Community > Forum > General Corporate > Truly a game of chance

Truly a game of chance

Posted by: 1_kermit - Sun Apr 09, 2006 8:39 pm
Post new topic Reply to topic
 [ 29 posts ] 
Truly a game of chance 
Author Message
Rocket Constructor
Rocket Constructor
User avatar
Joined: Wed Apr 05, 2006 5:35 pm
Posts: 7
Post Truly a game of chance   Posted on: Sun Apr 09, 2006 8:39 pm
This is truly a game of chance. I just LOST to someone who's prediction of the high temperature was over 15° lower than what it actually was. My prediction was 46.75° and his was 47.75°. The real high was 63°. What a difference!! We both got our information from professional weather forecasts. If professional weather men are off by more than 15°, then how can I be expected to do any better. :roll: This is truly a game of chance. For chance to fall in your favor 17 times in a row, you will have to be very lucky! Hope my luck improves soon! 8)


Back to top
Profile
Space Walker
Space Walker
avatar
Joined: Thu Dec 08, 2005 5:44 pm
Posts: 136
Location: Texas, USA
Post anecdotes weigh heavily   Posted on: Mon Apr 10, 2006 1:49 am
I think we will find that it is rare that the forecasts are off by 15 degrees. A heavy rain can really warm things up. If everyone is at the same skill level, then your shot at winning is exactly the same as everyone else's.

On the other hand, someone who can forecast within 1/2 a degree half the time and within one degress all the time will win or tie 87.5% of the time against someone who predicts in a uniform distribution within two degrees of the actual temperature.

Like a multiple choice test, if players can eliminate some bad answers, they can vastly improve the quality of their predictions.

What will make the game popular is the belief that the weather man is frequently wrong and that we can do better. What will make the game legal is that weather prediction is a skill, a hard skill, but one that will give skillful players their due.


Back to top
Profile WWW
Spaceflight Trainee
Spaceflight Trainee
avatar
Joined: Thu Apr 06, 2006 8:26 pm
Posts: 45
Post    Posted on: Mon Apr 10, 2006 5:40 pm
I disagree. The skill is in knowing which weather service to use and how you position your prediction. Of the 30 matched entries I have played at various levels 24 have been successful predictions (that's 80% success)! That does not include the sweeping victory I am about to have today, which will increase my success rate and carry me to level 5. Those lateral victories could just as well have been vertical ones, meaning I could have already won 17 times, because the skill level does not change as you increase in level (except that the skill of your opponent might).

I will not tell you how I achieve that rate until I've won my first flight, but it is not chance. Play on!


Back to top
Profile
Space Walker
Space Walker
avatar
Joined: Fri Jul 30, 2004 3:17 pm
Posts: 243
Location: So Cal, baby!
Post    Posted on: Tue Apr 11, 2006 1:26 am
Doesn't seem fair to me. If I lived in one of those buildings overlooking Central Park, I would have a very major advantage. Or, if the chosen spot was in Southern California I would have a distinct advantage. The target location should be variable: New York for a week, then Houston, LA, Tokyo, Moscow, Paris, London, etc...


Back to top
Profile
Space Walker
Space Walker
avatar
Joined: Thu Dec 08, 2005 5:44 pm
Posts: 136
Location: Texas, USA
Post Target location   Posted on: Tue Apr 11, 2006 2:33 am
The prediction location needs to be static. Players need to get a feel for the data there. Data sources need to be qualified. Predictions need to be comparable from day to day. Too much variability and participation will vary.

I don't think New Yorkers have an edge. There are central park web cams and very good data sources with recent data about central park. I do think that they have an interest nexus if they frequent the park. They may play more. If that is true such that the bulk of players are New Yorkers and we have sufficient liquidity for multiple sites, we will be adding additional geographic areas starting with Imperial Gardens, Tokyo.


Back to top
Profile WWW
Space Walker
Space Walker
avatar
Joined: Fri Jul 30, 2004 3:17 pm
Posts: 243
Location: So Cal, baby!
Post Re: Target location   Posted on: Wed Apr 12, 2006 3:01 am
dinkin wrote:
...we will be adding additional geographic areas starting with Imperial Gardens, Tokyo.


