Jump to content

Wisdom of Crowds - your predictions please!


MMATycoon

Recommended Posts

Haven't really analysed the results yet as I've just been programming the scripts that work everything out. The raw values really mean nothing - it's all gunna depend on the weighted values.

 

Top valued predictors so far are.

1. http://www.mmatycoon.com/managerprofilemanager.php?MgrID=79766 (8 of 10 right)

2. http://www.mmatycoon.com/managerprofilemanager.php?MgrID=66844 (6 of 7)

3. http://www.mmatycoon.com/managerprofilemanager.php?MgrID=69063 (5 of 6)

 

New event up now for your predictions.

Link to comment
Share on other sites

Been updating the event page this evening to show predictions / returns etc. Considering the early stages and relatively small number of people predicting, I think the results are very promising indeed :)

 

Bisping vs Rockhold

Favourite wins: 5
Underdog wins: 5
$10 on all favourites nets: $-34.16
$10 on all underdogs nets: $31.38
$10 on all our picks (raw): $20.05 (6 right, 3 wrong)

 

Shogun vs St Preux

Favourite wins: 7
Underdog wins: 3
$10 on all favourites nets: $0.43
$10 on all underdogs nets: $-24.79
$10 on all our picks (raw): $33.45 (8 right, 2 wrong)
$10 on all our picks (weighted): $23.68 (8 right, 2 wrong)

 

The weighting will be very inaccurate compared to what I want it to do, until maybe 10 or so events have gone by.

 

Anyway, I'm really happy with how it's gone so far. Gunna work on fighter and manager profiles for it next as well as making it more visually easy to take in the info.

 

N.B. Will probably work on (for the game itself), user inactivity tomorrow, to bring down hype for inactive users.... Just so you guys know I'm not totally obsessed by this ;)

  • Upvote 1
Link to comment
Share on other sites

Another two events in the books, another two significant profits for the weighted predictions. Couldn't be going much better at this point!

 

Particularly happy that the raw predictions (the bit anyone can do), is going badly but with the equations I've come up with, we're turning a profit :)

 

edit; damnit! when i run the script from the start again it does Werdum vs Hunt before Ortiz vs Bonnar and it screws up our weighted predictions :D trust me though, we had a positive result on Tito vs Bonnar of 4-1 :P

Link to comment
Share on other sites

Noob question as I just got around to trying this during UFC 180 and Bellator 131. But on the sidebar menu the link is shown as Wisdom of Crowds -(3), so how do I get rid of that? Or will it just go away after a few days? I obviously clicked on the link. But I figure the (3) is the addition of the upcoming UFC events?

Link to comment
Share on other sites

Mike - Are you going to further weight this (for gambling/prediction purposes) once you get enough data so that you have confidence in manager thresholds, ala - Manager A has a historic record of 80% correct predictions, therefore his predictions are weighted at (for example) 0.8 and Manager B with historic record of 20% correct gets weighted at 0.2?

 

I would expect this would remove those who have little idea of what they're talking about from having an excessive influence as with enough data you'd be able to work out what % of accuracy is statistically significant.

Link to comment
Share on other sites

Yes that's already happening in the weighted predictions we are using. Everyone has a rating worked out using multiple different factors.

 

I have it set up so I can run the entire thing from the start at one click, to change everyone's rating as I see fit, if I think of a better way of doing things, so we can play about with it quite a lot. You guys can see the rankings based on those ratings (though not the actual rating, as that's kinda like the coke's secret recipe)... so if you see anything after a few more events where you think x is being underrated or whatever, you can let me know.

Link to comment
Share on other sites

Very eager to see how this thing will evolve - It's just out of curiousity really, as i unfortunately can't bet on MMA here in France - But i'm happy to give my predictions for the project.

 

Mike maybe you could give out spin and win credits as an incentive to place predictions, but on the other hand it could make people predict whatever just to get the benefits....

Link to comment
Share on other sites

http://i.gyazo.com/8eb465579155dac286e08e21cb1dee5b.png

 

how do i get rid of this number. i have predicted the fights but those numbers are still there and its annoying me :D

I agree with this. Except Wolf is a lot more polite than me.

  • Upvote 2
Link to comment
Share on other sites

Mike - It may be worth giving a bit more of a brief description of what you consider 'knowing what you're talking about' means. I expect you've got people not participating because they don't want to throw out your numbers, yet they likely 'know more' or follow more than some of the people participating.

