December 2021 C.E.T. and EWP forecast contests -- start of the new contest year for 2021-2022

[damianslaw]

Another notably mild December. Becoming used to them now.. just about everything went wrong for potential cold weather. 

[The PIT]

Sunny Sheffield finished on 5.5C +0.7C above normal. Rainfall 111.2mm 127.1% of the monthly average

So a wet average month for us. Considering two record breaking days at 14C you would have thought it would have been higher.

[North-Easterly Blast]

December 2021 at 6.36 (6.4) CET was another pretty mild December.  I make it that in the last 100 years there were 12 warmer Decembers, and one with the same CET and 87 colder ones.  It did not make a top 20 warmest December ever, but it was still well above average despite some good signs for some cold weather.  Although there were some good building blocks with a fair amount of northern blocking in the second half, it still just did not come together to deliver the UK a significant cold spell.  I believe that the second half of December saw a fair amount of cold weather in NE Europe, but I do not know why it happens, the pattern did not quite come together to allow the cold air to reach the UK even though it wasn't far to the N and E of us.

[BornFromTheVoid]

A few records for the end of the month.
For the mean, both the 30th (11.4C vs 11.0C in 1834) and 31st (12.9C vs 11.1C in 1859) were record breakers. The 31st is also the joint 2nd warmest December day on record, beaten only by 13.1C on December 19th, 2015, and equal to December 12th, 1994.

The min was a record on the 31st (11.8C vs 9.6C in 1901) and the 2nd warmest minimum on record for any December day. This also means that the minimum on the 31st was warmer than the previous record mean for the day.
The max was a record on the 30th (13.9C vs 13.3C in 2015) and 31st (14.0C vs 12.4C in 1920), and a joint record on the 29th (13.8C also in 1925).

(edit due to typo spotted by RJS)

Edited January 1, 2022 by BornFromTheVoid

[damianslaw]

44 minutes ago, BornFromTheVoid said:

A few records for the end of the month.
For the mean, both the 30th (11.4C vs 11.0C in 1834) and 31st (12.9C vs 11.1C in 1959) were record breakers. The 31st is also the joint 2nd warmest December day on record, beaten only by 13.1C on December 19th, 2015, and equal to December 12th, 1994.

The min was a record on the 31st (11.8C vs 9.6C in 1901) and the 2nd warmest minimum on record for any December day. This also means that the minimum on the 31st was warmer than the previous record mean for the day.
The max was a record on the 30th (13.9C vs 13.3C in 2015) and 31st (14.0C vs 12.4C in 1920), and a joint record on the 29th (13.8C also in 1925).

To achieve a 2nd warmest min on record on the 31st is some feat! Gosh had the synoptics occured at the start of the month, it would have returned an even more extreme. I wonder whether the same synoptics would deliver record mins in other months of the year, or did we just strike perfect timing in terms of max impact at this time of year.

 

[Don]

6 hours ago, damianslaw said:

Another notably mild December. Becoming used to them now.. just about everything went wrong for potential cold weather. 

Yes a truly vile month and felt almost as bad as December 2015!

[summer blizzard]

December 2015 was stormy and wet, this was quite dry for a long stretch.

[Don]

1 minute ago, summer blizzard said:

December 2015 was stormy and wet, this was quite dry for a long stretch.

Still very dull and unpleasant though.

[North-Easterly Blast]

53 minutes ago, Don said:

Yes a truly vile month and felt almost as bad as December 2015!

 

Edited January 1, 2022 by North-Easterly Blast

[North-Easterly Blast]

1 minute ago, North-Easterly Blast said:

The December just gone was nowhere near as warm as December 2015 - it's overall CET was 3.3*C short, and it wasn't even one of the 20 warmest Decembers on record either.  I think the vile part of last month really was that there was a fair amount of northern blocking about, but the pattern just did not come together to allow the cold air to succeed in establishing over the UK, which is one of the worst aspects of the UK's weather patterns when that happens.

 

[Don]

1 minute ago, North-Easterly Blast said:

The December just gone was nowhere near as warm as December 2015 - it was 3.3*C short and it wasn't even one of the 20 warmest Decembers on record either.  I think the vile part of last month really was that there was a fair amount of northern blocking about, but the pattern just did not come together to allow the cold air to succeed in establishing over the UK, which is one of the worst aspects of the UK's weather patterns when that happens.

