Preditter
My partner Albert is a hacker. We didn’t invest very much in development platforms before Albert joined us (Twitter being somewhat of an exception) because we couldn’t eat our own dogfood. Albert changed all of that for us. He can not only evaluate development platforms but he can use them and provide feedback to the team.
Over labor day weekend, Albert hacked together a web app he calls Preditter. As he explains in this blog post, he built Preditter on top of three of our portfolio company platforms, Twitter, MongoDB, and Twilio.
If you want to make a prediction about a football game this weekend, just tweet the prediction to your followers and add @preditter to the end of the tweet. I just did that. Here is the tweet:
Jets over Ravens on monday night @preditter @garyvee
Let’s do Albert a favor and test Preditter out. And let us know what you think.
Comments (Archived):
hmm… well so far it seems to think the Ravens are going to win based on 1 prediction (me) even tough You and Gary V predicted Jets victory…
yeah, i noticed thati wonder if there is some delay in compiling the stats
More data needed obviously, but love the concept – especially since I need to submit a football pool sheet, and have a fantasy draft today! 😉
Maybe it “knows” something … it could be looking out at other twitter stream data too, right?
well, i checked it again and it was 50/50 based on 2 predictions so whoknows! :)Andrew Watsonhttp://tumblr.gleep.org/
bringing Vegas to the masses? (just my first thought what you can go with this…never mind the interactive transmedia architecture you could use this with sports programming)
With users having no skin in the game, I’m skeptical about how predictive Preditter will be. Time to call your friend Chuck Schumer and get him to loosen the rules on online betting so you can make Preditter more interesting.
Maybe instead of wagering money there could be a reward points / karma system for being right? Bragging Rights?
Maybe. It would be interesting to try that and compare its predictive ability to that of Intrade on the same questions. My bet is that Intrade would be more accurate because there’s money at stake.
Or conversely, those who pick losers could engage in community service as the “cost”. 😉
true
Maybe you could legally do this out of the UK now, if you found away to exclude American Twitter users?
wow, betting on twitter? that would be fantastic.
That’s why we have the best peer to peer betting platform in the world Dave http://www.betfair.com
Given the fiscal hole we’re in, I see no reason why we shouldn’t legalize online betting and tax it.
I like the concept. I must admit the return to command line interfaces for mainstream applications I find a bit puzzling.
Geek PR!Well done Albert
There was a time when people thought twitter could predict the weekend box office grosses:http://www.fastcompany.com/…But earlier this summer Scott Pilgrim was trending for weeks… and yet totally tanked.
well, that seems like an obvious example of something where the subject matter might resonate more with the twitter crowd than anyone outside of it…while 145MM+ people might be on twitter, the vast majority of usage is probably from a much smaller and more homogenous core group than the entire member list. (as @andyswan tweeted much more wittily about last week)
The tweet, for the uninitiated: http://twitter.com/AndySwan…
Perry beats Obama 2012 @preditter
who is Perry?i should follow him
In act of contrition/survival @fredwilson cuts max FEC check to http://www.rickperry.org/ by 2012 @preditter
or just a hedge 🙂
Political Armor.
funny, but no way Perry runs.
I’m very interested in your reasoning.. whatcha got?
1) he’s adamant that he’s not running. 2) he’s taking none of the necessary steps towards running. 3) he tells people privately that he’s not running.
I’m willing to wait until Feb ’11 before I follow the conventional wisdom on this one.
have you been reading Paul Burka too much? 😉
Governor of god’s gift to mankind, the greatest by-god State in the union!
Apparently someone that is going to beat Obama in a game of golf … that is what you were referring to right Morgan?
After watching the guy swing a club ( http://bit.ly/aFzXMv ) and throw a ball ( http://bit.ly/c2x0MB ), I’d suspect that’s been done approximately 147 times during his presidency: http://bit.ly/dhALMx
I don’t see a sub 90 round in his future.
Maybe front 9 🙂
Britney’s gonna surprise Fashion Week at Dasha Zukhova’s Pop magazine party at Don Hill’s Friday. @preditter
love it!
OMG I really think Jack is finally going to propose to Daphne this weekend!!!! @preditter
There is a difference between predicting and hoping. Will the app start to track individual performance and give greater weight to those that are right more often?You could sell access to view the profiles and predictions of those people that are most likely to be right.
i am hoping and predicting that the jets will win on monday night!
