2021 North Atlantic hurricane forecast contest

[Roger Smith]

Found these seasonal forecasts for reference (only the most recent version noted here):

CSU _____ 18 8 4 (June 3rd)

NOAA ___ 13-20 6-10 3-5 (May 20th)

 _ median _ 16.5 / 8 / 4

UKMO ___ 15* 7 3 (May 20th _ added 1 since they state 14 7 3 for official season)

NCSU ___ 15-18 7-9 2-3 (Apr 14)

_ median 16.5 / 8 / 2.5

TWC ____ 19 8 4 (May 13)

TSR _____ 18 9 4 (May 27)

UA ______ 18 8 4 (Apr 13)

avg of all _ 17 8 4

Will place that in the contest as "expert" since it comes pretty close to matching most of them anyway.

However, will also enter "NOAA median" as a separate entry. 

Our consensus so far appears to be closer to 20 10 5 (and our consensus is usually a bit more bullish than the experts, which works out well in active years and badly in low years). 

Last year if you didn't enter this or read the thread, we were all obviously low on named storms (30) although some had very good hurricane and major forecasts. I had to score by a beta method when the official contest scoring saw even the best calls reduced to zero with that absurdly high named storm count. 

So, this year, the official contest scoring will follow that precedent just in case we go off on a bender again with every swirl of cloud getting a name. Scoring as follows:

base score 100

_ less one quarter of (error of named storms) + (error squared of named storms)

_ less one half of (error of hurricanes) + (error squared of hurricanes)

_ less one half of (error of majors) + (error squared of majors)

... example follows ...

Your forecast 19 10 5 

Actual count 22 11 3

Your score = 100 - (0.25)* (3+9) -- (0.5)*(1+1) -- (0.5)*(2+4) = 100 - 3.0 -- 1.0 -- 3.0 = 93.0

(before it would have been 90.0)

... in a case more similar to last year ...

Your forecast 19 10 5

Actual count 28 11 3

... your score 100 - 0.25* (9+81) --1.0 --3.0 = 100 --22.5 --1.0 --3.0 = 73.5

(before it would have been 51.0).

________________________________________________

If this scoring system makes you consider any change in forecast, I won't be collecting forecasts for a table of entries until June 11th.

The contest deadline will be June 11th 06z (end of Thursday June 10th). Late entries will be docked one point per day, squared. (1, 4, 9 etc).

That means there is no point in entering beyond June 20th as your late penalty is then 100. (this late penalty will be worked out on an hourly basis from the 06z deadline). 

 

 

[Prospero]

7 hours ago, Roger Smith said:

So, this year, the official contest scoring will follow that precedent just in case we go off on a bender again with every swirl of cloud getting a name. Scoring as follows:

base score 100

_ less one quarter of (error of named storms) + (error squared of named storms)

_ less one half of (error of hurricanes) + (error squared of hurricanes)

_ less one half of (error of majors) + (error squared of majors)

 

I love mathematical formulas almost as much as a good storm! Funny how "math" connects ALL things.

But I am "John Dee", Prospero of Shake-Speare/Bacon...

[Cat Lady]

18/9/3

[Tezeta]

On 6/5/2021 at 8:30 PM, Prospero said:

I love mathematical formulas almost as much as a good storm! Funny how "math" connects ALL things.

But I am "John Dee", Prospero of Shake-Speare/Bacon...

Who

[Tom]

19/9/3

[hudsonvalley21]

21/11/6

[wxdude64]

19/7/4

[Looking to the skies]

19/8/7

[bkviking]

20/9/4

Ace 145

[WhirlingWx]

18/9/4

[Roger Smith]

The deadline for entries has now arrived. Late entries are accepted but with a late penalty of (days late)^2 (or days late squared), with the "days late" determined by hourly precision. Therefore if you read this within the first day of the late period, your penalty will be quite small (the contest is scored out of 100 as explained in a previous post). By day two you are looking at a rather significant late penalty increasing from 1 to 4 points and on day 3, from 4 to 9 points. 

This is a table of entries received. "Expert consensus" refers to a group of pre-season outlooks from noted hurricane research groups, all of which came up with forecasts close to this expert consensus. "NOAA median" refers to the average values of the ranges predicted by NOAA. The contest consensus is based on forecaster entries and does not include the expert consensus or NOAA median. The consensus is the mean of all forecasts submitted. 

The forecasts are listed from most to least named storms, and within ties, the numbers of hurricanes or majors will determine the order; if nothing separates the forecasts at all then order of entry is used. 

Table of forecasts __ 2021 hurricane seasonal contest

FORECASTER _________________ named __ 'canes __ Major 

Macintosh _______________________25 __ 14 __ 10

Tezeta ___________________________23 __ 14 ___ 7

Roger Smith _____________________ 23 __ 13 ___ 7

Southmdwatcher ________________ 22 __ 11 ___ 6

IntenseWind002 _________________ 22 ___ 9 ___ 4

hudsonvalley21 __________________21 __ 11 ___ 6

Newman ________________________ 21 __ 10 ___ 5

J_Keith_Lee _____________________ 21 ___ 9 ___ 4

Brian5671 _______________________ 20 __ 12 ___ 2

RJay ____________________________ 20 __ 10 ___ 5

WxWatcher007 __________________ 20 ___ 9 ___ 4

BKViking ________________________ 20 ___ 9 ___ 4

 