Will be hard pushed to get a December as warm as 2015 for some time, even with the warming climate.  Agreed ref to the northern blocking and not being able to get anything out of it, very frustrating!  

[North-Easterly Blast]

3 minutes ago, Don said:

Will be hard pushed to get a December as warm as 2015 for some time, even with the warming climate.  Agreed ref to the northern blocking and not being able to get anything out of it, very frustrating!  

The pattern was so extreme in December 2015 - just HP to the south-east of the UK and persistent low pressure to the NW of us from the month's start to finish, with virtually no day in that month with an airflow away from the sub-tropical Atlantic.  It was very much the "warm" extremity of the same month five years earlier.

[Roger J Smith]

All those daily records will of course find their way into the Almanac that I post but I won't put them in the existing posts since the Almanac is meant to indicate the records that exist going into each day. By the way the record high daily mean previously for 31st was in 1859. I had noticed that as part of a rather rare situation where three consecutive days had their record daily means from years prior to 1900 (30 Dec was in 1834, 31 Dec was in 1859 and 1 Jan was in 1851). Two of these have fallen and the third may also fall. But other than that three day stretch, the only other times that years from before 1900 held three or more consecutive records were these:

6-8 Jan (1898, 1890, 1858)

21-24 Jan (1796t1898, 1878, 1834, 1782)

(6-10 Feb the 8th was 1903 and all four of the others prior to that, so five in a row before 1904 there).

1-5 March (1878, 1777, 1777, 1859, 1834)

24-27 March (1776, 1777, 1777, 1777) and 23rd, 28th had ties from pre-1900 with 1945 and 1998, also 29th was pre-1900, 1813) ...

 ... so that would be a seven day interval where years before 1900 at least hold part of each record. 

27-30 April (1866, 1775, 1775, 1775) 

4-6 May (1834, 1800, 1867)

15-17 May (1833, 1808, 1833)

27-29 May (1788, 1847, 1780) after a tie on 26th (1784 t 2017). 

14-16 June (1814, 1858, 1858t1896) after a tie on 13th (1818 t 1989).

13-15 July (1808, 1808, 1825)

(31 Aug to 4 Sep all 1906 or earlier)

(6 to 10 Sep, 2021 broke up another string, on the 8th, previous to that it had been 1898, 1868, 1898, 1898, 1865).

25-27 Sep (1895, 1895, 1895)

27-29 Oct (1888, 1888, 1772 t 1984)

27-29 Nov (1818, 1828, 1818)

4-8 Dec (1888, 1898, 1898, 1856, 1856)

=====================================

May seem long a fairly long list but that is a bit more than half of the data (to 1906 it would be 135 of 250 years which is 54%).

Notice that the winter and spring months did better than summer or autumn in creating these ancient record strings. 

On the other side of the coin, the longest interval where no record from before 1900 survives would be 16 days, 25 July to 9 Aug, the oldest surviving records in that period are from 1948. Second longest is 16-26 Nov where the oldest records are from 1947. If today's record falls then we will be into a tied second longest such interval starting on 26 Dec and running to 5 Jan before a pre-1900 record on the 6th, the 24th and 25th Dec have older records too (24th from 1843 and 25th is a tie 1824, 1923)

 

 

 

 

[Don]

1 hour ago, North-Easterly Blast said:

The pattern was so extreme in December 2015 - just HP to the south-east of the UK and persistent low pressure to the NW of us from the month's start to finish, with virtually no day in that month with an airflow away from the sub-tropical Atlantic.  It was very much the "warm" extremity of the same month five years earlier.

Wasn't it just!

[Roger J Smith]

Almanac for 2nd January 

The almanac data will continue to appear in the Dec thread until 3rd Jan then go over to the January thread.

 

The highest daily mean (1772-2021) was 10.4 in 1948 which replaced 10.1 from 1935. The highest value since 1948 was 9.5 in 1992. Other high values included 9.8 in 1922, 9.7 in 1782 and 1921, 9.5 in 1916 (and 1992), and 9.3 in 1851.

The lowest daily mean (1772-2021) was -6.1 in 1786. Other low daily means were -5.0 in 1795, -4.6 in 1854, -4.5 in 1893, -4.1 in 1867, -4.0 in 1962, -3.8 in 1849, -3.6 in 1823, -3.5 in 1836, -3.3 in 1821, -3.2 in 1837 with -2.7 in 1979. The lowest value since 1979 was -1.8 in 1997 and 2002.

