Data is currently at
https://data.giss.nasa.gov/gistemp/tabledata_v4/GLB.Ts+dSST.csv
or
https://data.giss.nasa.gov/gistemp/tabledata_v4/GLB.Ts+dSST.txt
(or such updated location for this Gistemp v4 LOTI data)
January 2024 might show as 124 in hundredths of a degree C, this is +1.24C above the 1951-1980 base period. If it shows as 1.22 then it is in degrees i.e. 1.22C. Same logic/interpretation as this will be applied.
If the version or base period changes then I will consult with traders over what is best way for any such change to have least effect on betting positions or consider N/A if it is unclear what the sensible least effect resolution should be.
Numbers expected to be displayed to hundredth of a degree. The extra digit used here is to ensure understanding that +1.20C resolves to an exceed 1.195C option.
Resolves per first update seen by me or posted as long, as there is no reason to think data shown is in error. If there is reason to think there may be an error then resolution will be delayed at least 24 hours. Minor later update should not cause a need to re-resolve.
1,000@gonnarekt Thanks for sharing your prediction, as so few do.
Point wise I’ve be getting between 1.32 to 1.33, but I have a much more conservative prob. dist. now for my models than most others I’ve seen posted.
@gonnarekt this is my prediction as well when comparing the era5 month to month. However, my gut feeling says that it is more likely to see deviations to lower temperstures
Month has ended ....
Still a couple days left of ERA5 data but the GEFS ARIMAX adjusted errors have been fairly low so it doesn't look like my point prediction will change much more than maybe 0.01 degrees:
Right now March looks its going to be ~ 1.341 +- 0.055 C
(this rough 95% CI estimate is just from the past few predictions, but doesn't include a prediction for a month like March. It's a bit different from my past few predictions which relied on the RMSE from the validation set of a linear model's RMSE to correct biases from ERA5->GISTEMP -- which could also be a by month model if it validates, or an entire year model if it also validates. If I try to narrow the CI using an alternate method by narrowing the sample set from March using later cutoff dates for the data (instead of 1940 which earlier decades have larger uncertainty) the RMSE only gets down to about 0.04 -- this is still larger than the ~0.027 std. dev. above).
I'll have a final point prediction from this model when ERA5 data is final in 2-3 days.
Edit:
For reference 1.34 falls right in the middle of the majority polymarket bin which has it at 78%
Manifold market is a bit unluckier and it's possible it might be a near coin toss in the next ~ 10 days as to which run ends up being chosen if GISTEMP runs hover between 1.34 and 1.35.
GEFS and ECM now trending in opposite directions for end of month...
last ECM run unadjusted:
ECM_2025032012:
valid_date 2t
0 2025-03-21 14.597478
1 2025-03-22 14.533439
2 2025-03-23 14.457115
3 2025-03-24 14.422914
4 2025-03-25 14.437479
5 2025-03-26 14.417067
6 2025-03-27 14.338064
7 2025-03-28 14.271404
8 2025-03-29 14.226858
9 2025-03-30 14.241461
10 2025-03-31 14.257847
11 2025-04-01 14.266293
12 2025-04-02 14.326877
13 2025-04-03 14.407927
GEFS-BC runs from today unadjusted:
-- Processing ./gefs.20250320_00
-- Processing 105 files...
Global 2m temp average (C) for 03-20-2025: 13.820 (day over day delta: 0.000)
Global 2m temp average (C) for 03-21-2025: 13.749 (day over day delta: -0.070)
Global 2m temp average (C) for 03-22-2025: 13.726 (day over day delta: -0.023)
Global 2m temp average (C) for 03-23-2025: 13.745 (day over day delta: 0.019)
Global 2m temp average (C) for 03-24-2025: 13.841 (day over day delta: 0.096)
Global 2m temp average (C) for 03-25-2025: 13.904 (day over day delta: 0.062)
Global 2m temp average (C) for 03-26-2025: 13.930 (day over day delta: 0.026)
Global 2m temp average (C) for 03-27-2025: 13.936 (day over day delta: 0.006)
Global 2m temp average (C) for 03-28-2025: 14.000 (day over day delta: 0.065)
Global 2m temp average (C) for 03-29-2025: 14.122 (day over day delta: 0.122)
Global 2m temp average (C) for 03-30-2025: 14.191 (day over day delta: 0.069)
Global 2m temp average (C) for 03-31-2025: 14.286 (day over day delta: 0.095)
Global 2m temp average (C) for 04-01-2025: 14.391 (day over day delta: 0.105)
Global 2m temp average (C) for 04-02-2025: 14.457 (day over day delta: 0.067)
Global 2m temp average (C) for 04-03-2025: 14.472 (day over day delta: 0.015)
Global 2m temp average (C) for 04-04-2025: 14.489 (day over day delta: 0.017)
-- Processing ./gefs.20250320_06
-- Processing 105 files...