___ consensus __________________19.2 _ 9.8 __ 4.7

 

DonSutherland1 _________________ 19 __ 12 ___ 6

Ldub ____________________________ 19 __ 11 ___ 5 

magpiemaniac __________________ 19 __ 10 ___ 4

NCforecaster89 _________________ 19 ___ 9 ___ 4

Tom ____________________________ 19 ___ 9 ___ 3

Looking to the skies _____________ 19 ___ 8 ___ 7

wxdude64 _______________________19 ___ 7 ___ 4

TexMexWx ______________________ 18 ___ 9 ___ 4

Cat Lady ________________________ 18 ___ 9 ___ 3

snowlover2 ______________________18 ___ 8 ___ 4

Iceresistance ____________________18 ___ 7 ___ 3

Rhino16 _________________________ 17 __ 10 ___ 4

cptcatz __________________________17 ___ 9 ___ 4

NorthHillsWx ____________________17 ___ 9 ___ 3

____ "Expert consensus" _________ 17 ___ 8 ___ 4

TARCweather ___________________ 17 ___ 8 ___ 3

____ NOAA median _____________16.5 ___ 8 ___ 4

Yoda ___________________________ 16 ___ 9 ___ 5

LoboLeader1 ____________________15 ___ 9 ___ 4

Prospero ________________________14 __ 11 ___ 7

_____________________________________________________________

Posting at 0300z 11th, three hours ahead of the deadline. 

Will check for any amended forecasts or new entries within the deadline and incorporate them as required. 

Our consensus is just slightly higher than the expert consensus and actually quite similar to many of the individual forecasts within that consensus which has members at 18 and 19 named storms. 

While almost everyone found a unique forecast, one forecast did overlap an earlier one, in that one case, the earlier forecast would be considered the winner if those numbers are the best forecast. 

Thirty forecasts received so far, plus the two additional entries. 

Good luck to all !

 

[WxWatcher007]

Where’s @Windspeed?

[Windspeed]

Where’s [mention=845]Windspeed[/mention]?

I'll give it another go next season. Good luck!

[ncforecaster89]

9 minutes ago, Windspeed said:

17 minutes ago, WxWatcher007 said:

Where’s [mention=845]Windspeed[/mention]?

I'll give it another go next season. Good luck!

I vote to extend the deadline another hour for windspeed, without any penalty, if he were to wish to make an entry and his preferred numbers haven’t already been taken; which I presume may be the case.   
 

?

 

Edit: It appears there’s still almost two hours remaining for new entries without any penalty.  

[Windspeed]

Wasn't a time issue. I chose not to enter. Not that I would get lucky enough to repeat but someone else should win. Good luck.

[WxWatcher007]

3 hours ago, Windspeed said:

Wasn't a time issue. I chose not to enter. Not that I would get lucky enough to repeat but someone else should win. Good luck.

[Roger Smith]

The seasonal count has reached 8/3/1 with Grace briefly a major before (final) landfall and Henri now a minimal 'cane.

(note Aug 27, Ida has reached hurricane status and most forecasts assume it will become a major, which would make the count 9/4/2 at that point. The table is edited as of today to reflect that status).

(note Aug 29, Julian has come and gone as a tropical storm, table now edited again to reflect that, with the count 10/4/2).

(note Aug 30, Kate has formed and seems unlikely to become a hurricane, so I have redone the table yet again for 11/4/2).

(note Sep 2, Larry is already a hurricane and guidance seems bullish on its chances to become a major so with the count now 12/5/2 I have rejigged the table below for 12/5/3 (if Larry fails to become a major this will be adjusted later).

(note Sep 8, Mindy has arrived in the Gulf of Mexico, expected to remain only a tropical storm ... count once again adjusted). 

(note Sep 12, Nicholas active in the Gulf, waiting to see if this one becomes a hurricane before editing the table).

(note Sep 14, Nicholas did attain hurricane status near landfall, table now updated to reflect new count of 14/6/3).

The seasonal predictions you made now require these future outcomes to verify:

Required future tropical counts for seasonal predictions to verify

_ This table will be updated as new data change the seasonal count _

_ Aug 27 table changed to reflect likely 9/4/2 count within 1-2 days.

_ Aug 29 and 30 changes to 11/4/2.

_ Sep 2 change to 12/5/3 

_ Sep 8 change to 13/5/3

_ Sep 14 count is 14/6/3 

 

FORECASTER _________________ named __ 'canes __ Major _____ additional named, H and M required

Macintosh ______________________ 25 __ 14 __ 10 __________ 11 ___ 8 __ 7

Tezeta __________________________ 23 __ 14 ___ 7 ___________ 9 ___ 8 __ 4

Roger Smith _____________________23 __ 13 ___ 7 ___________ 9 ___ 9 __ 4

Southmdwatcher ________________22 __ 11 ___ 6 ___________ 8 ___ 5 __ 3 

IntenseWind002 ________________ 22 ___ 9 ___ 4 ___________ 8 ___ 3 __ 1

hudsonvalley21 __________________21 __ 11 ___ 6 ____________7 ___ 5 __ 3

Newman ________________________ 21 __ 10 ___ 5 ___________ 7 ___ 4 __ 2

J_Keith_Lee _____________________ 21 ___ 9 ___ 4 ___________ 7 ___ 3 __ 1

Brian5671 _______________________ 20 __ 12 ___ 2 ___________ 6 ___ 6 __(-1) (6 6 0 best outcome)

