2014 NATL Tropical-Hurricane-Seasonal Forecast Contest

[PSUBlizzicane2007]

October:  1/0/0

[metalicwx366]

October: 1/1/1

[GaWx]

Post your October monthly forecasts before 06z Wed Oct 1st.

I will update the scoring table located a few posts back, on the assumption that September finishes 2/1/1. The scores for actual entrants are generally not too bad, the highest was 15/16 for Andyhb (subject to any last minute revisions).

Roger/Folks,

There's a small chance that a subtropical storm (i.e., named) near Bermuda could be officially designated by tomorrow, meaning a potential effect on September for the contest depending on whether or not you're including subtropical storms.

Edit: The rules refer to "named storms" as opposed to "tropical storms". Since subtropical storms are a type of named storm, an argument can be made that these would count the same as tropical storms. It is ultimately Roger's call, of course. Opinions? Even if one doesn't form now, this would be informative for whenever one does form.

[Roger Smith]

Seems like this one isn't going to happen. If one does happen, I would have to ask around, what would people expect? When was the last time there was a named subtropical storm? I seem to remember there was one in 1978.

[wxmx]

Last year there was an unnamed Subtropical Storm...it was unnamed because it was found in post season analyisis that it acquired subtropical storm status. If they get subtropical storm characteristics they are named in operational forecasting. Last year featured Subtropical storm Beryl and Chris before they were upgraded to Tropical status.

 

Last occurrence of a purely subtropical featured storm (and named as well) without later acquiring tropical characteristics was Subtropical Storm Andrea in 2007.

[CandymanColumbusGA]

1/1/0

[Eskimo Joe]

0/0/0

[wxmx]

2/0/0

[Roger Smith]

Just for your amusement and entertainment, I have added in some provisional seasonal scores based on 7/5/1 which assumes that we will see two more named storms, one of which becomes a hurricane. That seems close to our current consensus anyway. As a reminder, the seasonal score is based on a max of 50 points and errors reduce that by the average of error and error squared. Example, if your seasonal storm count is off by 4, then the error reduction is 10 (from 16+4 divided by 2).

 

Based on that, Pluffmud had the highest score (48/50) for the original seasonal estimate of 8/4/1. As he has not been entering monthly forecasts, the highest provisional score overall is from Gawx who has not posted an October forecast yet, assuming that he(?) does then unless it's way off consensus, his score would likely continue to lead the contest (Nov-Dec is only worth 2/50). Andyhb is quite close behind on current provisional numbers.

 

Of course, if 2/1/0 additional is considerably underdone, then some of the people a bit higher up in the table would begin to do better at the expense of the higher scores below them. I don't think there is much hope for Normal, UncleW or myself but if there is, then we will have one hell of an October.

 

OTOH, if there's no further activity, all but the last two or three scores shown in the table would drop (the lower scores by more than the higher ones due to the increasing bite of the error squared factor) and PSUBlizzicane2007 might edge ahead of GaWx.

 

We shall see.

[GaWx]

October: 1/1/0

[PSUBlizzicane2007]

If I had just stayed with my original seasonal prediction I probably would have won lol

[Roger Smith]

Do you mean that in terms of the monthly forecastss you provided? That could be so. If you mean the actual score for your seasonal, you do get to keep the original score, the rules allowed updates that would count only if they scored higher after time penalties.

 

I think that system might have had some validity if people were adjusting to higher seasonal numbers by several in each category, the math does not work out well for low seasonal numbers that are adjusted slightly. Let's say it was a more normal season, and you had wanted to adjust 7/5/1 to 12/8/4. That would have been mathematically a good move as late as July 28th if 12/8/4 were correct, and even later if reality came in above that.

 

Anyway, you could still finish first I think, if your October forecast is on the money and if my provisionals are too high (actually you would probably finish first if we have a shutout or perhaps 1/0/0 now to end).

 

For next year, I think I will blend the two systems. What I liked about this system was that it pretty much counted on the pre-season forecast for all the points unless you made a significant adjustment early on. But if you allow any in-season adjustments, they should probably count on their own merits, the system used before probably gave those adjustments too much weight, so I would go with something like 15% for July 1 and 10% August 1 (weighted).

[GaWx]

Anyway, you could still finish first I think, if your October forecast is on the money and if my provisionals are too high (actually you would probably finish first if we have a shutout or perhaps 1/0/0 now to end).

