.96 or whatever), etc. Have a small standard deviation so local guys will still have a decently high total relative to others. Like I said, not sure I see this working as well, but it's a possibility when the local guys should be able to predict more heavily for a few teams, while the national guys should be spread out.
Something else I thought of that's semi-unrelated to the issue mentioned in the tweet, is to break down the predictions by team or conference. So if JC Shurburtt is correctly predicting a lot of kids to South Carolina, but not as many correctly predicted to Oregon, people can see his percentages as a team breakdown. That way, people can see what teams (or conferences) analysts might be more knowledgeable about.
Another unrelated thought is that you could somehow assign a certain # of points for each correct prediction based on the number of other people who also predicted it correctly. So if fewer people predicted it correctly, you get more points. As an example, if Analyst A predicted Jeb Blazevich to Georgia along with 17 other people, he could get 1/18 point for it (1 correct prediction divided by 18 total correct predictions). Whereas if Analyst B correctly predicted Chris Godwin to Penn State along with 1 other person, they could get 1/2 points for it (1 correct prediction, two total people). That encourages predictions for more (and possibly lesser rated) prospects.
I'll throw out some more brainstorms another time if the mood strikes me, but hopefully those thoughts help in some way. You've got a lot of potential with the Crystal Ball.
1st off all these new features, especially crystal ball, are awesome
whats the protocol for when a recruit silently commits and the insider knows it but the news is embargoed
that shouldn't count as a prediction, should it? or maybe they should get rewarded for having the info first
regardless great stuff 247 ... looking forward to seeing the continued evolution here
Thank you psubills62,
1, 3, 4 (though that specific version was not thought of) and the split points are all things we're considering. Average could be an option, we just need to figure out an average of what? Probably points for getting it right, further time out, divided by number of people with same prediction during the same time (If you got in first you get 100% during that period, when someone else joins your predictions, you get 50% during that period of the total and so on). Though someone who predicts late (final 36 hours) on 15 kids and is 100% correct is valuable, it's still not not the full intention of the crystal ball to our users IMO. Fortunately we'll have a lot of data when we finally get around to test this to see what's going to be a good formula.
Regardless, the guys are excited for it right now so we'll want to get something that continues to keeps it fun for the staff and other experts.
I'm testing a tool right now that will allow our editorial team to lock any new predictions for specific schools on a kid. If the word gets out like it did with Blazevich, they could lock any new predictions to Georgia because the 15th person to say Georgia when no other school is being selected on the day of his commitment isn't much value to the user.
I do think the first few people who tell you the school is valuable, especially the first. That insider worked hard to get that information and deserves some credit, even if it's just being displayed first on the page. There is a number after that though that adds no more value, and in those instances (like the one you pointed out), editorial will lock new predictions to that school.
I haven't read this whole thread, but what MrWoodson is saying happened yesterday with Bunting. Wiltfong changed his pick from ND to Michigan in the last minute. I don't fault or blame him, but it is misleading.
We'll have a tool that will mark time and be able to lock changes to prevent some of this. I can't fault a guy for changing his pick to the correct pick. I would if I knew no matter how late it was before the commit.
That is a solid idea.
That's coming, but it will be after user page and team pages.
247Sports In partnership with CBS Sports