The highest maximum (1878-2021) was 12.9 in 1910 and 1976. Since then 11.8 in 1982 has been the warmest. Other mild days include  12.5 in 1948, 12.4 in 1916, 12.3 in 1925 and 1932, 11.9 in 1933, 11.8 in 1886, 11.7 in 1922 and 1935, 11.4 in 1981 and 11.3 in 1950.

The highest minimum (1878-2021) was 8.9 in 1921. Other mild minimum readings were 8.5 in 1935, 8.3 in 1948, 7.9 in 1922 and 1992, 7.8 in 1896, 7.4 in 1988 and 7.3 in 1917.  

The lowest maximum (1878-2021) was -1.8 in 1893 (1786 may have been colder by 1-2 deg). Other cold values were -1.3 (1941),  -0.2 (1946), -0.1 (1891), 0.0 (1993), 0.2 (1940), 0.3 (1963), 0.7 (1962), 0.8 (1997), 1.0 (1890) and 1.2 (1908). 

The lowest minimum (1878-2021) was -8.7 in 1962 (although 1820 may have been slightly lower). Other cold minima include -7.5 in 1979, -7.1 in 1893 and 1979, -6.2 in 1887, -5.0 in 1890, -4.9 in 1980, -4.7 in 1879 and 1940, -4.4 in 1997, and -4.2 in 1971.

The wettest 2nd of January (1931-2021) in the EWP data was 14.93 mm (1984), followed by 14.41 mm (1998), and 11.20 mm (2018). 

The wettest week ending 2nd January (27th Dec - 2nd Jan) was 70.37 mm in 2003, the only value over 50 mm -- the second highest was 49.71 mm in 1949.

The least amount falling in a week ending 2nd January was 0.01 mm in 1993, then 0.54 mm in 1989, 0.89 mm in 1992, 0.98 mm (2009),  1.84 mm in 2019, and 2.13 mm in 1955. 

-----------------------------

(Today's map link is for 1921, during a fairly persistent if unspectacular mild spell.) 

https://www.wetterzentrale.de/reanalysis.php?jaar=1921&maand=01&dag=02&uur=1200&var=1&map=1&model=noaa

Edited January 3, 2022 by Roger J Smith

[Roger J Smith]

The EWP tracker added nothing for 31st and finished at 102 mm. I will wait until the 5th and the confirmed value before releasing scores but you can check the recent post with those forecasts in the range of the most probable outcomes. 

[J10]

Monthly

As mentioned by Roger, no players were spot on this month.

However 2 were within 0.1c. richie3846 and Kirkcaldy Weather

In total 9 were within 0.5c

Seasonal and Overall

For the first competition month, the Top 10 largely matches the monthly result., with BLAST FROM THE PAST in 11th due to a 5pt late entry deduction.

December 21 CET.xlsx Dec 2021 Summary.pdf

[Roger J Smith]

Almanac for 3rd January 

The almanac data will move over to the January thread for the next instalment on 4th Jan.

 

The highest daily mean (1772-2021) was 11.6 in 1932 which (until 1st Jan 2022) was tied with two other dates, 23rd 1834 and 25th 2016, as highest January daily mean -- looks likely all three will drop to t-2nd as provisional for 1st Jan 2022 is 12.7 C. Other mild daily means for 3rd include 10.6 in 1948, 10.1 in 1992, 9.9 in 1917, and 9.7 in 1950.

The lowest daily mean (1772-2021) was -7.7 in 1795. Other low daily means were -7.6 in 1786, -6.8 in 1867, -4.7 in 1827, -4.6 in 1789, -3.6 in 1811, -3.5 in 1941 and 1979, -3.2 in 1893, -3.1 in 1971 and -3.0 in 1864. The lowest value since 1979 was -1.7 in 1997.

The highest maximum (1878-2021) was 12.7 in 1932 and 1948. Since then 12.2 in 1999 has been the warmest. Other mild days include 11.7 in 1916, 1917, 1957, 1992 and 2007 and 11.5 in 1983. 

The highest minimum (1878-2021) was 10.5 in 1932. This was the January record high value but may end up being replaced by a higher reading for 1st Jan 2022 (to be confirmed). Other mild minimum readings were 8.8 in 1950, 8.5 in 1948 and 1992, 8.3 in 1935, 8.1 in 1917 and 1944, 7.7 in 1981, and 7.4 in 1933. 