 

For next year, I think I will blend the two systems. What I liked about this system was that it pretty much counted on the pre-season forecast for all the points unless you made a significant adjustment early on. But if you allow any in-season adjustments, they should probably count on their own merits, the system used before probably gave those adjustments too much weight, so I would go with something like 15% for July 1 and 10% August 1 (weighted).

 

Roger,

 Based on my own calculations (assuming I fully understand your scoring system) and assuming your listed month by month #'s through Sep. are correct, I have PSU winning if we go 0/0/0 from here to the end. However, I think that Andy and myself would tie for the win if we get 1/0/0 in Oct. and nothing afterward and be one point ahead of PSU. Please correct me if I'm wrong. I was looking at this last night.

[Roger Smith]

I think you're correct, but I have not done the calculations for the 2 points available for Nov-Dec which almost everyone usually predicts 1/0/0 and gets right.

 

If that also went 0/0/0 and you went 1/0/0 then you would fall another point behind, so probably your estimate is correct.

 

However, I doubt that October will actually go entirely storm-free, and you never know when another 1887 could come out of the blue (a whole season worth of storms after mid-October including December). It's not like 1887 was in some long-ago warm period either, the 1880s were damned cold in general. And there wasn't much solar activity, the 1883 peak was weaker than 2012 or whatever this one turns out to be. Yet 1886 and 1887 had prodigious tropical seasons. Go figure. 2005 had more of a season from Wilma to the end than we're having this year.

 

If you accept that anything is possible, my 3/2/1 October would then require a 5/2/0 Nov-Dec to reach my seasonal forecast. ... but there are some holding around 10/4/1 now who just need a boatload of weak tropical storms to come from behind and win this thing. Would not bet against them yet, this is the decade of weak tropical storms.

 

You three are in a somewhat tight space to win the contest, if my math is correct, you would have to enter and be correct for Nov-Dec to ensure a win if you were leading after October, and if October goes higher than 2/1/0 then other contestants begin to enter the picture, especially if Nov-Dec produces an oddball result. So be sure to enter that last phase, and avoid getting a provisional score that would be too low to guarantee that you stay ahead. If that provisional score was half of 1.0 (0.5) it might not be enough. Although I think the rules state that we go back to half-size penalty points for Nov-Dec so 1.0 would be scored by an actual entrant who was one off in two categories rather than just one.

 

BTW, Stebo and Ncforecaster have made monthly forecasts which, if correct in either case, would place them in the lead overall.

[wxmeddler]

Whoops:

1/0/0

[PSUBlizzicane2007]

Yeah just my original seasonal forecast which was with five named storms

[GaWx]

Roger,

 Based on my own calculations (assuming I fully understand your scoring system) and assuming your listed month by month #'s through Sep. are correct, I have PSU winning if we go 0/0/0 from here to the end. However, I think that Andy and myself would tie for the win if we get 1/0/0 in Oct. and nothing afterward and be one point ahead of PSU. Please correct me if I'm wrong. I was looking at this last night.

 

 The odds of 0/0/0 are now very low as STS Fay may be declared very soon. (It looks to start as a STD at 11 AM). So, a STS Fay would knock out PSU and be in Andy's as well as my favor among others.

[Roger Smith]

There are two points on the table for NOV-DEC and a correct forecast of an unusual outcome there could change the scores.

 

Let's just wait and see how the scoring unfolds, meanwhile, Fay has come into existence and I've decided that it will count whether it stays "sub-tropical" or becomes a tropical storm. It has a name and the contest was based on the premise of named storms, hurricanes and major hurricanes. If it becomes a sub-tropical storm of hurricane intensity then we have a discussion ahead. That hasn't happened before, has it? (barring the landfall stages of Sandy being a matter of debate about what kind of storm it was).

[Roger Smith]

The count is now officially 2/1/0 for OCT and appears to be heading for either 3/2/0 or 3/2/1 according to some model projections and the more certain risk of Gonzalo attaining at least hurricane status.  The season then is now at 7/5/1 heading for perhaps 7/6/1 in a few days and possibly 7/6/2 or even 8/6/2 (8/7/2 ?)

 

Based on that, I have updated the provisional seasonal scores to an estimate of 8/6/1 for now. Those are back about one page in the scoring table.

