First, we note that Gas Weighted Degree Days (GWDDs) for the month of September over time have very gradually declined, thanks in part to a large number of low-demand years last decade. Between 1980 and 2000 we only saw one year where September GWDDs came in below 220. Between 2004 and 2009 we then saw 3 of 6 years have September GWDDs below 220. Then, after that, we have remained clustered right around 255 GWDDs on average for the month. 

A standard deviation of only 21 GWDDs for the month shows how rather inconsequential it can often be for the natural gas market. Even a 2-SD weather deviation for the month only adds or subtracts around 42 GWDDs, which certainly is not much when you realize we are around 400 GWDDs below normal levels so far through 2017. Though a significant weather event could have a decent impact on end of storage, our modeling shows even a 2-SD September weather deviation would at most impact storage by around 60-70 bcf or so, certainly nothing crazy. 

As we move through the month, we will gradually see Utility Gas Weighted Heating Degree Days (UGWHDDs) overtake Population Weighted Cooling Degree Days (PWCDDs). We typically see about twice as many PWCDDs as UGWHDDs for the month as a whole, and nationally September 23rd is the date where historically UGWDDs first match up with PWCDDs before overtaking them as we head later into the fall season. 

The result is that warmer forecasts, especially across the South and Southeast, would technically be bullish through much of the month, but colder forecasts late in the month across the North and Northeast should become a bit more supportive as well. It generally becomes hard to get a perfect mix of either bullish or bearish weather through September, which likely explains why in the past weather rarely drives natural gas price action through the month.