RJay ____________________________ 20 __ 10 ___ 5 ___________ 6 ___ 4 __ 0

WxWatcher007 __________________20 ___ 9 ___ 4 ___________ 6 ___ 3 __ 1

BKViking ________________________ 20 ___ 9 ___ 4 ___________ 6 ___ 3 __ 1

 

___ consensus __________________19.2 _ 9.8 __ 4.7 __________ 5.2_3.8 _1.7

 

DonSutherland1 _________________ 19 __ 12 ___ 6 ____________ 5 ___ 6 __ 3 (6 6 3 best outcome)

Ldub _____________________________ 19 __ 11 ___ 5 _____________5 ___ 5 __ 2

magpiemaniac ___________________ 19 __ 10 ___ 4 ____________ 5 ___ 4 __ 1

NCforecaster89 _________________ 19 ___ 9 ___ 4 _____________5 ___ 3 __ 1

Tom _____________________________ 19 ___ 9 ___ 3 _____________ 5 ___ 3 __ 0

Looking to the skies _____________ 19 ___ 8 ___ 7 _____________ 5 ___ 2 __ 4 (533 best possible)

wxdude64 _______________________ 19 ___ 7 ___ 4 _____________ 5 ___ 1 __ 1

TexMexWx _______________________ 18 ___ 9 ___ 4 _____________ 4 ___ 3 __ 1

Cat Lady _________________________ 18 ___ 9 ___ 3 _____________ 4 ___ 3 __ 0

snowlover2 ______________________ 18 ___ 8 ___ 4 _____________ 4 ___ 2 __ 1

Iceresistance ____________________ 18 ___ 7 ___ 3 ______________ 4 ___ 1 __ 0

Rhino16 __________________________ 17 __ 10 ___ 4 _____________ 3 ___ 4 __ 1 (441 best possible)

cptcatz ___________________________17 ___ 9 ___ 4 _____________ 3 ___ 3 __ 1

NorthHillsWx ____________________ 17 ___ 9 ___ 3 ______________ 3 ___ 3 __ 0

____ "Expert consensus" _________ 17 ___ 8 ___ 4 _____________ 3 ___ 2 __ 1

TARCweather ____________________ 17 ___ 8 ___ 3 _____________ 3 ___ 2 __ 0

____ NOAA median _____________ 16.5 ___ 8 ___ 4 ____________ 2.5 __ 2 __ 1

Yoda _____________________________ 16 ___ 9 ___ 5 _____________ 2 ___ 3 __ 2 (best outcome 3 3 2)

LoboLeader1 _____________________ 15 ___ 9 ___ 4 _____________ 1 ___ 3 __ 1 (3 3 1 best outcome)

Prospero _________________________ 14 __ 11 ___ 7 ______________0 ___ 5 __ 4 (best possible 4 4 4)

____________________________________________________________________

As of Kate (Aug 30) now three forecasts are impossible to verify but those three can still win the contest by hitting their best possibles. With Larry now a major, a fourth forecast would fall into this category. And with Mindy's arrival three more have drifted into the Cannot Fully Verify Narrows. Most can still win the contest though if they hit their best possibles. Nicholas did not add any new members to the "cannot fully verify" category but drew even with our lowest prediction of number of storms. We are still one short of the lowest number of hurricanes predicted, and had already seen one forecast passed by the number of majors with three others now level at 5. 

 

(Aug 22) If we assumed that one third of the season has occurred and two equal thirds lie ahead, the outcome would be 

24 9 3

If we assumed that 40% of the season has now occurred but that the remaining 60 per cent would be twice as productive of hurricanes and majors, the outcome would then be 20 9 3

The mean of these two assumptions is 22 9 3. 

(Aug 27) If we now assume 45% of season has occurred and remaining 55 per cent would be 1.8 times as productive of hurricanes and majors, the new target is then 20 9 4.5 

(Sep 2) If Larry became a major and the count was 12 5 3 and we assumed this was the halfway point of the season with an equal second half, the outcome would be 24 10 6. Only one forecast goes higher on named storms, several have higher H and M components. 

(Sep 8) The forecast models are not too bullish on additional tropical developments in this peak portion of the season so a more subjective outlook now might be around 21 11 5. Just my speculation though. 

[Prospero]

Quote

Prospero ________________________14 __ 11 ___ 7 ______________ 6 ___ 8 __ 6 (best possible 7 7 7)

LOL!

I knew I went low on named storms. A few surprised me; Anna, Bill, Claudette, and Danny. Elsa, Fred, and Grace were maybe in my numbers. Henri is another surprise.

So for me to win, almost every named storm will be hurricane and mostly a major. Figuring late September and October start, I was likely way off. August did fire up a bit.

Good thing I am not a met with people depending on my predictions!

 

 

[Roger Smith]

The table produced last week showing the counts you need now to reach your forecast targets has been updated on the assumption that Ida does make it to major (the first two numbers in this revised count are already verified). This table will be adjusted if Ida does not become a major, and also at any point that the count changes again (with two likely candidates out there this could be soon although I might wait for those to show their potential before doing the adjustments). (added later _ currently 10/4/2 with Julian, waiting to see what happens to TD 10 and these two both seem likely to remain only TS intensity, but will adjust the table after they are done with their work). (later _ Julian has already been declared post-tropical after a brief spell as a TS -- will now update the table for this known element).

