Annual CET competition - October update

Apologies for the delay. Here is the October table. No change at the top but Saint Snow and AP Snowman are creeping closer to the top 3.

TABLE

I know I didn't post enough for the annual competition, but I'm pleased my cumulative prediction error for Sep and Oct amounts to only 0.76C and on both occasions I posted early. 

No gloat. Just happy.

 

---

Annual CET competition - October update

Apologies for the delay. Here is the October table. No change at the top but Saint Snow and AP Snowman are creeping closer to the top 3.

IMAGE. Members enable at bottom of page

TABLE

19th.

To think I was actually in 2nd place in May.

---

Annual CET competition - October update

Apologies for the delay. Here is the October table. No change at the top but Saint Snow and AP Snowman are creeping closer to the top 3.

IMAGE. Members enable at bottom of page

TABLE

 

Cheers, GW 

---

Thanks for the update GW. It appears I'm a very average person this year - that makes a change from the usual abnormal tendencies .

The year's been far from it, though - the 1991-2020 long-term average would now be 22nd in the table, with a cumulative error of 11.76°C.

I wonder what the biggest impact month of the year will prove to be, in terms of how much it affects the leaderboard when omitted? Eyeballing it, February and April are strong contenders. For example, dropping February sends me up 6 places to 8th, while dropping April drops me 6 places to 20th!

So yeah, I've learned that April was the antidote to February for me, despite having still been a nearly 0.8°C error. Funny how things go sometimes.

---

Thanks for the update GW. It appears I'm a very average person this year - that makes a change from the usual abnormal tendencies .

The year's been far from it, though - the 1991-2020 long-term average would now be 22nd in the table, with a cumulative error of 11.76°C.

I wonder what the biggest impact month of the year will prove to be, in terms of how much it affects the leaderboard when omitted? Eyeballing it, February and April are strong contenders. For example, dropping February sends me up 6 places to 8th, while dropping April drops me 6 places to 20th!

So yeah, I've learned that April was the antidote to February for me, despite having still been a nearly 0.8°C error. Funny how things go sometimes.

One major trend in this competition is that the general standard of predictions has improved and the overall spread is less than it used to be. This year has been a very challenging one and where I have deviated from the mean prediction, the months have been well above or below average, but I went the wrong way! April and September were the ones that did it for me. Very well done to all these contestants near the top of the tree  

---

Good observations GezM 

 

I realised this morning that I'm able to answer my own question regarding impact of each month on the table. I summed the total change (regardless of direction) in rankings caused by omitting each month in turn. Results:

Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct
78	126	114	130	112	111	106	123	100	117

As you can see, by this measure, April had the biggest impact, followed by February, then August.

I did poorly in two of those but relatively well (still not great) in April. February was one of my widest ever misses, perhaps the widest, owing to the very cold spell not only ending about a week sooner than anticipated but being shortly followed by extremely mild weather.

Out of interest, I've been exploring what would happen if there was a 2°C error cap. Some interesting impacts which I'll present in early January. I had been thinking that it would give an unfair advantage to bold predictions, but then realised that it's also a cap on the maximum gain relative to others in the competition, so I think it balances out (not 100% on it though!).

---

Checking again. February was an awful month for me and many others too. Sucked in by the forecast of a long cold spell which was swept away by one of the most extreme winter warm ups in UK history! It's been an interesting year if you're trying to predict the weather!

---

Checking again. February was an awful month for me and many others too. Sucked in by the forecast of a long cold spell which was swept away by one of the most extreme winter warm ups in UK history! It's been an interesting year if you're trying to predict the weather!

 

 

Looking at the table to the end of October, looks like Windy Willow won the Feb month. He's propping everyone up now.

Frank H had a stormer of a start to the year, and was only about 0.6c cumulatively out at the end of March. He's still bang up there in contention - but will need the models to flip to an extreme Arctic output pretty quickly to stay there!!

---

If we look at the mean prediction error by month for all those who are currently still in the competition we find the following:

Jan 0.72C
Feb 1.70C
Mar 0.54C
Apr 2.07C
May 1.17C
Jun 0.45C
Jul 0.89C
Aug 1.01C
Sep 1.59C
Oct 0.85C

So yes April was easily the worst month and June was the best so far this year.

The other statistic I like to look at is who is the most consistently accurate predictor measured by number of months with a prediction error less than 0.5C. That doesn't necessarily equate to the leader in the competition because you could have a couple of terrible months but do really well in all the others and therefore not be near the top of the table.

A significant number of people have had 4 months with a prediction error less than or equal to 0.5C. Interestingly our leader SSWD only just squeezes out 3 months in that category with his October prediction error of exactly 0.5C. Second to sixth in the table all have 4 months with prediction errors under 0.5C, as do the ninth and eleventh placed people. The lowest ranked person with four predictions under 0.5C is marting in 20th place. The only person who has failed to achieve a prediction error under 0.5C this year is Darren S  who is usually near the top of the table. His smallest error is 0.51C in March.

 

 

 

Looking at the table to the end of October, looks like Windy Willow won the Feb month. *She's propping everyone up now.

Frank H had a stormer of a start to the year, and was only about 0.6c cumulatively out at the end of March. He's still bang up there in contention - but will need the models to flip to an extreme Arctic output pretty quickly to stay there!!

 

I believe I said at the time it's unlikely I would do so well for the rest of the year lol

I may even end up off the chart after this month, along with a few others, not quite the wooden spoon but very close 

Oh well it's all good fun 

 

Thanks for the chart GW  and all your efforts.

---

Apologies, Willow 

This place is so male-dominated, I just assumed ; I should have read your profile.

 

PS - didn't mean any harm with the propping up the table comment, was just an observation that Feb was a tough month and the person who won it had struggled in others. Now I'm waffling. Sorry. And hope it's not put you off the comp 

---

Apologies, Willow 

This place is so male-dominated, I just assumed ; I should have read your profile.

 

PS - didn't mean any harm with the propping up the table comment, was just an observation that Feb was a tough month and the person who won it had struggled in others. Now I'm waffling. Sorry. And hope it's not put you off the comp 

lol we're good  I don't tend to take offense easily  This is all just a bit of fun. 

---

---

 

lol we're good  I don't tend to take offense easily  This is all just a bit of fun. 

Can I take offence at "offence" having an "s" in it? 

 

Perhaps I should take part in this and then at least you'd have someone else underneath you! 

---

Thanks for the interesting inside stats GW . 0.5 is what I consider a very good result, 1.0 a decent one, based on the typical variability of the CET. By that measure, having 4 or more in a year is truly phenomenal, even if it is with a bit of luck along the way... which is inevitable, whoever you are - the CET is such a sensitive soul, sometimes completely switching moods at the slightest regional change!

 

---