Global 2m temp average (C) for 03-21-2025: 13.762 (day over day delta: -0.298)
Global 2m temp average (C) for 03-22-2025: 13.728 (day over day delta: -0.035)
Global 2m temp average (C) for 03-23-2025: 13.736 (day over day delta: 0.008)
Global 2m temp average (C) for 03-24-2025: 13.840 (day over day delta: 0.104)
Global 2m temp average (C) for 03-25-2025: 13.927 (day over day delta: 0.087)
Global 2m temp average (C) for 03-26-2025: 13.954 (day over day delta: 0.027)
Global 2m temp average (C) for 03-27-2025: 13.972 (day over day delta: 0.018)
Global 2m temp average (C) for 03-28-2025: 14.070 (day over day delta: 0.098)
Global 2m temp average (C) for 03-29-2025: 14.204 (day over day delta: 0.134)
Global 2m temp average (C) for 03-30-2025: 14.193 (day over day delta: -0.012)
Global 2m temp average (C) for 03-31-2025: 14.356 (day over day delta: 0.164)
Global 2m temp average (C) for 04-01-2025: 14.440 (day over day delta: 0.084)
Global 2m temp average (C) for 04-02-2025: 14.506 (day over day delta: 0.066)
Global 2m temp average (C) for 04-03-2025: 14.545 (day over day delta: 0.038)
Global 2m temp average (C) for 04-04-2025: 14.543 (day over day delta: -0.002)
-- Processing ./gefs.20250320_12
-- Processing 105 files...
Global 2m temp average (C) for 03-21-2025: 13.753 (day over day delta: 0.011)
Global 2m temp average (C) for 03-22-2025: 13.711 (day over day delta: -0.041)
Global 2m temp average (C) for 03-23-2025: 13.707 (day over day delta: -0.004)
Global 2m temp average (C) for 03-24-2025: 13.806 (day over day delta: 0.099)
Global 2m temp average (C) for 03-25-2025: 13.890 (day over day delta: 0.084)
Global 2m temp average (C) for 03-26-2025: 13.912 (day over day delta: 0.022)
Global 2m temp average (C) for 03-27-2025: 13.921 (day over day delta: 0.009)
Global 2m temp average (C) for 03-28-2025: 13.992 (day over day delta: 0.071)
Global 2m temp average (C) for 03-29-2025: 14.127 (day over day delta: 0.135)
Global 2m temp average (C) for 03-30-2025: 14.339 (day over day delta: 0.212)
Global 2m temp average (C) for 03-31-2025: 14.296 (day over day delta: -0.043)
Global 2m temp average (C) for 04-01-2025: 14.373 (day over day delta: 0.077)
Global 2m temp average (C) for 04-02-2025: 14.451 (day over day delta: 0.078)
Global 2m temp average (C) for 04-03-2025: 14.470 (day over day delta: 0.019)
Global 2m temp average (C) for 04-04-2025: 14.452 (day over day delta: -0.018)
-- Processing ./gefs.20250320_18
-- Processing 105 files...
Global 2m temp average (C) for 03-21-2025: 13.767 (day over day delta: 0.788)
Global 2m temp average (C) for 03-22-2025: 13.713 (day over day delta: -0.055)
Global 2m temp average (C) for 03-23-2025: 13.693 (day over day delta: -0.020)
Global 2m temp average (C) for 03-24-2025: 13.791 (day over day delta: 0.099)
Global 2m temp average (C) for 03-25-2025: 13.880 (day over day delta: 0.089)
Global 2m temp average (C) for 03-26-2025: 13.893 (day over day delta: 0.012)
Global 2m temp average (C) for 03-27-2025: 13.880 (day over day delta: -0.013)
Global 2m temp average (C) for 03-28-2025: 13.927 (day over day delta: 0.048)
Global 2m temp average (C) for 03-29-2025: 14.029 (day over day delta: 0.101)
Global 2m temp average (C) for 03-30-2025: 14.278 (day over day delta: 0.249)
Global 2m temp average (C) for 03-31-2025: 14.230 (day over day delta: -0.048)
Global 2m temp average (C) for 04-01-2025: 14.318 (day over day delta: 0.088)
Global 2m temp average (C) for 04-02-2025: 14.397 (day over day delta: 0.078)
Global 2m temp average (C) for 04-03-2025: 14.428 (day over day delta: 0.031)
Global 2m temp average (C) for 04-04-2025: 14.435 (day over day delta: 0.008)Polymarket is very confident in the 1.32-1.36 bin...