Some of our forecasts look very promising at this point with quite plausible "rest of season" counts to verify. 

[Roger Smith]

I have now updated the table a few posts above to reflect the arrival of Larry, and the table assumes Larry will become a major, which can be edited if this is not the case by the weekend. Based on the fact that September 7 is the statistical midpoint of the hurricane season, the eventual count would be 24/10/6 if the second half resembled the first half, and would be closer to 24/12/7 if the second half was equal in storm production but more potent in development.

Quite a few forecasts remain very viable at this point and it's almost anybody's race to win at this point. 

[Roger Smith]

I have updated the "storms now required" post a few posts back in the thread, for Mindy's brief appearance. 

A growing number of forecasts are now somewhat adrift as their numbers cannot fully verify, but I don't think very many forecasts are eliminated despite that, and it's going to be a very competitive end to the contest. I mentioned in that post that with a rather lackluster tropical outlook in the current models at this peak time of the season, perhaps 21 11 5 is the most likely finish we might expect (but it's just speculation). 

Count is now 13/5/3. 

(edit Sept 12)

With the appearance of Nicholas, the count is 14/5/3 -- will wait to see if Nicholas happens to become a hurricane before updating the table of "storms required to verify."

[Roger Smith]

Nicholas did reach hurricane status around landfall last night, so the new count is 14/6/3. Here's a copy of the table edited above for the number of storms each forecaster now requires to meet their predictions. That table will be locked and this new one will see the next round of edits. 

Odette, Peter and Rose have changed the count to 17/6/3 with just slight chances of any of them becoming hurricanes (Odette is not necessarily done because part of its circulation may redevelop tropical status, but even so that looks only tropical storm at most which would change nothing). ...

 

Required future tropical counts for seasonal predictions to verify

 

_ This table will be updated as new data change the seasonal count _

_ Sep 14 count is 14/6/3 (see previous version for earlier counts)

_ Sep 18 count is 15/6/3 (Odette came and went although there's some chance of a part two).

_ Sep 20 count is 17/6/3 (Peter and Rose both tropical storms, little chance of hurricane status apparently).

_ Sep 24 count is 18/6/3 forecast to become 18/7/4 within days (this has now verified). 

_ Sep 25 count is 19/7/4 after the brief non-rampage of Teresa.

_ Oct 2 count is 20/7/4 as Victor remained only at tropical storm intensity at its peak. 

 

FORECASTER _________________ named __ 'canes __ Major _____ additional named, H and M required

Macintosh ______________________ 25 __ 14 __ 10 ___________ 5 ___ 7 __ 6 (best outcome 7 7 6)

Tezeta __________________________ 23 __ 14 ___ 7 ___________ 3 ___ 7 __ 3 (best outcome 6 6 3)

Roger Smith _____________________23 __ 13 ___ 7 ___________ 3 ___ 6 __ 3 (best outcome 6 6 3)

Southmdwatcher ________________22 __ 11 ___ 6 ___________ 2 ___ 4 __ 2 (best outcome 4 4 2)

IntenseWind002 ________________ 22 ___ 9 ___ 4 ___________ 2 ___ 2 __ 0

hudsonvalley21 __________________21 __ 11 ___ 6 ____________1 ___ 4 __ 2 (best outcome 3 3 2)

Newman ________________________ 21 __ 10 ___ 5 ___________ 1 ___ 3 __ 1 (best outcome 3 3 1)

J_Keith_Lee _____________________ 21 ___ 9 ___ 4 ___________ 1 ___ 2 __ 0 (best outcome 2 2 0)

Brian5671 _______________________ 20 __ 12 ___ 2 ___________ 0 ___ 5 __(-2) (4 4 0 best outcome)

RJay ____________________________ 20 __ 10 ___ 5 ___________ 0 ___ 3 __ 1 (2 2 1 best outcome)

WxWatcher007 __________________20 ___ 9 ___ 4 ___________ 0 ___ 2 __ 0 (2 2 0 best outcome)

BKViking ________________________ 20 ___ 9 ___ 4 ___________ 0 ___ 2 __ 0 (2 2 0 best outcome)

 

___ consensus __________________19.2 _ 9.8 __ 4.7 _________- 0.8_2.8 _0.7 (3 3 1 best outcome)

 

DonSutherland1 _________________ 19 __ 12 ___ 6 ___________-1 ___ 5 __ 2 (4 4 2 best outcome)

Ldub _____________________________19 __ 11 ___ 5 ___________ -1 ___ 4 __ 1 (3 3 1 best outcome)

magpiemaniac ___________________ 19 __ 10 ___ 4 __________ -1 ___ 3 __ 0 (2 2 0 best outcome)

NCforecaster89 _________________ 19 ___ 9 ___ 4 ___________ -1 ___ 2 __ 0 (2 2 0 best outcome)

Tom _____________________________ 19 ___ 9 ___ 3 ___________ -1 ___ 2 __-1 (2 2 0 best outcome)

Looking to the skies _____________ 19 ___ 8 ___ 7 ___________ -1 ___ 1 __ 2 (1 1 1 best outcome)

wxdude64 _______________________ 19 ___ 7 ___ 4 ___________ -1 ___ 0 __ 0 (0 0 0 best outcome)