That would be a start. Webcams don't cut it... do you have web-chemoreceptors? web-tactile-sensors? People whom are natural weathervanes use all of thier senses.


Back to top
Profile
Space Walker
Space Walker
avatar
Joined: Thu Dec 08, 2005 5:44 pm
Posts: 136
Location: Texas, USA
Post New York feeling   Posted on: Wed Apr 12, 2006 5:57 pm
One of our leaders is in Washington State and the other is in Louisiana. If we get two New York winners in a row, I will move the game.


Back to top
Profile WWW
Spaceflight Enthusiast
Spaceflight Enthusiast
avatar
Joined: Thu Apr 13, 2006 1:56 am
Posts: 1
Post Not chance   Posted on: Thu Apr 13, 2006 2:16 am
I will start off by saying that I am a meteorologist. There are some ways I believe I have an advantage over those who aren't. On the other hand, the fact that anyone could go and "adopt" the forecast from a weather service (such as the Weather Channel, NWS, or AccuWeather) does sort of even the field a bit. The reason is those services use the same data I can use to come up with their forecasts. Thus, by using their forecasts a person is indirectly using some of the data I would use to do my forecasts.

Where I may gain an advantage is that I generally have a good idea of the weaknesses and strengths of the forecast models (also known as model biases). But, what we have to remember is that sometimes we are talking about a difference in high temperature forecasts of a degree or two, which is especially the case on days when no significant weather is expected. In such cases, whether the temperature hits 71 or 72 degrees can make all the difference, and it is generally impossible for someone to make an argument for forecasting 72 rather than 71 degrees, or vice versa. Thus, those with no meteorological background using a weather service forecast may have just as much skill in forecasting the high as I do in these situations.

This is, of course, a bit different when there is significant weather expected (such as rain, strong winds, clouds, etc..). In these cases arguments can be made for forecasting 65 degrees rather than 71 degrees, or vice versa. This is also where knowing model biases as well as other factors can make a difference. Even so, sometimes the weather can do things we don't always predict correctly (though there is usually someone who does get it right on occasion).


Back to top
Profile
Launch Director
Launch Director
avatar
Joined: Thu Apr 27, 2006 3:39 pm
Posts: 13
Post    Posted on: Thu Apr 27, 2006 4:40 pm
I don't know... I'm willing to bet that the skill level, experiences, and effort of the winners are going to vary quite a bit.

Early on, the winners are going to be the ones that can "choose" the right forecast model or CCM company.

Future winners may rely a lot more on statistical derivation or will be extremely selective on when they use their plays (for instance, I'm leaning toward never playing when precip is forecast, as it throws all my temp forecasts off to an unpredictable degree). In this case, the advantage might go to a local who can look out the window and look for specific forecasting keys.

At some point, I expect somebody to attempt to build a mesoscale Richardson model and tweak it appropriately. I wonder if I still remember how to do that -- computers and data aquisition are a lot faster than they were when I was an undergrad taking numerical prediction.

Alan Sheets
asheets@frii.com


Back to top
Profile
Launch Director
Launch Director
avatar
Joined: Wed Apr 12, 2006 3:12 am
Posts: 16
Post    Posted on: Thu Apr 27, 2006 6:45 pm
So what would you guys suggest to make the game more skillful? Do you think having more levels would do it? I don't want to compete through more levels, unless the entry was less or something. It sounds like the people that can make it through a couple of levels aren't going to just be guessing. I would take my chances any day against a guy who was just randomly guessing. I do agree that it often feels like chance, but I am a math major in college and I would have to say over many levels the statistics of anyone wining the whole thing by random guessing is nearly impossible. I haven’t crunched the numbers yet but I am sure it is nearly impossible.

As whether people in New York have an advantage, I think that’s just aesthetic. Of course you think you have an advantage if you live there or a disadvantage if you don’t. But, the national weather service does not have weather predictors sitting in central park every day. They send the info to people sitting behind desks just like us. We review the info the same as they do. Now, being a meteorologist would defiantly be an advantage, don’t want to play against that guy, but I am not afraid of people in New York predicting.

It does sound cool to have some other locations.