I don't really have a clue what the final temperature will be so precisely, but the splitting the adjusted forecasts puts a point forecast at ~ 1.33 in my model which at least agrees with the general symmetry and most probable bin of Polymarket. I still have a much lower confidence though than Polymarket's current odds do.
The remainder of the month should be interesting: there has been a small dip so far in recent ERA5 data, and GEFS has slightly adjusted in last couple days but GEFS still indicates a trend upwards, opposite of ECM that wants to trend downwards.
I have adjusted methodology for the GEFS forecasts (using an adaptive averaging approach rather than averaging all past forecasts to generate the predicted day average for GEFS (weighting last couple days as the average (slight exp. decay more towards recent ones, but discarding the older forecasts) when all 8 forecasts vary significantly from the larger pool's forecast std. dev).
Also switched to ARIMAX rather than using OU after doing much more testing with the data; not much change indicated (ARIMAX is more conservative over OU at least for the remainder of the month: wanting to bring down GEFS temps by ~0.06 vs ~0.10 adjustment over last 8 days)
Despite all that, GEFS and and ECM are still far apart; even if you assume the average of the remaining ERA5 temps (Mar 24-31) will end up somewhere in between, the difference is then 0.59 / 2 ~= 0.3 C. (This might be a bit exaggerated if the ARIMAX adjustments both perform poorly in trying to adjust to the correct ERA5 temps).
The ECM dip looks too extreme but the GEFS rise only looks slightly less plausible after the recent dip)
(the GEFS (red line) and the ECM in the charts both have forecasts adjusted by ARIMAX)
EdiT: The latest ARIMAX adjusted forecasts to ERA5 using ECM, GEFS look much more smooth and plausible (continuous) intuitively to my human eyes with the latest ERA5 temps.
ECM still showing a dip on the 28th of ~ 0.1 to 0.15 degrees from the prior day, and the remainder of the days for the month are on the cool side relative to today. Still disagreement between the two about the trend for the remainder of the month. (GEFS_ARIMAX only shows the most minor of a dip around then)
(RED is ERA5 historical up until Mar. 24 then GEFS_ARIMAX followed by statistical long run extrapolation; BLUE are last 5 ECM_ARIMAX runs):
Latest 4/5 (adjusted) runs from ECM (last 2 days) show the rest of the month on average nearly plateauing ... consequently, splitting the difference requires downwardly adjusting the GEFS offset quite a bit further than I thought might happen two days ago...
@LeonardoParaiso As I said previously where before (last week) ECM and GEFS were in somewhat decent agreement on the trend of rising temps for the end of the month, now there is a bifurcation: recent ECM runs says there will be (roughly) a plataeu, while GEFS continues having a trend of rising temps (albeit from the current much lower temperature than I anticipated even a few days ago).
The sharp rise during the middle of the month that GEFS predicted a week ago hasn't happened though which doesn't bode well for GEFS....
I think the Manifold's current odds are more plausible though than Polymarket's high certainty, but I won't be surprised if there is more flip-flopping, so I still favor an even flatter distribution than what even Manifold has given the uncertainty....
@SiebeNijland Yeah, the trend as of yesterday for recent ECM runs is lower ... (GEFS adjusted reference - ERA5_ARIMA model runs)
model_init avg_diff
0 2025031612 0.171753
1 2025031700 0.183720
2 2025031712 0.252170
3 2025031800 0.277377
4 2025031812 0.334323Done putting in some limit orders for the night, @gonnarekt or anyone else if you wanna bet more
Vibe betting always works! (lets see how well it works this month):
Expecting 2025 to supercede 2024's record of 1.39 for March based on mostly statistical extrapolation for missing forecast data (last 12 days).
Expecting current forecast up until middle of month to track or be slightly above 2024 on average. The second half of March 2024 had an abnormal dip, so on average following the trend I expect something more like this for March (ignore rest of year past March), using the OU as a guess for the offset:
You can imagine the line being within (very roughly) +- 0.6 C for the last ~2 weeks and roughly +- 0.3 for the next 2 weeks.