 

(edit Oct 14) -- That table has been updated to show seasonal scores for 8/6/2. A new table has been added later in this thread showing provisional total scores for 9/7/2 and 3/2/1 for OCT.

[ncforecaster89]

For the benefit of full disclosure, I wanted to go with 2/1/0 for this month, but chose 2/0/0 as my best chance of possibly overtaking Stebo since he had already posted 2/1/0, himself (and he entered this month with a very slight lead over me).  That said, it now looks as though we both (more me than him) will come up short with this unexpected burst of activity that is likely to exceed our expectations on TC intensity-as Roger noted in the preceding post.

[GaWx]

For the benefit of full disclosure, I wanted to go with 2/1/0 for this month, but chose 2/0/0 as my best chance of possibly overtaking Stebo since he had already posted 2/1/0, himself (and he entered this month with a very slight lead over me).  That said, it now looks as though we both (more me than him) will come up short with this unexpected burst of activity that is likely to exceed our expectations on TC intensity-as Roger noted in the preceding post.

 

I could be mistaken, but it looks to me like Stebo would likely win if we get in Oct. to 3/1/0, 3/2/0, 3/1/1, 3/2/1, or 3/2/2, among other things.

[ncforecaster89]

I could be mistaken, but it looks to me like Stebo would likely win if we get in Oct. to 3/1/0, 3/2/0, 3/1/1, 3/2/1, or 3/2/2, among other things.

 

Thanks for the post, GaWx.  I must admit that I haven't reviewed the various statistical possibilities since I posted my October forecast a couple of weeks ago.  At that time, it appeared that Stebo and I were in the best position to win the contest-hence my need to offer a slightly different projection to possibly catch him.  That said, I didn't think 3/1/0 or anything higher than 2/1/0 (which was the expectation I had for this month) would've materialized. It now appears I will very likely be wrong!  I appreciate you catching my error in suggesting that Stebo might also come up just short, when in fact, he appears to currently be in the best position to win this thing-as you so thoughtfully pointed out.

Edit:  Apologies to Stebo for not accurately characterizing his current position in the contest.

[Roger Smith]

If you're looking at possible scores, or if not, please keep in mind these two minor points:

 

1. The original rules (4) stated that for monthly scores, a minimum progression would be in place for any cases that ended up otherwise low-scoring for the field. That minimum progression means that highest score in OCT can be no lower than 8.0, second highest can be no lower than 7.0, third 6.0, fourth 5.0 and from there on down at intervals of 0.5 ... if the scores as normally calculated are higher than those minimum values at any point in the progression, the rule still applies further down the order. This could apply if we get to 3/2/1 or anything higher despite the fact that I would happen to score 10 on that. The scores for "Normal 1989-2014" and "Consensus" are not counted as entries to establish ranking, but receive whatever score they would have gained as entries. Ties would count in total rather than in placement, as one example, if three people tied for fourth and they were all boosted to 5.0 from some lower value, the next highest score would be eligible to be increased to 3.5 rather than 4.5.  This will likely all be academic if there is no third named storm in October.

 

2. In post 85 I clarified that June, July and NOV-DEC would be scored with half-scale error points (notice, not OCT). This would mean that one error point would reduce the available 2.0 to 1.5 which would then go to 1.6 by the minimum scoring rule above. It's quite possible that an oddball result in Nov-Dec would require the use of the minimum progression rule again, and it would run 1.6, 1.4, 1.2, 1.0 then 0.9, 0.8 etc (with ties factored in so it might involve quite a few breaks in that progression).

[Roger Smith]

Scoring to end of October (edited Nov 2 )

 

-- This table will be updated as greater certainty develops on actual scoring outcomes. There will still be 2 points available for your NOV-DEC forecasts. Those forecasts are now incorporated and provisionally scored for an outcome of 0/0/0 consistent with the provisional seasonal scores in this table.

 

-- This table excludes max landfalling forecasts, see the previous table in post 122. ... set at 100% to prevent line overflow.

