 |
|
| View previous topic :: View next topic
|
| Author |
Message |
SloggingScotsman
Joined: 18 Jul 2006
Posts: 2640
|
 Posted: Wed Mar 01, 2017 9:37 am Post subject: Good Judgement Project |
 |
|
Anyone else here doing this?
I have been waiting for today to let a little experiment time out before I get stuck back in, and am curious to know if others have had similar experiences to me.
A year and a half ago when this opened up to the public from the black vault of the American intelligence agencies (and the Superforcasting book was published, which I noticed in The Economist), I decided to give it a go, and have done ok. For those who don't know it's a maths based prediction scorer. The original aim, iirc was to find Americas top global event predictors in the population and the results showed, again iirc, that housewives etc can do better than intelligence agents!
Now this is significant as the way I went around doing this is not that far from how I tackle remote viewing.
Anyhow I decided to try ten Predictions, which I know is a small sample, but I wanted some feedback from the project before I dived in, and it can take a year and a half for questions to resolve.
In essence I split my sample into two. The first I followed the projects protocols, basically revising my prediction as new information comes to light, while the other half I made an initial prediction and by the seat of my pants left it.
My conclusion is that the projects protocols do generally work. By revising Predictions you do make better Predictions. However this is based on a small sample, which is skewed by one of my ten Predictions being worse than a guess and worse than the median prediction. In every other case I beat the average by a fair margin mathematically.
Now the final prediction is currently in the nomans land of being computed, and it will worsen my score a little (Back in Oct 2015, I made the prediction that there was a 10% chance of Bashar al-Assad ceasing to be president by today).
However there is still the lingering question of Trump, which was not part of my Predictions here. Way before anyone else thought it possible he could become president I was making a tit of myself at Oxford University confidenly predicting his win. Yet about a month before the vote I decided to follow the good judgement projects protocols and waivered, declaring that I didn't know at that point in time (which was true) and would have gone the other way if I were voting due I guess to the belief of the press. So this tells me that I can't trust the good judgements protocols completely. Relying on my gut can be accurate.
As you know throughout my life I have made predictions that people, including a manager in Price Waterhouse, and separately a human resource manager at a multinational, laughed in my face about, but time has proven me accurate time and time again.
As you might imagine I am now looking to find ways of distilling this to capitalise on it, something that I have failed to do throughout my life. But with the good judgement project providing independent mathematical evidence, it has motivated me to work through those who laugh, then later "knew it all along".
But I am very conscious that ten Predictions proves nothing. Hence I was wondering if anyone else here is doing this project and are prepared to offer any insights from their own experience with it?
For those unfamiliar with the project, a brier score of 0.5 equates to chance, the lower the better, a superforcecaster is around iirc 0.2. My predictions have come in at:
0.000
0.001
0.005
0.020
0.060
0.064
0.073
0.115
0.500
0.583
Now updated fro all ten Predictions.
Last edited by SloggingScotsman on Fri Mar 03, 2017 10:39 am; edited 1 time in total |
|
| Back to top |
|
 |
Tony Stark
Joined: 26 Apr 2007
Posts: 2458
Location: Milton Keynes
|
| Posted: Wed Mar 01, 2017 11:23 pm Post subject: |
|
|
Isn't this like saying that if I could place my bets when the horses were half way round the Grand National then I would have a better chance of picking the winner? 
|
|
| Back to top |
|
|
SloggingScotsman
Joined: 18 Jul 2006
Posts: 2640
|
| Posted: Thu Mar 02, 2017 7:45 am Post subject: |
|
|
| Tony Stark wrote: | | Isn't this like saying that if I could place my bets when the horses were half way round the Grand National then I would have a better chance of picking the winner? |
tony??? I don't understand?
I am using a once US government funded, now commercialised, mathematical based prediction project.
Here, I am asking if anyone else is doing the same so that we can learn from each other.
If you mean that it is cheating to be able to change your predictions as time passes, well those are the rules established by the US government and then transferred into the commercial world. Who am I to argue? (Behave now folks).
For me personally I have done respectably without changing my predictions, though I do concede that changes do improve things, normally but not always.
I stumbled across it on a book review in The Economist, so it must have some credibility!
|
|
| Back to top |
|
|
explorerJC
Joined: 20 Oct 2005
Posts: 15242
Location: Farthingstone
|
| Posted: Thu Mar 02, 2017 8:36 am Post subject: |
|
|
you are still a sucker for a name, Sloggers...first it was a couple of Tories, then it was your lodge....recently it has been the US and the economist...
The Economist has to fill its pages every week and once the US government funded the likes of Henry Kissinger, so, at face value, another former US backed project is nothing more than that to me...
i haven't read the model, but do I assume that if you suggest that there is only a 10% chance of something happening and it doesn't happen then you are technically, according to the model, 90% correct???