Even with a slightly less likely 0 offset for the error for the rest of the month (less likely since I expect it will take some time for the error to drift back to 0, based on the recent trend, giving on average a positive offset for the rest of the month), on average it looks like 2025 March will beat 2024 just by eyeballing it, so at the moment I think the last two bins should be the most likely:
Past March 24 I fill in statistically (you can ignore the rest of the year past March):
The end of the 3rd week for the GEFS forecast shows the temperature rise levelling off, so I look at some statistics for the last week of March in ERA5 given the GEFS forecast now covers until March 24.
Now the statistics for last week of March (the last datapoint in the two charts below is the 0 offset prediction for 2025):
This shows almost always (81/85 = ~95% of the time) the average temperature in the last 7 days of March shows a rise compared to the March 24 temperature (an average of ~0.147 C higher). (The 2025 guess has it about 0.117 higher than March 24 so not too far from the mean and actually a bit under)
With the absolute temp average for last week the last week of March 2025 a bit higher than 2024 (based on extending the GEFS forecast statistically).
As I still need to offset the error for part of the GEFS forecast (although it has been decreasing) I do some basic calculations for a scenario where we assume the offset of 0 is correct for the GEFS portion but the statistical forecast rise is wrong for the last 7 days, and for instance the average temperature from March 25-31 is the same as March 24 or less: this is the 5% chance mentioned above that corresponds to any offset ~ <-0.037 (= -(.117) * (7/22)) in the chart below:
The OU meta-prediction is still predicting a slightly positive offset for the remainder of the month (0-0.09) looks plausible), but at the moment it seems likely (with low confidence) the anomaly for March will still be broken (currently at 1.39 for March 2024) given the GEFS forecast and the above...
Its still too early to say what offset is correct, but the odds are tilted towards still being positive ~0.05:
March is the hardest month to predict though so we shall see...
Polymarket for reference (has completely different bins) and a completely different forecast at the moment (currently predicting something below or near last year's anomaly temp):
@parhizj gopfan saw all of this and went against everything you just said 😂 that dude bats a thousand. gopfan more like goatfan
@LeonardoParaiso That’s the whole point of markets. At least the individual bin probabilities do not seem wildly overconfident to me at the moment; Poly is at 61%.
Did some more work today including deterministic ECM data to try to come up with an even yet more elaborate prediction system combining ECM and GEFS after all the final adjustments are all done by splitting the difference of the average for the remaining forecasted temps.
In doing so I did understand the source of some of the discrepancy at least between what I have been predicting using GEFS for March and what seems clear others are predicting using I believe ECM data:
Comparing the last 5 runs using ECM (wherein I try fit to the ERA5 data and subsequently further adjust the forecast offsets using ARIMA (as using OU did not work out with ECM)) with the GEFS reveals ECM predicts lower temps for the middle of the month finally matching GEFS towards the end of the month.
Using a (mixed) best guess for the offset for GEFS (red line) it looks like this:
I thereafter calculate the average difference for GEFS (with offsets) - each ECM model run (last 5 model runs) across common dates (after all offsets already):
model_init avg_diff
0 2025031012 0.153844
1 2025031100 0.203933
2 2025031112 0.152150
3 2025031200 0.110761
4 2025031212 0.175303
Mean of average differences: 0.1592I then split the difference in offsets to calculate a new split offset (this I hope has the effect of weighting very roughly equal the final adjusted ECM and GEFS (mapped) forecasts)
This corresponds to an offset that is now slightly negative (which more closely matches the last couple errors for gefs predictions - ERA5), which looks like this now:
This corresponds to the grey line as the split difference offset below, which shows roughly equal odds for the highest 3 bins (with the highest bin with more probability mass).
I don't particularly like the look of the last couple few days though from GEFS:
This didn't have any serious consequences for my recent betting in hindsight, but it does show that it is especially sensitive to the next week or so rather than the end of the month (will it continue near-plateuing like ECM, jump up as high as GEFS suggests, or somewhere in between)
For any interest bettors, a simple linear squares regression to debias from ECM->ERA5 leaves quite a bit to be desired (this is only using the 00Z init forecast and 0,6,12,18h valid time samples)
Mean Squared Error (MSE): 0.06
Root Mean Squared Error (RMSE): 0.25
Mean Absolute Error (MAE): 0.21
OU did not validate well but the time series dependencies of the runs on the residuals also is obvious even without running autocorrelation and partial autocorrelation, thus resorting to ARIMA which seems ok but I haven't put any instrumentation to monitor its performance.