TexMexWx _______________________ 18 ___ 9 ___ 4 ____________-2 ___ 2 __ 0 (0 0 0 best outcome)

Cat Lady _________________________ 18 ___ 9 ___ 3 ____________ -2 ___ 2 __-1 (0 0 0 best outcome)

snowlover2 ______________________ 18 ___ 8 ___ 4 ____________ -2 ___ 1 __ 0 (0 0 0 best outcomes)

Iceresistance ____________________ 18 ___ 7 ___ 3 _____________-2 ___ 0 __-1 (0 0 0 best outcome)

Rhino16 __________________________ 17 __ 10 ___ 4 _____________-3 ___ 3 __ 0 (2 2 0 best outcome)

cptcatz ___________________________17 ___ 9 ___ 4 _____________-3 ___ 2 __ 0 (1 1 0 best outcome)

NorthHillsWx ____________________ 17 ___ 9 ___ 3 ______________-3 ___ 2 __-1 (1 1 0 best outcome)

____ "Expert consensus" _________ 17 ___ 8 ___ 4 ______________-3 ___ 1 __ 0 (1 1 0 best outcome)

TARCweather ____________________ 17 ___ 8 ___ 3 _____________ -3 ___ 1 __-1 (1 1 0 best outcome)

____ NOAA median _____________ 16.5 ___ 8 ___ 4 ____________-3.5 __ 1 __ 0 (0 0 0 best outcome)

Yoda _____________________________ 16 ___ 9 ___ 5 _____________-4 ___ 2 __ 1 (best outcome 0 0 0)

LoboLeader1 _____________________ 15 ___ 9 ___ 4 _____________-5 ___ 2 __ 0 (0 0 0 best outcome)

Prospero _________________________ 14 __ 11 ___ 7 ______________-6 ___ 4 __ 3 (best outcome 2 2 2)

____________________________________________________________________

As of Kate (Aug 30) three forecasts were impossible to verify but those three could still win the contest by hitting their best possibles. With Larry a major hurricane in early September, a fourth forecast fell into this category. And with Mindy's arrival three more drifted into the Cannot Fully Verify Narrows. Most can still win the contest though if they hit their best possible outcomes as shown. Nicholas did not add any new members to the "cannot fully verify" category but drew even with our lowest prediction of number of storms. We are still one short of the lowest number of hurricanes predicted (7 by wxdude64 and Iceresistance), and had already seen one forecast passed by the number of majors (Brian5671, 2) with five others now level at 3. 

Various projections were made in the previous version of the table. Assuming the season is now 60% complete and that the second portion will match that for relative intensity, the projected end point would be (rounded off) 24 10 5 which implies a further count of 10 4 2. An assumption of higher intensity by one third would yield 24 12 6, a further count of 10 6 3. My subjective estimate is that the season may end around 24 13 5. (10 7 2 yet to come) ... the bottom three forecasts are just about eliminated from any chance to win at this point by the progression of possible outcomes that match their seasonal count, or beyond. By 16 the highest possible count is 16 8 5 so Yoda could win if that happened and then nothing else happened later. That would require storms 15 and 16 to be major hurricanes. By the time we reach 17 with any combination of possible H and M, the three forecasts at 14-16 are eliminated as each outcome would have a lower total error count with a higher-storm-number forecast. (by 17 the possible outcomes are these:

The possible outcomes now from 19 7 4 would be 

then 20 7 4, 20 8 4, 20 8 5.

and 21 7 4, 21 8 4, 21 8 5, 21 9 4, 21 9 5, 21 9 6

_________________ 

With the brief existence of TS Odette, three more forecasters have drifted into the shallows where their numbers cannot verify, but probably none of them are yet eliminated from any chance of winning the contest. After Peter and Rose, many more (most of the contest field) found themselves in the same shallows, so it's beginning to look like a few who still have possible verifying numbers are moving into the lead now. As the projected additional values for Sam are 1 1 1 there were no new forecasts pushed into the cannot verify shallows. (there could have been if anyone had been sitting at 110 required, or 100 required, but nobody was there). Teresa however has finished the search for perfection for six more forecasts, and there are now only three forecasts left standing (although they are not the only three that could win the contest) -- wxdude64 is currently at his predicted numbers of 19 7 4.

Note: new feature below is current and projected scoring for today's count and any possible future counts for the next two storms (if there are two more). 

[Tezeta]

Come to papa

[Roger Smith]

Tracking actual scores as the contest approaches its end phase

The scoring system was explained earlier, basically you lose only half as many points from your base score of 100 for named storm errors as you lose for hurricane and major errors.

The named storm error reduction is one quarter the value of (error) + (error squared). For example, you lose 1.5 for an error of 2 ... ( 2 + 4 ) / 4

That same error for hurricanes or majors would cost you three points ( 2 + 4) / 2. 

To make the scoring a bit easier to follow, I will arbitrarily assign 40 to named storms and 30 each to hurricanes and majors. You can go into negative scoring territory, but no forecast is currently there. This is the same as applying all three reductions to 100 but gives us a visual comparison for the three categories. 

These scores are now in scoring order for the current count.