_________________
www.appliedtri.co.uk Tri and Du coaching
www.naturalrunningform.co.uk Natural Running Form Coach
2018 Training Camps http://www.appliedtri.co.uk/training-camps/
|
|
| Back to top |
|
|
SloggingScotsman
Joined: 18 Jul 2006
Posts: 2640
|
| Posted: Thu Mar 02, 2017 9:18 am Post subject: |
|
|
| explorerJC wrote: | you are still a sucker for a name, Sloggers...first it was a couple of Tories, then it was your lodge....recently it has been the US and the economist...
The Economist has to fill its pages every week and once the US government funded the likes of Henry Kissinger, so, at face value, another former US backed project is nothing more than that to me...
i haven't read the model, but do I assume that if you suggest that there is only a 10% chance of something happening and it doesn't happen then you are technically, according to the model, 90% correct??? |
ejc, I do put trust in governments (even if I do often disagree with policies) and in The Economist. Never been a member of a lodge btw. Though I can understand how you could reasonably make that assumption given what I achieved in that sphere.
That was an interesting question re Assad and I will admit I am curious how it will score out, should find out in a few days.
Unless you know a better way Ejc, this project is the best way that I know to develop a proveable track record on predictions on global affairs. Absolutely and relatively in terms of others who made comparable predictions.
|
|
| Back to top |
|
|
explorerJC
Joined: 20 Oct 2005
Posts: 15242
Location: Farthingstone
|
| Posted: Thu Mar 02, 2017 9:55 am Post subject: |
|
|
| SloggingScotsman wrote: | | explorerJC wrote: | you are still a sucker for a name, Sloggers...first it was a couple of Tories, then it was your lodge....recently it has been the US and the economist...
The Economist has to fill its pages every week and once the US government funded the likes of Henry Kissinger, so, at face value, another former US backed project is nothing more than that to me...
i haven't read the model, but do I assume that if you suggest that there is only a 10% chance of something happening and it doesn't happen then you are technically, according to the model, 90% correct??? |
ejc, I do put trust in governments (even if I do often disagree with policies) and in The Economist. Never been a member of a lodge btw. Though I can understand how you could reasonably make that assumption given what I achieved in that sphere.
That was an interesting question re Assad and I will admit I am curious how it will score out, should find out in a few days.
Unless you know a better way Ejc, this project is the best way that I know to develop a proveable track record on predictions on global affairs. Absolutely and relatively in terms of others who made comparable predictions. |
what value to society is there in predicting whether Assad remains in power?
_________________
www.appliedtri.co.uk Tri and Du coaching
www.naturalrunningform.co.uk Natural Running Form Coach
2018 Training Camps http://www.appliedtri.co.uk/training-camps/
|
|
| Back to top |
|
|
SloggingScotsman
Joined: 18 Jul 2006
Posts: 2640
|
| Posted: Thu Mar 02, 2017 10:11 am Post subject: |
|
|
| explorerJC wrote: | | SloggingScotsman wrote: | | explorerJC wrote: | you are still a sucker for a name, Sloggers...first it was a couple of Tories, then it was your lodge....recently it has been the US and the economist...
The Economist has to fill its pages every week and once the US government funded the likes of Henry Kissinger, so, at face value, another former US backed project is nothing more than that to me...
i haven't read the model, but do I assume that if you suggest that there is only a 10% chance of something happening and it doesn't happen then you are technically, according to the model, 90% correct??? |
ejc, I do put trust in governments (even if I do often disagree with policies) and in The Economist. Never been a member of a lodge btw. Though I can understand how you could reasonably make that assumption given what I achieved in that sphere.
That was an interesting question re Assad and I will admit I am curious how it will score out, should find out in a few days.
Unless you know a better way Ejc, this project is the best way that I know to develop a proveable track record on predictions on global affairs. Absolutely and relatively in terms of others who made comparable predictions. |
what value to society is there in predicting whether Assad remains in power? |
ejc, come on man you are playing with me again.
First you take the proverbial over unproven prediction claims then you take the proverbial when I find a once government backed method of proving my claims.
Debating with you is a lose lose situation.
Why does every thread have to turn into knock big stu over the head with a hammer? I do doing my best to use and improve my skills to earn a living, you may not like it, but the government changed the rules and I have to comply.
If you are not involved with the good judgement project yourself Ejc, why not back off then perhaps if anyone else here who is using it might share experiences to mutual benefit.
So is your strategy to bully me off this forum with your persistent many months long persistent bashing, putting words in my mouth, making inaccurate assumptions, etc etc.