Back to top
Profile
Launch Director
Launch Director
avatar
Joined: Thu Apr 27, 2006 3:39 pm
Posts: 13
Post    Posted on: Thu Apr 27, 2006 9:14 pm
I think the contest would be a bit more interesting if wins/losses weren't determined just by predicting the high temp, with low/precip, and rh used as tiebreakers. Since I was more of a meso/micro scale specialist when I was active, I'll probably get blasted by those who's specialty is more temperature based. But, if the deciding factor was on winds or precip, I'd stand a better chance.

Instead, I'd rather see the contest based on a full-forcast, where each of the 4 elements constitute 25% of the score. Whomever does the best job with a full forecast wins that round.

You could make it even better by adding additional elements, such as average wind speed/velocity, predicted cloud cover, predicted watch/warning elements, etc.


Back to top
Profile
Space Walker
Space Walker
avatar
Joined: Thu Dec 08, 2005 5:44 pm
Posts: 136
Location: Texas, USA
Post Rubric requirements   Posted on: Fri Apr 28, 2006 4:15 pm
Critical to the rubric is that ensuring players of ordinary skill are competitive. If only a weather forecaster can predict all four excellently, then that is an argument for keeping the rubric the same. Remember it is not enough for a game to help the more skillful beat the less skillful often. There also has to be enough hope for an average player to attract lots of average players to the game.


Back to top
Profile WWW
Launch Director
Launch Director
avatar
Joined: Thu Apr 27, 2006 3:39 pm
Posts: 13
Post    Posted on: Wed May 03, 2006 10:05 pm
A multi-element forecast and scoring system was what was always used in the prediction contests I took part in as an undergrad. Quite a few freshmen beat grad students back then.


Back to top
Profile
Space Walker
Space Walker
avatar
Joined: Thu Dec 08, 2005 5:44 pm
Posts: 136
Location: Texas, USA
Post 10 variable   Posted on: Wed May 03, 2006 10:24 pm
I have seen multi-city challenges on the web where you have to match the ones digit of each city.

There is a form on National Weather Service that has the high temperature in Central Park to the tenth of a degree.

There are many choices for game design. Why do you recommend a multi-variate approach? What rubric does it use?


Back to top
Profile WWW
Launch Director
Launch Director
avatar
Joined: Thu Apr 27, 2006 3:39 pm
Posts: 13
Post    Posted on: Wed May 03, 2006 11:48 pm
I recommend a multivariate approach simply because it is more fun that way. Every single play I have made, save 1, has been decided by hi temp only. After looking at some of the other competitors, I note that they have had the same experiences.

According to NYU-Albany's wx school, the main genesis of wx prediction contests was to get away from the simple reliance on simple regression, numerical models, etc. From that point of view, judging a winner solely based on high temperature means that a guess is as good as a regression, which is as good as a weather model, which is as good as simply quoting a CCM or the NWS. As it stands now, this is a game of chance -- which is boring to me. If you want to keep me interested, and pumping money into your company, make me challenge my skillset a little bit more.

As for contest rubrics, there are several. A few that I've found right away include:

* max, min, and PoP -- each as 1/3 of the day's score (NYU Albany)
* max, min, precip amount -- scored as 80-(1 point per degree error)-(4points per inch error) (National COllegiate Weather Forecasting Contest, which, BTW, I participated in from high school [when I had no interest in weather, BTW] up until they shut it down 4 years ago)
* precip only contests (too many to list)
* tropical storms prediction contests (many different variants,, including # of storms, # of landfalls, # of particular types, nowcasts, etc -- I won some of these as an undergrad, ticking off a few people at NHC, BTW.)
* extended forecasts (predicting max and min for a rolling 5 day period -- a bit more challenging, I think -- maybe too challenging to the point that at day 5 a guess based on climatology is statistically better than one from a computer model)
*long-term precip contests -- Accuweather is running run of those right now...

There's lots of different ways to go about this that would be fun AND inclusive of the non-trained individual. You'll notice that I haven't included a bunch of wx elements that are possible for a contest, but highly advanced.[/list]


Back to top
Profile
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 29 posts ] 
 

Who is online 

Users browsing this forum: No registered users and 15 guests


© 2014 The International Space Fellowship, developed by Gabitasoft Interactive. All Rights Reserved.  Privacy Policy | Terms of Use