SMCJB 

NG Margin increase effective COB September 10th, 2021 Maintenance margin's increasing as follows (Note: Non-member initial margin rates will be 110% of these) Tier 1 / Oct21 from $3500 to $4100 +$600 / +17% Tier 2 / Nov21 from $3400 to $4050 +$650 / +19% Tier 3-5 / Dec21-Feb22 from $3300 to $3950 +$650 / +20% Tier 6 / Mar22 from $3100 to $3550 +$450 / +15% Tier 7 / Apr22 from $1450 to $1800 +$350 / +24% (!!!) Tier 8 / May22 from $1300 to $1500 +$200 / +15% Tier 9-13 / Jun22-Oct22 from $1300 to $1400 +$100 / +8% Tier 14-18 / Nov22-Mar23 from $1100 to $1300 +$200 / +18% Tier 19-30 / Apr23-Mar24 from $750 to $800 +$50 / +7% all others unchanged. https://www.cmegroup.com/notices/clearing/2021/09/Chadv21-317.html And it hasn't even started getting cold yet!

what are some of the incentives for nymex to increase margin req? could it be due to one sided positions getting heavy?

bob666s 

what are some of the incentives for nymex to increase margin req? could it be due to one sided positions getting heavy?

Volatility

SMCJB 

Volatility

noted with thanks

Tried to run a valuation estimate with COT positions and storage, using neural networks. The output came out still close to the earlier estimates, around high 3's for the next couple of months. It feels too low, even if the coming HDD turns out about where expectations lie.

details of the method are posted here: https://bastion.substack.com/p/ng-quantamental-st-fv-fcasts

Does anyone have any idea why the quantitative approach is likely off, or wrong by so much? I'll take any suggestions regarding how to adjust this for improved accuracy. thanks.

NatGas Margins going up AGAIN!

NG Margin increase effective COB September 14th, 2021

Maintenance margin's increasing as follows (Note: Non-member initial margin rates will be 110% of these)

Tier 1 / Oct21 from $4100 to $4300 +$200
Tier 2 / Nov21 from $4050 to $4300 +$250
Tier 3-4 / Dec21 & Jan22 from $3950 to $4300 +$350
Tier 5 / Feb22 from $3850 to $4275 +$425
Tier 6 / Mar22 from $3550 to $3700 +$150
Tier 7 / Apr22 from $1800 to $1950 +$150
Tier 8 / May22 from $1500 to $1600 +$100
Tier 9-13 / Jun22-Oct22 from $1400 to $1500 +$100
Tier 14-18 / Nov22-Mar23 from $1300 to $1450 +$150
all others unchanged.

https://www.cmegroup.com/notices/clearing/2021/09/Chadv21-324.html

bob666s 

Tried to run a valuation estimate with COT positions and storage, using neural networks. The output came out still close to the earlier estimates, around high 3's for the next couple of months. It feels too low, even if the coming HDD turns out about where expectations lie. details of the method are posted here: https://bastion.substack.com/p/ng-quantamental-st-fv-fcasts Does anyone have any idea why the quantitative approach is likely off, or wrong by so much? I'll take any suggestions regarding how to adjust this for improved accuracy. thanks.

Ever think about making the data you used downloadable on your substack page?

Comments

1. Nobody should take Neural Network advice from me, as I have a very high failure rate on using them. I always find that changing the initial seed completely changes my results which obviously means my model isn't stable or predictive.
2. Not sure the value of OI, especially if you are only looking at NYMEX NG and not including ICE HNG. Also I think you will find OI is seasonal.
3. With the EIA data are you looking at outright levels or change vs say 5 year average. Outright levels are highly seasonal. Change vs 5 year gives a better quick summary of fundamentals. Obviously if you have month (or even week) as a variable then a good ML model should identify this relationship but it could be a better feature. At a minimum include actual data and 5 year average data and let the model imply the change the difference.
4. Price vs Oil. This is becoming a lot more prevalent now but several years ago was irrelevant. One of the reasons US gas inventories are low is because of the amount of LNG we are exporting. Doesn't matter that US prices are at multi year highs/$5-6 when international prices are at $20. Exports will happen as fast as they can load the vessels. This definitely WASN'T the case historically.
5. Finally your charts all show the last 2-4 years. The US has been in over supply situation during this period. If you look back say 10-15 years, when the US was in a supply deficiet you will see extremely different price relationships. Given the fundamental difference this year, maybe the last 2-4 years aren't the best to train data on.

Like the work. Keep posting.

Sorry this one is a few days late. Was out of town end of last week.

Yes NatGas Margins going up AGAIN!

NG Margin increase effective COB September 17th, 2021

Maintenance margin's increasing as follows (Note: Non-member initial margin rates will be 110% of these)

Tier 1-4 / Oct21-Jan22 from $4300 to $4800 +$500
Tier 5 / Feb22 from $4275 to $4800 +$525
Tier 6 / Mar22 from $3700 to $4200 +$500
Tier 7 / Apr22 from $1950 to $2150 +$200
Tier 8 / May22 from $1600 to $1750 +$150
Tier 9-13 / Jun22-Oct22 from $1500 to $1650 +$150
Tier 14-18 / Nov22-Mar23 from $1450 to $1500 +$50
other months going up between $0 and $75

https://www.cmegroup.com/notices/clearing/2021/09/Chadv21-238.html

Interesting read. The bearish NG case....

https://seekingalpha.com/article/4455892-natural-gas-could-crash-without-a-cold-winter

SMCJB 

Interesting read. The bearish NG case.... https://seekingalpha.com/article/4455892-natural-gas-could-crash-without-a-cold-winter

Thanks.

I love the Cordier photo.

KOLD is trading at 11.15 this morning. Last Friday when he wrote article it was 9.52.

What ways will others here use to get short NG.

SMCJB 

Interesting read. The bearish NG case.... https://seekingalpha.com/article/4455892-natural-gas-could-crash-without-a-cold-winter

I do have to wonder about the data he used for weather. It looks like he didn't use population weighted weather data. He said this,



But pwCDD for July were below the long term average and below prior years. The heat was in the less populated western part of the US.

August was hotter than prior years and the long term avg.