But to answer your question Ejc, the whole project is about finding people who can predict global events more accurately than most. The funny thing is that the years of once secret research proves iirc that housewifes and postmen can outperform intelligence officers, which when you think of all the spook stuff they have access to, is quite a revelation. Why that specific question, set in that particular way, you could only get an accurate answer from the project people that took over when the spooks commercialised it.
|
|
| Back to top |
|
|
explorerJC
Joined: 20 Oct 2005
Posts: 15242
Location: Farthingstone
|
| Posted: Thu Mar 02, 2017 10:52 am Post subject: |
|
|
So far on this thread, two questions have been asked by me, one to establish an understanding on the methodology and the other, more loosely, to identify its value.
Both questions were about the concept, not the person...
I made some slight mockery of your regular choices of source however, and i believe these are open to debate in terms of reliability and value....as are anyone else's in an open forum.
If you really believe that this is bullying, you are, of course, entitled to draw your opinion, however, I would put myself into the safe hands of the wider audience on that.
_________________
www.appliedtri.co.uk Tri and Du coaching
www.naturalrunningform.co.uk Natural Running Form Coach
2018 Training Camps http://www.appliedtri.co.uk/training-camps/
|
|
| Back to top |
|
|
Tigger
Joined: 25 Apr 2005
Posts: 2775
Location: Hampshire
|
| Posted: Thu Mar 02, 2017 4:52 pm Post subject: Re: Good Judgement Project |
|
|
| SloggingScotsman wrote: | | But I am very conscious that ten Predictions proves nothing. |
Absolutely agree with you there. And, to be honest, if the population is large enough then neither do 20.
|
|
| Back to top |
|
|
SloggingScotsman
Joined: 18 Jul 2006
Posts: 2640
|
| Posted: Thu Mar 02, 2017 5:14 pm Post subject: Re: Good Judgement Project |
|
|
| Tigger wrote: | | SloggingScotsman wrote: | | But I am very conscious that ten Predictions proves nothing. |
Absolutely agree with you there. And, to be honest, if the population is large enough then neither do 20. |
given that it has been over two decades since I sat a stats exam Tigger, do you [anyone with stats knowledge] know offhand what a reasonable number would be to statistically confident?
As questions can take a year and a half to resolve, this is a long term endeavour, but it would be nice to know at what point my stats become meaningful.
Obviously I am self selecting the questions I want to answer from the dozens to hundreds in the project (I haven't counted recently).
Once I have a statistically significant result (assuming I can achieve that) then there must be scope for earning some wedge from predicting global events.
|
|
| Back to top |
|
|
GrahamO
Joined: 10 Apr 2005
Posts: 10003
Location: United Arab Emirates or an airport
|
| Posted: Thu Mar 02, 2017 5:40 pm Post subject: |
|
|
I blame the NHS for their inability to keep him in his cell and properly medicated.
_________________
Arguing with engineers is like wrestling in mud with a pig - after a while you realise they like it
newMan : I don't know why you think you can address me like that! Don't ever reply directly to any of my posts again please.
|
|
| Back to top |
|
|
Tigger
Joined: 25 Apr 2005
Posts: 2775
Location: Hampshire
|
| Posted: Thu Mar 02, 2017 6:24 pm Post subject: Re: Good Judgement Project |
|
|
| SloggingScotsman wrote: | | do you [anyone with stats knowledge] know offhand what a reasonable number would be to statistically confident? |
I think it is more complicated than that.
Have a look at this article about why 225,000,000 is not enough if the wrong question is asked.
|
|
| Back to top |
|
|
SloggingScotsman
Joined: 18 Jul 2006
Posts: 2640
|
| Posted: Thu Mar 02, 2017 6:46 pm Post subject: Re: Good Judgement Project |
|
|
| Tigger wrote: | | SloggingScotsman wrote: | | do you [anyone with stats knowledge] know offhand what a reasonable number would be to statistically confident? |
I think it is more complicated than that.
Have a look at this article about why 225,000,000 is not enough if the wrong question is asked. |
thanks Tigger, I have skim read it for now. I do understand probability, standard deviations, variance, and how eg fund managers can look good through sheer chance and the law of big numbers etc. And I accept that I might just be experiencing positive variance at this time.
What I dont know is, assuming my trend continues, at what point would staticians say, how he is doing this might be worth looking at.
For the first time I am now able to show independently and mathematically that I am making better predictions than others comparably. (At this time a brier score of 0.156 compared to a comparable median of 0.4) Up until now all I could prove would be to tell you to talk to specific people who once laughed in my face.
|
|
| Back to top |
|
|
|
Share
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Albums
Memberlist
Bloglist
Wiki
Profile
Log in to check your private messages
Log in