 

 

This table is (now) based on 0/0/0 in NOV-DEC for an 8/6/2 seasonal outcome. November-Decembers are now scored for that outcome,  0/0/0 worth 2.0, 1/0/0 1.5, 1/1/0 scores 1.0. ... Total scores in brackets represent the "what if they had issued all monthly predictions as listed" comparison which essentially doubles all provisional scores marked * after monthly values that appear before the scores in each case. Official contest results are the first of the two projected scores. You can check the seasonal scores (worth 50/100 of the total) that appear with the seasonal forecasts to compare them separately. There's a table at the bottom of this post for comparisons of total monthly scoring.

 

Easy summary -- these scores are what you have now, if nothing else happens, including the scores you would receive for NOV-DEC forecasts.

 

FORECASTER _________ SEASONAL____JUNE___JULY___ AUG____SEP____OCT____NOV-DEC

 

___________________ S _ H _ M ____ SHM___SHM____SHM____SHM_____SHM_____ SHM_____ TOTALS

 

Normal 1989-2013____ 14_7_3 27.0 __ 100 3.5__100 5.5__421 8.0__531 7.0__32110.0__000 2.0____63.0

 

 

Uncle W ____________ 14_7_1 27.0 __ 100 3.5__310 4.5__530 5.0__42112.0__110 5.0__000 2.0____59.0

 

Roger Smith _________13_8_2 32.0 __ 000 4.0__210 5.5__32110.0__432 9.0__32110.0__210 0.0____70.5

 

Jim Marusak _________13_6_1 34.0 __ 100 3.5__100 5.5__31010.0__531 7.0__220 8.0__100 1.5____69.5 

 

Rjay _______________ 13_5_2 34.0 __ 100 3.5__210 5.5__421 8.0__42112.0__200 5.0__000 2.0____70.0 

 

IsentropicLift ________ 13_5_1 33.0 __100*1.8__100*2.8__21011.0__532 6.0__100 3.0__000*1.0____58.6 (64.0)

 

Solak ______________ 12_5_2 39.0 __210 2.0__100*2.8__421*4.0__421*6.0__210*3.5 __ 000*1.0____58.3 (75.5)

 

CPC/NHC May outlook@ 12_5_2 39.0 _100*1.8__100*2.8__421*4.0__421*6.0__210*3.5__000*1.0____58.1 (77.0)

 

Ground Scouring _____ 12_4_2 37.0 __100*1.8__200 5.0__31010.0__421*6.0__211*4.0__000*1.0____64.8 (77.5)

 

metalicwx366 ________12_3_1 33.0 __ 100 3.5__110 6.0__32110.0__42112.0___111 6.0__000*1.0___71.5 (72.5)

 

Harry ______________11_5_2 43.0 __100* 1.8__100*2.8__311*4.5__421*6.0__210*3.5__000*1.0____62.6 (82.0)

 

Mike Ventrice ________11_4_1 40.0 __ 000 4.0 __100*2.8__310*5.0__421*6.0__210*3.5__000*1.0____62.3 (80.5)

 

CandymanColumbusGA 11_3_1 37.0 __ 200 2.5__210 5.5__31010.0__32013.0__110 5.0__000*1.0____74.0 (75.0)

 

Thewxmann _________11_2_0 31.0 __ 200 2.5__210 5.5__410 8.0__41012.0__000 3.0__000 2.0____64.0

 

Eyewall 2005 ________10_5_2 46.0 __100*1.8__100*2.8__321*5.0__421*6.0__210*3.5__000*1.0____66.1 (86.0)

 

Isotherm ___________ 10_4_2 44.0 __ 100 3.5__210 5.5__311 9.0__32114.0__210*3.5__100 1.5____81.0 (84.5)

 

Srain ______________ 10_4_2 44.0 __100*1.8__100*2.8__311*4.5__421*6.0__210*3.5__000*1.0____63.6 (83.0)

 

wxmeddler _________ 10_4_1 43.0 __ 100 3.5__210 5.5__31010.0__42112.0__100 2.9__000 2.0____78.9

 

Peabody Flood ______ 10_4_1 43.0 __100*1.8__100*2.8__310*5.0__421*6.0__210*3.5__000*1.0____63.1 (83.0)

 

U Thant ____________10_3_2 41.0 __ 100 3.5__210 5.5___311 9.0__531 7.0__100 3.0__000*1.0____70.0 (71.0)

___________ (9-2) __ 11_6_2 (44.0x.08=3.5)

 

superstorm93 _______ 10_3_1 40.0 __100*1.8__100*2.8__310*5.0__411*6.5__210*3.5__000*1.0____60.6 (81.0)