The lowest maximum (1878-2021) was -0.9 in 1941 (1786 and 1795 may have been around -3 to -5). Other cold values were -0.7 (1971), -0.1 (1894), 0.1 (1979), 0.6 (1879, 1893 and 1908), 0.9 (1946), 1.0 (1997), 1.2 (2010), 1.3 (1965) and 1.4 (1889 and 1963).

The lowest minimum (1878-2021) was -7.8 in 1962 (although 1786 and 1795 likely around -10 to -12). Other cold minima include -7.1 in 1979, -6.9 in 1893, -6.1 in 1993, -6.0 in 1940 and 1941, -5.6 in 1995, -5.4 in 1971, -5.2 in 1980, -5.0 in 1879, -4.7 in 1908, -4.5 in 1997, and -4.0 in 1965.

The wettest 3rd of January (1931-2021) in the EWP data was 16.20 mm (1983), followed by 13.31 mm (2016), 13.20 mm (1980), 12.93 mm (1994) and 10.01 mm (2014). 

The wettest week ending 3rd January (28th Dec - 3rd Jan) was 69.86 mm in 2003, followed by 55.42 mm in 1949, 51.65 mm (1998), 51.44 mm (2016), 50.32 mm (1994) and 50.31 mm (2014).

The least amount falling in a week ending 3rd January was 0.87 mm in 2009, followed by 1.55 mm in 1989 and 1.78 mm in 2019. Several long dry spells ended on this date and have left the list (notably 1991-92 and 1993). 

-----------------------------

(Today's map link is for 1932, one of the mildest days in January.) 

https://www.wetterzentrale.de/reanalysis.php?jaar=1932&maand=01&dag=03&uur=1200&var=1&map=1&model=noaa

 

Edited January 4, 2022 by Roger J Smith

[Roger J Smith]

EWP scoring for Dec 2021

My suspicions were correct, the adjusted table value is 106.4 mm (4.4 higher than the tracker).  

 

Rank _ Forecaster _________ Fcst __ Err ____ Score __ penalty, order of entry deductions

 