Tracking the scores as we approach the end, you will see your current score from the known count (sometimes including 2-4 day forecast developments which will be tagged and adjusted if they don't come to pass). Then off to the right of your current score will be potential scores for the next two systems in whatever combination they appear -- they can only add 1 0 0, 1 1 0, 1 1 1, then 2 0 0, 2 1 0, 2 1 1,  2, 2, 0, 2 2 1 or 2 2 2 so there will be nine potential scores. And of course, there could be more than two more systems so even this won't tell you everything about final outcomes (by November it might). 

note: scores in the table currently based on 21/7/4 after Wanda's recent designation. ...

also 21/8/4 score was retained in the table in case Wanda became a hurricane (it did not but I left the column in for your interest, or in case there's a last minute re-classification about that or any other system in the contest year) ...

those would become current contest scores if so and 21/7/4 scores would be removed from table.

__ __ FC = future counts (possible) for remaining 22 storm outcomes

___ ___ ___ current contest scores are in green (first total scoring column)

 

FORECASTER __________ Predictions __ Scores (40 30 30 TOT) 2174 (2184) __ 2274 2284 2285 2294 2295

 1 wxdude64 _____________19 ___ 7 ___ 4 _____ 38.5 _30 __30__ 98.5 _ (97.5)___ 97.0_ 96.0_ 95.0_ 94.0_ 93.0_

 2 J_Keith_Lee ___________ 21 ___ 9 ___ 4 ______40 __ 27 __30__ 97.0_ (99.0) ___ 96.5_ 98.5_ 97.5_ 99.5_ 98.5_ 

t3 IntenseWind002 _______22 ___ 9 ___ 4 _____39.5 _27 __30 __ 96.5_ (98.5) ___97.0_ 99.0_98.0_100.0_99.0

t3 WxWatcher007 _______ 20 ___ 9 ___ 4 ______39.5 _27 __30__ 96.5_ (98.5)___ 95.5_ 97.5_ 96.5_ 98.5_ 97.5_ 

t3 BKViking ______________ 20 ___ 9 ___ 4 _____ 39.5 _27 __30__ 96.5_ (98.5) ___ 95.5_ 97.5_ 96.5_ 98.5_ 97.5_ 

t5* snowlover2 ____________18 ___ 8 ___ 4 ______ 37 __29 __30__ 96.0 _ (97.0)___ 94.0_ 95.0_ 94.0_ 94.0_ 93.0_ 

t5* Iceresistance __________18 ___ 7 ___ 3 _______ 37 __30 __29__ 96.0 _ (95.0)___ 94.0_ 93.0_ 91.0_ 91.0_ 89.0_   

 7 NCforecaster89 _______19 ___ 9 ___ 4 _______38.5 _27 __30 __ 95.5_ (97.5)___ 94.0_ 96.0_ 95.0_ 97.0_ 96.0_ 

 8 Tom ___________________19 ___ 9 ___ 3 _______ 38.5 _27 __29 __ 94.5_ (96.5)___ 93.0_ 95.0_ 93.0_ 96.0_ 94.0_ 

 9 TexMexWx ___________ 18 ___ 9 ___ 4 _______ 38.5 _27 __30 __ 94.0 _ (96.0)__ 92.0_ 94.0_ 93.0_ 95.0_ 94.0_ 

(t 9) _ "Expert consensus" _17 ___ 8 ___ 4 _______35 __ 29 __30 __ 94.0 _(95.0)___ 91.5_ 92.5_ 91.5_ 91.5_ 90.5_ 

t10 Newman _____________ 21 __ 10 ___ 5 ________40 __ 24 __29__ 93.0_ (96.0) ___ 92.5_ 95.5_ 96.5_ 97.5_ 98.5_

t10 CatLady ______________ 18 ___ 9 ___ 3 _______ 37 __ 27 __29__ 93.0 _(95.0) ___ 91.0_ 93.0_ 91.0_ 94.0_ 92.0_

t10 TARCweather _________ 17 ___ 8 ___ 3 _______ 35 __ 29 __29__ 93.0_ (94.0) ___ 90.5_ 91.5_ 89.5_ 90.5_ 88.5_ 

____ NOAA median _____ 16.5 ___8 ___ 4 ________ 36.1__29 __30__ 92.8 _(93.8)___ 90.1 _ 91.1_ 90.1 _ 90.1_ 89.1_ 

__ consensus __________ 19.2 _ 9.8 __ 4.7 _____ 38.7_24.7 _29.4__ 92.8_(95.6)___ 91.4_ 94.2_ 94.6_ 96.0_ 96.4_ 

t13 RJay _________________20 __ 10 ___ 5 _______ 39.5 _24 __29 __ 92.5_ (95.5) ___ 91.5_ 94.5_ 95.5_ 96.5_ 97.5_ 

t13 magpiemaniac _______19 __ 10 ___ 4 _______ 38.5 _24 __30 __ 92.5_ (95.5) ___ 91.0_ 94.0_ 93.0_ 96.0_ 95.0_ 

15 cptcatz ________________17 ___ 9 ___ 4 _______ 35 __ 27 __30 __ 92.0 _(94.0)___  89.5_ 91.5_ 90.5_ 92.5_ 91.5_ 