All you needed was a safety. :)10 to 9 was more like a baseball score.
we needed an offense
I think this is pretty cool.If it had categories built in with # tags or something to sort the predictions and some simple graphic design work it might find some angel funding pretty easily 🙂
My kids pee-wee soccer team is gonna kick Bedford’s ass on Saturday. @preditter
AVC.com will have more than 300 comments related to Preditter post today @Preditter
Ha! momentum’s not there yet.maybe I can come up with 150.nope….I have work to do.
at this rate, you will be right
Would the @preditter posts on twitter be more than the comments here on AVC.com
tru dat
Standing-room-only crowd at the school’s Fall Fundraiser and $180,000 raised! @preditter
Interesting – using twitter and twilio as input engines for “wisdom of the crowds” decision making.Like an open source survey monkey
The programming side of this is cool….showing how multiple portfolio companies can be integrated into a single app. Cool stuff by Albert.Aside from that, if the goal is to actually get more accurate predictions, it’s not going to work. If the goal is to let people have a little fun, it might….especially if it went with prizes and so on.The reason it will fail as a predictor is that users are putting up no capital. No risk. Money is the ultimate vote of confidence. My $100 bet should be 100x more indicative than Tereza’s $1 bet. You can’t do that with fake money or badges or anything else.To anyone on this board, I will bet you $(insert whatever amount you want here) that within 2 years you cannot come up with prediction software that BEATS this: http://bit.ly/b0PTFNThe same can be said for the stock market. You cannot build a prediction market without using real money that is better than the most liquid and finely tuned prediction markets in the history of the world: The stock markets themselves and their derivative counterparts.I like where Albert is going on the app side though. A lot. It will inspire other ideas of how to use his portfolio companies in applications….and I’m guessing THAT is what this is really about 😉
If you were to add reputation and accuracy of prediction as a form of capital it would work a lot better.The more often you’re right the more strongly weighted your opinion is.
I’ve seen so many of these on the finance side that I can confidently say you’re right….it would work “better”….but still confidently say that it still wouldn’t WORK. 🙂
Totally agree, but for a slightly different reason.I’ve played with lots of predictive math for gambling apps (poker, sports betting, et al) and there is usually lots of money to be made in a theoretical sense. There’s a lot of dumb money gambling.The key issue is that the vig the casino takes is extremely high, so it’s very very hard to be a long term winner.
there are definitely tons of long-term statminers who make tons of money betting sports.
By beating the lines, do you mean the final lines as game predictions? Assuming you’re counting the vig as part of the bet and you must bet on every game, I agree with you that Vegas lines are pretty much unbeatable. But beating them >50% of the time is certainly doable, and if you build a line-shopping algorithm it’s possible to get up into the profitable range although you probably won’t be betting on many games and would likely be running a ton of arbitrage plays.This is a really cool idea though and I think it’d be interesting to see the intelligence of various crowds on such items- it’s clear that many times ESPN.com polls are inaccurate due to the biases inherent in ESPN’s coverage which are pushed through to the unwitting viewers. This seems most clear during March Madness where the popular underdog picks seem to be talked up to such a fervor they become overrated, and in small conference vs. big conference bowl games where the big name teams are likely to be overrated due to ESPN covering these teams much more and having a vested interest in keeping their golden geese as golden as possible.
Selected lines are fine, and be up by more than the yield on a 10-year Tbill on a -104 vig after 500 predictions/bets.
That’s interesting. I think you’re right Andy. Be it money or some other scarce currency it lacks teeth.Without it it’s a PR mechanism.You win if you put in more money into contests with the largest most engaged audiences.What I do like in Foursquare is re-setting the rankings every week. Said another way — don’t let the currency compound because I think it skews the predictability.So each user gets handed a fixed amount each week (e.g. 1000 points). Then it’s really interesting to see what people did starting with the same (but finite) asset and with equivalent opportunity costs.You can track cumulative results in a separate leaderboard but basically what i’m saying is that if Andy has $100 and I only put in $1, that works — as long as i have $99 going into alternative bet(s). If I only had $1 to start with then it doesn’t work.
re-read your comment and realizing I misread.still deciding how I feel about it needing to be real money.it does need to be a scarce resource that means something. Usually that means money. But it doesn’t have to.Also if Tereza bets $1 on things day in day out over and over and then one day bets $100, does my variation from the norm mean anything?(had I paid better attention in Finance 602 I’d probably know the answer to this question)
If you’re a proven sharp bettor, your variation from the norm could mean something- but in this case it can’t mean anything other than your sudden increase in confidence for this pick is likely an overreaction- when betting, you can determine a basic unit size and then vary it on each individual bet based on your estimated probability of winning the bet using the Kelly Criterion formula, set to your desired risk of ruin (loss of your entire betting bankroll). You’d have to basically be 100% sure of winning to bet 100x your normal bet.
Or that $100 (or more) just fell in my lap.Or the place I usually put my other $99 is no longer available.
No wisdom of crowds?