_01_ _ seaside60 ___________ 105 __ -1.4 ___ 10.00 

_02 _ _ Mapantz ____________ 110 __ +3.6 ____ 9.82 

_03 _ _ freeze _______________110 __ +3.6 ____ 9.70 __ 2nd ent 0.12

_04 _ _ Feb91Blizzard _______102 __ --4.4 ____ 9.47 

_05 _ _ MidlandsIceAge _____102 __ --4.4 ____ 9.35 __ 2nd ent, 0.12

_06 _ _ davehsug ___________ 102 __ --4.4 ____ 9.23 __ 3rd ent, 0.24

_07 _ _ AWD ________________ 101 __ --5.4 ____ 8.94

_08 _ _ February1978 ________112 __ +5.6 ____ 8.77

_09 _ _ Polar Gael ___________100.2 _ -6.2 ____ 8.60

_10 _ _ John88b _____________ 100 __ --6.4 ____ 8.41

_11 _ _ Booferking ___________ 99 __ --7.4 ____ 8.06^

_12 _ _ BlastFromThePast ___100 __ --6.4 ____ 7.99 __ 2nd ent, 1d late 0.42

_13 _ _ Sleety _________________98 __ --8.4 ____ 7.88

_14 _ _ Typhoon John ________115 __+8.6 ____ 7.70

_15 _ _ Emmett Garland _____ 97 __ --9.4 ____ 7.53

_16 _ _ Godber1 _____________116 __+9.6 ____ 7.34

_17 _ _ summer18 ___________ 96 __--10.4 ____ 7.16

_18 _ _ Ed Stone ______________95 __--11.4 ____ 6.98

_19 _ _ Mulzy _________________95 __--11.4 ____ 6.86 __ 2nd ent 0.12

_20 _ _ dancerwithwings _____ 95 __--11.4 ____ 6.74 __ 3rd ent 0.24

_21 _ _ Roger J Smith ________118.8 _+12.4 ____ 6.45

_22 _ _ Reef _________________ 119 __+12.6 ____ 6.28

_23 _ _ ProlongedSnowLover_119 __+12.6 ____ 6.16 __ 2nd ent 0.12

_24 _ _ Weather26 ___________ 120 __+13.6 ____ 5.94

_25 _ _ syed2878 _____________120 __+13.6 ____ 5.82 __ 2nd ent 0.12

_26 _ _ J10 ___________________ 120 __+13.6 ____ 5.70 __ 3rd ent 0.24

_27 _ _ Don __________________ 120 __+13.6 ____ 5.58 __ 4th ent 0.36

_28 _ _ Coldest Winter _______121.4 _+15.0 ____ 5.25 

_29 _ _ Let It Snow! ___________ 90 __--16.4 ____ 5.02^ 

_30 _ _ daniel* ________________91 __--15.4 ____ 4.89 __ 1d late 0.30

_31 _ _ virtualsphere _________ 89 __--17.4 ____ 4.68

_32 _ _ noname_weather _____88 __--18.4 ____ 4.51

_33 _ _ Stewfox ______________125 __+18.6 ____ 4.40

_34 _ _ JeffC __________________ 85 __--21.4 ____ 4.16

_35 _ _ Dr(S)No _______________85 __--21.4 ____ 4.04 __ 2nd ent 0.12

_36 _ _ Norrance _____________ 85 __--21.4 ____ 3.92 __ 3rd ent  0.24

_37 _ _ froze were the days ___85 __--21.4 ____ 3.80 __ 4th ent 0.36

_38 _ _ Jonboy ________________84 __--22.4 ____ 3.47

_39 _ _ Mr Maunder __________84 __--22.4 ____ 3.35 __ 2nd ent 0.12

_40 _ _ Stargazer ____________  83 __--23.4 ____ 3.13 

_41 _ _ Leo97T ______________ 130 __+23.6 ____ 3.01

_42 _ _ stevew _______________ 82 __--24.4 ____ 2.78

_43 _ _ BobD29 ______________ 81.6_--24.8 ____ 2.61

_44 _ _ StationaryFront ______ 132 __+25.6 ____ 2.50

_45 _ _ SteveB ________________79 __--27.4 ____ 2.27

_46 _ _ KirkcaldyWx __________ 77.6_--28.8 ____ 2.09

_47 _ _ weather-history _______77 __--29.4 ____ 1.92

_48 _ _ Frigid _________________ 75 __--31.4 ____ 1.72

_49 _ I Remember Atlantic252_140 __+33.6____ 1.63

_50 _ _ rwtwm ________________ 72 __--34.4 ____ 1.40

_51 _ _ snowray _______________ 72 __--34.4 ____ 1.28 __ 2nd ent, 0.12

_52 _ _ The Pit _________________ 70 __--36.4 ____ 1.08

_53 _ _ Neil N __________________ 66 __--40.4 ____ 0.90

_54 _ _ summer blizzard _______ 60 __--46.4 ____ 0.72

_55 _ _ shillitocettwo ___________59 __--47.4 ____ 0.54

_56 _ _ DiagonalRedLine _______ 50 __--56.4 ____ 0.36

_57 _ _ Earthshine ______________45 __--61.4 ____ 0.18

_58 _ _ Ross Andrew Hemphill _ 45 __--61.4 ____ 0.06 __ 2nd ent 0.12

-----------------------------------------------------------------------------------

^ two scores are generated from one lower scoring interval than the rank attained because the next score down included a late penalty, despite being generated from a higher scoring interval. 

Scores are generated from equal scoring levels, then adjusted down for additional deductions which may include those shown (order of entry, late one day) ... other deductions occur for rank to reward high ranking forecasts, from .01 2nd-3rd down to .10 after 20th. Then deductions are applied for increasing deviations from best forecast's error in percentage terms, these are .01 to .05 deductions in five categories, then a further deduction applies to "wrong sign" forecasts, to reward those who had the correct side of normal. That deduction runs from .01 to .06 in the range just on the wrong side of normal then stays .06 for the rest. You can see the effect of that where above normal forecasts (right sign) are scattered through larger groups of below normal forecasts with similar errors. The interval looks different because the higher forecasts are not penalized for wrong anomaly sign.

Order of entry deductions are not a penalty. Your score is generated from the first forecast at that value, so you only drop about 2/3 of a scoring interval per entry. Example, four forecasts at 85 mm all scored from the top scoring interval, by the time you reach the fourth entry, the drop is 0.36 instead of about 0.52 from the equal intervals. Those resume for the next forecast and score. 