 

 

 

Consensus (median) _ 10_3_1 40.0 __ 100 3.5__110 6.0___311 9.0__42112.0__210 7.0__000 2.0____79.5

 

 

 

TropicalAnalystwx13 _ 10_2_1 36.0 __100*1.8__100*2.8__31010.0__32013.0__100*1.5__000*1.0____66.1 (73.0)

 

wildweatherman179 _ 10_2_0 34.0 __100*1.8__200 5.0__310*5.0__410*6.0__100*2.5__000*1.0____55.3 (71.5)

 

wxmx _____________ 9_4_1 45.0 __ 200 2.5__210 5.5___421 8.0__521 9.0___200 5.0__100 1.5____76.5

 

Isohume / CSU # ___ 9_3_2 43.0 __ 100#3.5__100#5.5__311#9.0__321#14.0_110#5.0__000#2.0___82.0

 

blazess556 _________ 9_3_1 42.0 __ 100 3.5__100*2.8__311 9.0__321*7.0__110*2.5__000*1.0____67.8 (81.0)

 

hurricaneman _______ 9_3_1 42.0 __ 100 3.5__210 5.5__421 8.0__32114.0__100 3.0__100 1.5____77.5

 

ncforecaster89 ______ 9_3_1 42.0 __ 100 3.5__210 5.5__311 9.0__42112.0__200 5.0__100 1.5____ 78.5

 

Stebo _____________ 9_3_1 42.0 __ 100 3.5__210 5.5__311 9.0__32114.0__210 7.0__110 1.0____82.0

 

Icebreaker5221 _____ 9_3_1 42.0 __ 100*1.8__210 5.5__32110.0__32114.0__110 5.0__100 1.5____79.8 (81.5)

 

bkviking ___________ 9_2_0 36.0 __ 100 3.5__100 5.5__31010.0__41012.0__000 0.0__000 2.0____ 69.0

 

Amped ____________ 9_1_1 33.0 __100*1.8__100*2.8__300*4.0__311*7.5__100*1.5__000*1.0____51.6 (70.0)

 

Pluffmud ___________ 8_4_1 46.0 __100*1.8__100*2.8__210*5.5__321*7.0__110*2.5__000*1.0____66.6 (87.0)

 

GaWx ______________8_3_1 43.0 __ 100 3.5__110 6.0__31010.0__42112.0__110 5.0__000 2.0____81.5

______________7-1__ 8_4_1 (46.0x.7=32.2)

 

andyhb ____________ 8_2_1 39.0 __ 100 3.5__200 5.0__310 9.0__32114.0__100 3.0__100 1.5____75.0

 

torchageddon _______ 8_1_0 32.0 __ 100 3.5__000 5.0__200 9.0__510 9.0__000 0.0__000 2.0____60.5

 

Ed Lizard ___________7_3_1 42.0 __1.5 0 3.1__100*2.8__210*5.5__211*8.0__110*2.5__000*1.0____64.9 (84.6)

 

biodhokie __________ 6_3_1 40.0 __ 000*2.0__100*2.8__210*5.5__211*8.0__110*2.5__000*1.0____61.8 (83.5)

 

PSUBlizzicane 2007___ 5_1_0 26.0 __ 100 3.5__200 5.0__21011.0__32013.0__100 3.0__000*1.0____62.5 (63.5)

_____________ 7-31 _ 7_3_0 (40.0x.4=16.0)

_____________ 9-2 __ 7_5_0 (45.0x.07=3.2)

 

Notes:

 # Isohume monthly provisionals are scored full value since he was just relaying the FSU forecast. This allows for maximum available scoring points for true comparison with actual forecasts.

 

@ added Oct 18th, after reading in hurricane thread that CPC/NHC May outlook was 12/5/2, which happens to be same as Solak, I have added them in. Solak predicted June and missed the other months so there is a slight difference there.  I have gone back to the other table and added them in there as well. I suppose that we should count the NHC forecast on full monthlies for the comparison, as with the CSU-Isohume entry.

 

Forecaster name in italics -- all monthlies were issued at start of season (one or two may have been re-issued under the rules, but basically these contestants gave their monthlies "blindfolded" in terms of how the season was actually unfolding.