I’ll bet against the crowd everytime
I think this could also be used to play a game like ESPN streak for cash. You could set up a league to see who picks the most games correctly in a row. Also sometimes called a “Survivor League” in fantasy football. Doing it through preditter is more fun than a league because you can have a conversation/trash talk around your league. I am always a little surprised by the lack of social in fantasy football. Yahoo and ESPN give you “trash talk” options, but they just aren’t a good platform.Regardless, I like the idea, very cool use of portfolio companies! Great to see an investor hacking something together like this. Excited to see how it progresses.
Kudos to him for putting out a MVP to market test before spending too many resources. Other startups could learn from him.
Nice use of api’s but more than anything to me it shows you the potential that Twitter has if it mined it’s users data and added some algorithms and tasty search functionality.My main use for Twitter is trending topics that I’m interested in. So for example is it worth me upgrading my iPhone to IOS 4.1 based on other twitter users reactions to it ?
Before I go hear shofar:Z+ Brand, jason wu’s down line, will have a large cult following after NY fashion week @preditterIt will probably be right, but I don’t think this is a good way of measuring Jason Wu’s popularity
l’shana tova shanawe had a very lively and fun service in our shul this morning
Same to you and your family Fred.Link is to a photo of a Rabbi on Via Rodeo Drive in Beverly Hills this evening with his Shofar.Only in BHhttp://awaldstein.tumblr.co…
I love that photo (and a happy new year to you too)
Thank Fred, a Good Year to you too. I wish I could say the same for mine, I think if I go back to the same minyan I will bring paper balls for every time the cantor goes “AYAYAYAY”
Very cool. Just roll in the moneyline vs aggregated predictions and you’ll have a pretty interesting app for arbitraging sports bets.
Something like this could be used to determine which game of a regional broadcast to show in certain areas, if local blackout restrictions allowed. Might be really cool during March Madness when often a top seed’s first round game is on but no one really cares about it in comparison to a 5-12 battle on at the same time in a much smaller area. Although March Madness On Demand seems largely to have ameliorated this problem.Thought of this because in clicking around on a few of the teams, I noticed tonight’s game had several more predictions than the others (for obvious reasons) indicating it is a more popular matchup at this time. Small sample size disclaimers apply.
A system like @preditter can be also good to know, for example, what people think about stocks intersecting the results with technical analysis…
I like this a lot. A “thing that makes you say … hmmmmm” and great example of how stitching things together in creative ways can often result in value in excess of collective sum..
Predicting box office success a la Hollywood Stock Exchange (http://www.hsx.com/) — shouldn’t they be using Twitter/Preditter?Cantor Fitzergerald finally abandoned plans to securitize films this summer, which was wildly complicated. BUT doesn’t change the fact that markets do a decent job predicting box office, better than when experts/critics call it out.Can do for any new product or website launch — will they hit 100k users in 100 days?
Love it! In sports, I see it as a cool tool to be used in fantasy leagues. With a graphical UI on top of the data, it would be a quick snapshot of what others are thinking across players. A personalized widget for use inside of a personal league could be interesting too.
Thanks for all the comments here! It’s really much more of a weekend project to get first hand experience with technologies that we have invested in than a serious attempt at getting better NFL predictions …
What language? And can you share the source (gist/github repo)?
USV is going to have to start funding itself…. 😉
i thought the same thing this morning when i saw preditter
I think $PREDICT may work better to organize Fred and Albert at least for stock market
Cool stuff…I built a game like this last summer ( http://gawk.it ) but it was very hard to get people to play it (command line games are hard to make stick)…still I like seeing people play in this space and so will def. applaud the efforts and do what I can to push it too! 🙂
Today is the beginning of an awesome New Year for many of the terrific AVC people! @preditter (For the rest, you have to wait until next January 1 or you can count the last one retroactively.)
Fred, you’re nuts. There’s no way the Jets are beating the Ravens on Monday night.
I am 2 for 3 so far on @preditter. Yay Saints, Bummer Verdasco, Geaux Tigers! and Vamos Rafa! See you on TechCruchTV later. PS: Anybody got extra USOpen tix for sale for Saturday or Sunday??
Interested to know, Twitter and Mongo are obvious for this app. How does Twilio fit in?
Nevermind, answered my own question (RTFM…LOL). You can text a winner to a phone number as well. Nice!
Interesting post and awesome idea! My co-founder and I (during the same time frame) put together a weekend project using Grails/Groovy called SportsCred (www.sportscred.com) that also does NFL and other sports predictions.It was also an experiment in new technologies (mobile web & graph API), integrations and virality. I guess it’s not that surprising because of the time of the year, but great to see another hacker whipping up something during their free time. Would love to hear from Albert how the idea came about and where he wants to go with it.