The normal used is 1981-2010. If we get outcomes very close to normal in any month then the same principle applies but a more generous scale of deductions is applied. 

 

Edited January 5, 2022 by Roger J Smith

[Roger J Smith]

Best combined forecasts for Dec 2021 __

 (late forecasts rank 1 lower in CET for this determination, already factored into EWP ranking)

1. Freeze __________________ CET 4th __ EWP 3rd ____ 7 combined (t14 all time _ 4 yrs + 1 mo)

2. seaside60 ______________ CET 15th _ EWP 1st ____ 16 combined

3. Typhoon John _________ CET 14th _ EWP 8th ____ 22 combined

4. Blast from the Past ____ CET 11th _ EWP 12th __ 23 combined

5. AWD ____________________ CET 17th _ EWP 7th ___ 24 combined

6. syed2878 _______________ CET 3rd _ EWP 25th ___ 28 combined

[Dancerwithwings]

Hi Roger, regarding the EWP, you’ve stated me  95 - 11.4 - 6.74 __ 3rd ent 0.24 

it’s just to be honest, I’m not totally with you with this ...3rd ent 0.24

I’m taking it as....3rd entry and a points deduction of 0.24 because of this ?

If this is the case, I’m sure I only had one guess.

If I’m wrong, you have permission to call me stupid

 

Edited January 5, 2022 by Dancerwithwings

[Roger J Smith]

Hi dancerwithwings, the deduction is applied to the score of the first entrant at your forecast level. In the CET contest you would go down a full scoring interval. In the EWP contest the score is reduced by two thirds of a scoring interval for later entries at the same level. So it's not really a deduction applied to your forecast once we accept that order of entry is part of the scoring process. Meanwhile your average error is ranked as being equal to those of the earlier (and later) entrants. If the EWP contest was scored exactly the same way as the CET contest, your score would have been -0.34 relative to Ed Stone who was first in at 95 mm. So the 0.24 drop is from his scoring level (18th). Does that explain it better? Also, in case it seemed like this meant you had 6.74 minus 0.24, not the case, it's a score of 6.74 derived from the 6.98 minus 0.24. 

One other place where EWP scores can be more lenient than CET is that in the rare cases where two forecasts have the same error on opposite sides of the outcome, they are scored equally and I don't look at order of entry. But they might not stay equal if one is right side of normal and the other is not. 

Edited January 5, 2022 by Roger J Smith

[Dancerwithwings]

Not making excuses but It’s my first go at the EWP, didn’t realise one had to or best to submit both CET & EWP guesses an the same time.....but thanks for making it clear and the rest of what you said was very clear also, cheers  

[Roger J Smith]

Those order of entry deductions have been 0.12 for all recent contests and sometimes could have been slightly different if I went two-thirds of every scoring interval, this current contest had a scoring interval of 0.172 (approx). Back when we had a flood of forecasts due to the site promotion of the contest in Dec 2019, I went down to 0.05 as the scoring interval was 0.08. But otherwise it has remained 0.12 through a small range of scoring intervals .16 to .20. ... some forecast contests don't apply the order of entry principle and Joe Blow who entered with the same as John Doe but six days later would be tied in the standings. I wanted to make the EWP scoring a sort of mirror image of the CET contest which is why I brought in some other small deductions for error size, anomaly sign and close results. The principle is the same but instead of giving bonus points, I take away fewer of an otherwise standard reduction. As you move down the table, it is possible to see an overall deduction of 0.21 from the scoring interval (before any consideration of late penalties, those are applied to the net score). You could drop 0.10 for being ranked lower than 20th (a higher rank number that is), and 0.05 for having a forecast whose error percentage is more than 50% larger than the contest winner (who often has a very small error so basically if you are 50% out, you'll take the maximum penalty there), and 0.06 for being on the wrong side of normal. Here again there is a bit of protection for those in order of entry deduction situations. You only get the deductions for rank and sign that apply to the first level. In this case, as Ed Stone was 18th, all three of the 95 mm forecasts had .09 rather than .10 taken off for rank, and .02 instead of .02, .03 and .04 for anomaly sign (those deductions don't suddenly appear on the wrong side of normal, they come in six steps so that forecasts very close to normal are not considered as far off in that sense as those a long way from normal). So in net terms, your score is actually .13 higher than if we just went with scoring intervals by rank. Taking increased deductions may seem like a different approach until you realize that in the CET, the winner gets the most bonus points so it's really a mirror image if you say the actual winning scoring interval is for practical purposes closer to 110 than 100 there. In fact the CET scoring system may put a bit more emphasis on the bonus points since they are given some partial weight in the scoring system; I think to get a totally equivalent scoring system using my method I would probably need to use somewhat larger deductions and ease into them with an accelerated scale.