16 Looking to the skies __ 19 ___ 8 ___ 7 _______ 38.5 _ 29 __24 __ 91.5_ (92.5)___ 90.0_ 91.0_ 94.0_ 90.0_ 93.0_

17 NorthHillsWx _________ 17 ___ 9 ___ 3 ________ 35 __ 27 __29 __ 91.0 _(93.0)___ 88.5_ 90.5_ 88.5_ 91.5_ 89.5_ 

18 Rhino16 _______________ 17 __ 10 ___ 4 _______ 35 __ 24 __30 __ 89.0 _ (92.0)___ 86.5_ 89.5_ 88.5_ 91.5_ 90.5_ 

19 Yoda __________________ 16 ___ 9 ___ 5 _______ 32.5 __27__29 __ 88.5 _(90.5)___ 85.5_ 87.5_ 88.5_ 88.5_ 89.5_ 

20 Ldub ___________________19 __ 11 ___ 5 ________ 38.5 _ 20__29 __ 87.5_ (91.5) ___ 86.0_ 90.0_ 91.0_ 93.0_ 94.0_ 

21 hudsonvalley21 _______ 21 __ 11 ___ 6 ________ 40 __ 20__ 27 __ 87.0_ (91.0) ___ 86.5_ 90.5_ 92.5_ 93.5_ 95.5_

t22 LoboLeader1 __________15 ___ 9 ___ 4 ________ 29.5__27__30 __ 86.5 _(88.5)___ 83.0_ 85.0_ 84.0_ 86.0_ 85.0_ 

t22 Southmdwatcher _____ 22 __ 11 ___ 6 ________ 39.5 _20 __27 __ 86.5_ (90.5) ___ 87.0_ 91.0_ 93.0_ 94.0 _96.0

24 Brian5671 _____________ 20 __ 12 ___ 2 ________ 39.5 _ 15 __27 __ 81.5_ (86.5) ___ 80.5_ 85.5_ 82.5_ 89.5_ 86.5_ 

25 DonSutherland1 ________19 __ 12 ___ 6 ________ 38.5 _ 15 __27 __ 80.5_ (85.5) ___ 79.0_ 84.0_ 86.0_ 88.0_ 90.0_ 

26 Roger Smith ___________ 23 __ 13 ___ 7 ________ 38.5 __ 9 __24 __ 71.5_ (77.5) ___ 72.5_ 78.5_ 81.5_ 83.5 _86.5

27 Prospero _______________14 __ 11 ___ 7 __________26___ 20 __24 __ 70.0 _(74.0)___ 66.0_ 70.0_ 73.0_ 73.0_ 76.0_ 

28 Tezeta _________________ 23 __ 14 ___ 7 ________38.5 __ 2 __24 __ 64.5_ (71.5) ___ 65.5_ 72.5_ 75.5_ 78.5 _81.5

29 Macintosh _____________ 25 __ 14 __ 10 _______ 35 ___ 2 ___ 9 __ 46.0_ (53.0) ___ 48.0_ 55.0_ 61.0_ 61.0 _67.0

 (scoring S H M counts) ______________________________________2174 _ (2184) _____2274_2284_2285_2294_2295_

reductions for error (H, M 30) 1 3 6 10 15 21 28 (S 40 minus 0.5 1.5 3.0 5.0 7.5 10.5 14.0)

* note: ranks for non-member forecasts (NOAA, expert consensus, and consensus) do not change contest ranks. 

* note also ranks after t3 (three forecasts) jog by one position upward, as two of the forecasts in that group of three are identical, so as a contest entrant points out, further ranks should perhaps be for distinct forecast entries rather than numbers of people. So there are 29 ranks now for 30 forecasts. 

Future possible count scoring leaders in bold type. 

_________________________________________________________________

THIS TABLE WILL BE UPDATED AFTER EACH NEW STORM. 

Oct 31 update _ The count is now 21 7 4 as shown, and remaining options for 21 8 4 and five possible outcomes for 22 storms are now in the table. The situation for counts of 23 can be reviewed in another post (scroll down) at this point. 

[Tezeta]

Im so confused at that explanation but know i am crushing most of you in this contest. 

[Roger Smith]

On 9/24/2021 at 1:40 PM, Tezeta said:

Im so confused at that explanation but know i am crushing most of you in this contest. 

Not yet, could eventually but you and I have almost the same forecasts and we're nearly bottom of the scoring table until we do get up around 23 named storms, then the problem will be that from 19/7/4 the most optimistic outcome would be 23/11/8 and a more realistic one is 23/9/5. From 23/11/7 (or even 24/11/7) our scores would be 3-4 still behind two lower named storm forecasts who would have smaller error totals on H/M. The named storm errors only cut in half as much as the H and M errors. If we get to 25/11/7 then we do get into a narrow lead. For you to pass me, there would have to be 14 or more hurricanes, our errors are the same on the other two, and realistic counts that reach 14 hurricanes go well above the 25 named storms from McIntosh who has a higher major prediction so you can still finish ahead of McIntosh up to around 27/14/8, the H count will not matter but above 27/8 on the others would favor McIntosh. We are of course nowhere near those numbers yet and probably won't add enough H or M for us to be too concerned about it anyway although we could easily verify our 23 as best sub-forecasts. 

I have an excel file to back up the scoring (confessing that I did the first table in my head then decided to call in Mr Excel to help me out. It was relatively easy to calculate scores for all options that can still verify (from 19/7/4, for example 25/14/10 of McIntosh cannot quite verify and our two forecasts cannot exactly verify as we need six more hurricanes from four named storms. So for the possible options of 21 and 22 named storms, you can see in the table above who would win the contest if that's where the numbers end up. For the 23 options this is what I see as possible and the highest scores calculated by the excel program ...

