I think its really only fair that we remain consistent and given the overflowing thread last year, I see not why we shouldn't have one this year. Now we know that the OPI turned out to be wrong on a scale that rivals Labor's economic policy but that in of itself is not a reason to completely discredit it just like a decent historic correlation was not a reason to credit it. 

Still the website seems to be down with a 404 service error: http://app.til.it/opi/

Perhaps the website being hosted on a car company's website was reason for concern after all. In any case I'm sure more people will be skeptical of it this time. Still, its worrying if this was a kid in a garage making stuff up because so many people were taken in by it (here and the other place).

 

---

I smelt the proverbial rat from the start. It was the claim of near perfect correlation set my nostrils flaring.

---

That claim could actually have been correct, the problem is; that is not impressive. I could construct a model that has a 100% correlation with historical winters just by trial and erroring a bunch of variables until they fit perfectly, but really all I have done is found a setup that works by chance and honed in on it like constructing a dartboard around a dart. 

 

---

After last years fiasco I think I'll give this predictive tool a wide berth, more so as I was one to champion it last year.

The OPI.... That's a name I've not heard in a long time... 

At least it gave me lots to talk about during my winter updates last year. 

---

I take it the OPI is going to be like the odd numbered star trek movies. We just don't talk about it! 

---

 

I'm very thankful the Tories turned 0.6 trillion of debt Labour left into £1.5 trillion with the forecast for it to increase year on year.

The important thing for both parties is that the bankers got their money and we pay for it.

You see the bankers create a bubble through massive fraud and manipulation of the markets - the size of the bubble represents the false value of the market and so when they burst the bubble they take the false value and we are left with the real value - the banker bailouts and national debts accrued as a result are pretty much the difference between the two values.

The value and money which was never really there is taken by the bankers and we pay the difference but for that to happen our government must borrow the money from the banks to pay for the bailouts (give them the money back) and transfer that debt onto the public books.

In actual fact the debt cannot be repaid and is not meant to  be - it is merely there as measure of financial control over the people.

After 6 years of austerity, that is massive unprecedented cuts to public services and lowering of the standard of living the debt has increased almost £1 trillion and is forecast to grow year on year for the foreseeable even with more draconian cuts to public spending. Within a few years we will be spending around the same on our own peoples welfare as we were while recovering from a global war and worldwide depression.

So even though austerity is plainly not working we will continue with the policy because for some reason people cannot fathom the left/right paradigm and how the British people have been sold out to globalist financiers and bankers.

The idea that Brown or Blair were left wing or socialist is as laughable as the idea the Tories are conservatives.

 

Meanwhile the OPI, yes the OPI, should not be dismissed because of last year but clearly as with all these possible background drivers and hindcast methods it needs to not be taken in isolation, it is just another measure to consider for those brave/foolish enough to make seasonal forecasts.

 

 

The authors never claimed a 100% success rate. They demonstrated (not just claimed, but demonstrated) a 90% historical correlation. So even as a hindcasting tool they acknowledged that one year out of every ten it would not work.

If something has a 90% historical correlation, but does not correlate the first time you try it, then binning it in a tantrum doesn't reflect well on the user imo.

If it fails to correlate this year and the next, then fine. It's clearly missing something major, though it would still have been a useful exercise in isolating which factor might be missing.

---

I think it was meant to be a serious attempt at a useful predictive tool based on the correlation between the polar vortex in October with the following winter values of the Arctic Oscillation which in turn has a correlation to UK winter temperature

As I recall they had an incomprehensively complex mathematical formula to determine the ellipticity of the PV (or degree of deviation from circular) in order to get the OPI. All the references from last year now turn up a 404 error which does not bode well

Anyway I have/had a graph of the OPI vs following AO which showed a poorish performance in 2013 and an epic fail in 2014 even though prior to 2013 they had a 90% correlation (all by hindcasting)

---

Exactly ...................a fools game

---

I never really got to the bottom of the OPI and what the situation was. There were suggestions of meetings and collaboration with Cohen but I can't recall whether these were verified. I recently Googled the topic and it seems there is very little information on it which suggests it may have been discontinued. The authors of the OPI are members here so perhaps they will post in this thread. A number of things have come and gone since TWO started and we may end up coming full circle to the Wax Wings etc. At least those reports provided some amusement. 

---

No theory should be discounted until it's been thoroughly tested. Retrospective yes, but especially prospective over several years.
Conclusions about the OPI theory can't be made for some time yet and shame on anyone discarding it as a joke after one year.
Saying that, I never really bought into the algorithm but if it's not the hot topic this silly season you can bet something else will come along to take it's place in the stack of straws we all seem to cling to.

---

Conclusions were made were made at the time it became the topic of the season last year. Claiming a probability of > 0.90 (I think it was 0.94?) in the method was pretty astonishing and to be frank, a tad outlandish. Had there been any basis of truth in that statistic, you would reasonably expect that winter 2014-15 would have been near exactly as was predicted by the OPI advocates.

 

 

---

OPI didnt seem to have much more funding that a University project. Therefore all the talk about why it was hosted on a car rental site, in my eyes is to be ignored. It had to be put somewhere cheap and accesible to all. Its likely someone donated some web-server space to achieve UAT.

The fact that it didnt do very well last year - again shouldnt be taken in isolation. The formula was a theory, and for that reason i wouldnt be surprised if it wasnt taken off line to study why it was so wide of the mark. You learn alot from problems, probably more so than if they got the results spot on.

Credit to the guys that did it - if it was a con, it was the most boring one ever, if it was a worthy science discovery, then it perhaps needs to be reviewed and give it another go. Its gotta be better than other LRFs which flip flop every month.

We need to encourage ventures like this, especially going on the budget they appear to not have.

---