It could well be that in the scores you see this month, those around 8 would be closer to 7, those around 7 would be a bit below 6, and then from there down the scores would gradually converge on the same end point, so 4 would be about 3.5, 2 would be about 1.8, if this was an exact duplication of the CET scoring. The problem with trying for that is while CET errors are always units of 0.1 C, errors in this contest are on quite a flexible scale, some months see a lot of forecasts within 20 mm on either side like this month where consensus was one of the higher scores. Other months see a spectrum of errors between 20 and 100 mm. What is rare in the CET (an outcome like Dec 2015) is fairly common in this contest where few forecasts outside the 50-150 per cent of normal range are submitted, but outcomes are quite often outside those boundaries. There has been a bit of a scatter of points totals and average error, last year a few regular entrants were 10-15 ranks apart in the two different measures. But that is the same as the CET contest, the average error statistic does not read out in exact linear correlation with the order of finish. 

Another slight difference between contests is that the EWP contest is based on total score and has no average score component in scoring. If you have eleven good scores in the CET you would be sure to win; here you might finish 2nd-3rd with eleven good scores and you might have had some benefit from the 3 out of 10 for the missing forecast. Reef demonstrated that last year, he missed July, later said he would have gone with near normal, if he had entered that it would have only scored 3 or so (after late penalty) but those three points would have moved his rank from 9th to 6th. Conversely, you could win the EWP contest or at least rank high in it despite missing three contests, in the CET you are relegated from the Premier to the First Division or whatever they now call the second division after your third missing forecast. Some of those 9/12 or 8/12 entrants still generate enough points to finish mid-table but they aren't in the table any longer. 

[Roger J Smith]

17 minutes ago, Dancerwithwings said:

Not making excuses but It’s my first go at the EWP, didn’t realise one had to or best to submit both CET & EWP guesses an the same time.....but thanks for making it clear and the rest of what you said was very clear also, cheers  

Oh okay, I overlooked that aspect, but your order of entry was based on when you submitted the EWP forecast as you know, this is why in the table of entries it probably gives two numbers for order of entry. Sometimes also the opposite happens, somebody changes their CET but leaves their EWP forecast unchanged. So that retains its earlier order of entry for the EWP contest. Since that revision created a new order of entry and vacated the old one, I then take the former value minus 0.5 to maintain the same overall order, so let's say you entered ninth and revised your CET in the 51st entry, then your order of entry stats would be (51, 8.5) just in case I need to refer back to the EWP order; this is also shown in the summary below the main table of entries, which lists EWP forecasts in order of values, the shared forecasts are listed there in order of entry. 

No particular requirement to submit both together, it's just what most people do. There have been one or two revisions of EWP forecasts without touching the CET forecast as well. Some avoid the whole order of entry thing by adding a decimal which makes it very unlikely you will have any company. As to whether it's better to avoid duplicating or riding with a crowd, the statistical logic on that seems neutral. In your own case, you could have gone with 94.9 and taken the full drop, or 95.1 and become solo 18th, pushing the other two down to 19th and 20th. At random the two outcomes are equally likely, and (now that you see what's happening) you know when submitting if you happened to spot the duplicate entries, that this was the proposition, so you are saying, I will duplicate which means if Ed Stone is nth then I will have the points of (n+1)th for sure, whereas if I go 0.1 on either side, at this point I could be nth or (n+2)th. So these are long-term the same proposition on average. It makes no long-term difference to outcome if you avoid duplication or welcome it. But it does reward early entry; in the same set of propositions, Ed Stone can only finish nth or (n+1)st, except that if all duplicates follow the same procedure and separate out, then his outcomes become the same range as theirs. 

Don often finishes very well in these contests and is quite often in a duplicate forecast group, so it doesn't seem to be a hindrance. I have never duplicated and ... well it hasn't helped much.