For the last set of options I show scores for myself and Tezeta as we begin to close in on the leaders at that point. 

 

22 7 4 ___ IntenseWind007 22 9 4 (97.0) ______________ 23 7 4 ___ IntenseWind007 22 9 4 (96.5)

22 8 4 ___ IntenseWind007 22 9 4 (99.0) ______________ 23 8 4 ___ IntenseWind007 22 9 4 (98.5)

22 8 5 ___ IntenseWind007 22 9 4 (98.0) ______________ 23 8 5 ___ IntenseWind007 22 9 4 (97.5)

22 9 4 ___ IntenseWind007 22 9 4 (100.0) _____________ 23 9 4 ___ IntenseWind007 22 9 4 (99.5)

22 9 5 ___ IntenseWind007 22 9 4 (99.0) _______________23 9 5 ___ IntenseWind007 22 9 4 (98.5)

22 9 6 ___ Newman 21 10 5 (97.5) ______________________ 23 9 6 ___ IntenseWind007, Newman, JKeithLee (tied 96.5)

22 10 4 __ IntenseWind007 22 9 4 (99.0) ______________ 23 10 4 ___ IntenseWind007 22 9 4 (98.5) 

22 10 5 __ Newman 21 10 5 (99.5) _____________________ 23 10 5 ___ Newman 21 10 5 (98.5)

22 10 6 __ SouthMDwatcher 22 11 6 (99.0) ____________ 23 10 6 ___ SouthMDwatcher 22 11 6 (98.5)

22 10 7 __ SouthMDwatcher 22 11 6 (98.0) ____________ 23 10 7 ___ SouthMDwatcher 22 11 6 (97.5)

_______________________________________________________23 11 4 ___ IntenseWind, SouthMD tied (96.5) RS 91 Tez 88

_______________________________________________________23 11 5 ___ SouthMDwatcher (98.5) RS 94 Tez 91

_______________________________________________________23 11 6 ___ SouthMDwatcher (99.5) RS 96 Tez 93

_______________________________________________________23 11 7 ___ SouthMDwatcher (98.5) RS 97 Tez 94

_______________________________________________________23 11 8 ___ SouthMDwatcher (96.5) RS 96 Tez 93

(so Tezeta, we are blocked by SouthMDwatcher all the way to the top of the 23 named storm pile although we gradually get closer to the top).

Above 23 named storms, you would figure the count almost surely must add some H and M so I won't bother to list these yet, and the plausible ones start probably around 24/10/5. 

(edit Oct 2, the fact that Victor did not become a hurricane eliminates the 22 10 x and 23 11 x options from the possible outcomes). 

(edit Oct 31, now that Wanda has arrived, could possibly become a hurricane but unlikely to be a major hurricane, 22 9 6, 23 10 7 options are eliminated). 

[Tezeta]

7 minutes ago, Roger Smith said:

Not yet, could eventually but you and I have almost the same forecasts and we're nearly bottom of the scoring table until we do get up around 23 named storms, then the problem will be that from 19/7/4 the most optimistic outcome would be 23/11/8 and a more realistic one is 23/9/5. From 23/11/7 (or even 24/11/7) our scores would be 3-4 still behind two lower named storm forecasts who would have smaller error totals on H/M. The named storm errors only cut in half as much as the H and M errors. If we get to 25/11/7 then we do get into a narrow lead. For you to pass me, there would have to be 14 or more hurricanes, our errors are the same on the other two, and realistic counts that reach 14 hurricanes go well above the 25 named storms from McIntosh who has a higher major prediction so you can still finish ahead of McIntosh up to around 27/14/8, the H count will not matter but above 27/8 on the others would favor McIntosh. We are of course nowhere near those numbers yet and probably won't add enough H or M for us to be too concerned about it anyway although we could easily verify our 23 as best sub-forecasts. 

Your system is kinda sus

[StormchaserChuck!]

lol, 18,18.17,19

[Roger Smith]

The last two posts appear to be somewhat in coded format, meanwhile I added some further info to my last post to show who would win the contest at all currently possible outcomes of 22 and 23 named storms. Some outcomes are no longer possible, you can't add more hurricanes or majors than named storms to 19/7/4 (unless Peter came back to life as a hurricane). 

I am happy to answer any formulated questions about the scoring but it was explained in June and the reason why named storms are not given as large an error rating as the other two categories is because last year the equal weighted system we used in all other years crashed out and only one forecaster even had a positive total because of the huge error stats for the named storm portion. So I devised this less punitive (but still expanding exponentially as I think it should) system ... you do of course realize that the 40-30-30 category scoring is just for illustration, I could have made it 30-35-35 and come up with the same totals, or stuck to 100 minus total error. Also I hope people realize that the first set of scores only apply to the current situation and are not contest "standings" in any other sense. 

[Roger Smith]

Now that the New England storm has become STS Wanda, we are at 21/7/4. Later today I will edit changes into previous posts about scoring, removing now-excluded possible outcomes. Wanda could become a hurricane (not the current NHC prediction) but seems unlikely to be a major hurricane so that 21/7/4 and 21/8/4 scoring will remain as well as options for 22/7/4, 22/8/4 and 22/8/5 and 22/9/4, 22/9/5 -- previous scoring tables are now updated.