 

 

Contest leaderboard (of those who played all or most of the monthly segment) would then be as follows:

 

 

Stebo ____________82.0

Isohume (CSU) ____82.0

GaWx ___________ 81.5

Isotherm _________81.0

Icebreaker5221 ___ 79.8

 

Consensus _______ 79.5

 

wxmeddler _______78.9

ncforecaster89 ____78.5

hurricaneman _____77.5

wxmx ___________76.5

andyhb __________75.0

candymanCSGA____74.0

metalicwx366 _____71.5

Roger Smith ______70.5

U Thant _________ 70.0

Rjay ____________ 70.0

Jim Marusak ______69.5

bkviking _________ 69.0

 

Normal __________63.0

 

PSUBlizzicane07 ___62.5

Torchageddon ____ 60.5

 

 

Some of the provisional scores were into the 60s, so this is not exactly the final leaderboard. Will publish that at the end of the contest. Same goes for the "what if they had played" leaderboard where the two highest scores were 87 for Pluffmud and 86 for Eyewall2005.

 

____________________________________________________

 

 

At this point, the highest monthly totals for players who tried all or all but one including all who played JUN to NOV for comparison ...

(using one provisional at 50%)*

(using all provisionals as implied) # 

 

Isotherm _____ 40.5*

Stebo ________ 40

Isohume ______39#

Icebreaker5221_39.5*

metallicwx366 _ 38.5*

Roger Smith ___38.5

Gawx _________38.5

Candyman ____ 37*

PSUblizzicane__ 36.5*

ncforecaster ___ 36.5

andyhb _______ 36.0

RJay _________ 36

 

Normal _______ 36

 

wxmeddler ____ 35.9

hurricaneman __ 35.5

Jim Marusak ___ 35.5

Thewxmann ___ 33

bkviking ______ 33

Uncle W ______ 32

wxmx ________ 31.5

U Thant ______ 29*

torchageddon__ 28.5

 

A few of those were all issued in June and either left unedited or given one late edit, so well done if you placed up in the table from that distance out. Isentropic Lift would have scored 31 playing June and July plus NOV-DEC at full scores, Tropicalwx analyst13 would have scored 37 and blazess556 39, playing June, July, October and NOV-DEC at full scores.

 

The highest monthly from provisionals at full score would have been 43.5 for biodhokie whose score is entirely theoretical and 42.6 for EdLizard who did enter June and is otherwise a fully theoretical monthly. Otherwise most of the doubled theoretical values would have placed between 37 and 40 (but actually scored 18.5 to 20).

 

All of these scores will change again if there is more activity -- this is what you would score if nothing else happens in NOV / DEC.

 

 

 

[Stebo]

Nov-Dec

 

1/1/0

[Roger Smith]

I believe I can control the weather just by predicting it ... not really but I need a 3/2/0, 1887 style. And there go my 2 points out the window.

[Isotherm]

Nov-Dec:

 

1/0/0

[andyhb]

Nov-Dec: 1/0/0

[Roger Smith]

OCT count back up to 3/2/1 as speculated ... so I changed the last provisional scoring table yet again to show scores as of today.

 

This includes a full report on the monthly segment.

 

The full contest is now quite close between Stebo, Isotherm and Gawx with several others in the mix, especially if we get any further activity. This means your 2-pt NOV-DEC forecast could be the margin at the end of the contest.

 

I had indicated 3/2/0 above but will now be going 2/1/0 (just playing what I need in the seasonal to move into the lead, would need the extra 2 points to maintain it). I really think it could happen too, in fact there are some signs of a 4/2/1 or even 4/3/1 finish for OCT which is going to decimate the monthly scores considerably (and force me to edit the darned table again).

 

If you're doing the math in your head, the 10 scores for OCT if the yellow x becomes a storm (and I saw where it did on one model at least, but close to the month boundary) would drop to either 9 or 8, the 8 scores would drop to either 5 or 4 but might then rise a bit due to the mercy rule, and most of the others would drop quite low or to zero.  As most of the leaders have the same OCT scores, it would appear that only your host would benefit from any of that.

[icebreaker5221]

Meh, wanted to go 1/1/0 but models in disagreement w.r.t. MJO and week-2 Kelvin wave could hold together or just as easily fall apart.  Plus ENSO and overall seasonally base state argue against it.  Guess I'll conform:

 

1/0/0