Link to comment
Share on other sites

A couple of random thoughts :

 

- Maybe you already did it, but you should eliminate a small percentage of the most extreme predictions (on both ends), to get rid of errors and whatnot. This is commonly used to refine an analysis, although it kinda goes against the WOD theory... but maybe you should try with and without this, and see how it goes.

 

- Not that i'm an expert by any means, but refrain yourself from tweaking the algorithm too much : One could be tempted (consciously or not) to tweak it in every way with too little data until he's "proven right". Great results don't necessarily mean that the algorithm is perfect, but would most likely mean that it "fits" with the few event you've got data from.

 

- One thing i'm affraid of : If / when the system is proven to be working, the fact that the crowd opinion will be visible could affect the whole thing : If we have existing data visible before giving our predictions, our picks could be (again, unconsciously) biased by the people's overall opinion. EDIT : same thing about the "Manager rankings" and "WOC profile" : Hopefully most of us don't give a damn about our rankings, but the will to climb the laddr could lead to un-natural predictions. Plus seeing how we do individualy kinda goes against the principle :)

  • Upvote 1
Link to comment
Share on other sites

The only thing I have any concern about is point 3.

 

1. The value weighting for users negates any problems with point 1. Actually, if someone goes all out on a limb, that's really important and needs to be factored in. That's probably where a lot of the value comes from.

2. I disagree with you on this one. Once we have lots of results in, I really should be looking to change the equations to make the data "proven right". The raw data is never going to change / lie, it's the interpretation of the data that's where the money is at :) It's my job to change those equations as much as necessary to maximise profits :)

3. I was concerned about this too. However, after thinking about it, competitive nature isn't a bad thing. If it makes people put more effort/thought into their predictions then good. The only problem would come if people started looking at others' opinions (e.g. Sherdog). If lots of people did that, we'd be weighted towards a certain prediction and we'd have lost the independent nature of people's predictions. So yeah, be as competitive as you like but don't look at other people's opinions. Still though, all's not lost if people do that because the weighting system (WOC rank) irons out poor predictions. If people read articles that are right beforehand, good. If they read articles beforehand that are wrong, they get marked down for upcoming events.... Still effectively a win win over a longer period of time; just a negative for the event where they listened to someone else's poor prediction.

Link to comment
Share on other sites

Gotta admit your points make sense...

 

 

Can i be curious about the way you pounder the results with the WOC rank? I mean, how much the best ranked's opinion does count compared to the last ranked one? You're talking about ironing out poor predictions, but aren't they one of the bases of the WOC theory? if the weighting is too strong, you could get back to a system where you trust more "the expert" than "the crowd".

 

Please don't see my comments as critics, I'im just a whore for theories and statstics, and i've been reading some stuff about WOC in the past too.. i guess i'm just happy to discuss this shit here rather than boring my wife with it :)

 

 

Edit: also, do you use managers' confience for the WOC rankings ? here's an interesting read if you haven't stumbled upon it yet : http://www.technologyreview.com/view/528941/forget-the-wisdom-of-crowds-neurobiologists-reveal-the-wisdom-of-the-confident/

Link to comment
Share on other sites

Damn it, well we had to have a negative event sooner or later! We had a rush of predictions for Jared Rosholt in the couple of hours before the event, otherwise the pick was on Oleinik and we'd have made another profit :P Oleinik was slated to be the #1 value bet, so we'd have put $18ish on him and returned a $33 profit! :D

 

I was worried about this event beforehand though cos the bet values were very low compared to previous events. Still, it's all good info to find some of the pitfalls and perhaps we'll be suggesting "Don't bet on x fight".

 

@Manta - I can't reveal the secret recipe :P

We have the unweighted predictions to compare to, when it comes to the issue over over-weighting the top predictors. We'll go with whatever system produces the best results over a sustained period.

 

Thanks for the link to the article, I'll check it out. Yes we do factor in confidence into predictions and the WOC rankings.

Link to comment
Share on other sites

Looking like there isn't much value at the top of the card on the next event. No odds out yet for the undercard but hopefully that will be better.

 

Of the 6 events so far we have three 2 starers, a 1.5 and two 1s.

 

 

 

If you haven't checked out the stats page yet, when I encorporate "no bet for crap bet ratings", we're up at $120, instead of our current $91, with a much smaller total amount bet.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...