Money:Tech — September Retail, and Making Money from the Weather

Lots of unhappiness about weak September U.S. retail results. Some of that can be blamed on a fading consumer, but there is also ample reason to think it has something to do with the weather.

Check the following figure (click for a larger version) from Weather Trends for a somewhat complex graph that shows the recent weather/retail relationship, tracking comp sales data against precipitation and average temps in major metros. The upshot? September was much warmer and drier than seasonal norms, likely driving poor cold weather clothes sales.

 

 

There will be oodles more meaty stuff on making money from weather data at my Money:Tech conference in New York this coming February.

Related posts:

  1. Making Money from the Weather
  2. Making Money from Weather Data
  3. Making Money from the Weather, Part III
  4. Making Money from the Weather
  5. Weather Channel Buys (Weather) Bonk

Comments

  1. Mike Mothner says:

    Who would have ever thought comp sales could be related to weather? Once there are specialists in this area of financial combined with meteorological analysis we’ll know that we have truly entered a new era.

  2. Umm, yeah, but:
    1) You only know the weather after it’s happened
    and
    2) There’s lots and lots of opportunities for seeing patterns which aren’t really there.
    People have been making weather-related bets ever since farming was invented. I wouldn’t say that couldn’t be refined with better data. Obviously it can. But it seems more a very narrow specialized application in practice.

  3. Fair and sober thoughts, Seth. But a) we’re getting retail data today, and we’ve known about September weather for almost two weeks, b) the utility of weather data in financial markets is eminently testable, and c) as a single tradable factor it’s narrow, but as a component of a multifactor bottoms-up model, it’s much broader utility.

  4. Adam says:

    I find the weather effect hard to interpret from the graph shown. I would like to see the correlation coefficients or ANOVA table before drawing conclusions about weather’s impact on comps.

  5. Hey says:

    For a prediction model, you need to calculate variance from target weather of items for sale. September sales are for cold & foul weather items – hotter & drier is very bad. Major population centers had 100 degree effective temps on October 8th – and these were the key centers of cold & foul weather!
    For March you’re selling items that are for warmer weather with a mix of dry and wet weather goods. So warmer weather should drive sales, while wetter won’t harm sales since product mix will be mostly spring, rather than summer, apparel and you’ll pick up additional sales of inclement weather gear.
    December was hampered by warm and dry weather, but holiday kept it so it was average comps instead of weak comps. Cold January helped drive sales, could have been even better with more snow.
    Great tool, but you do really need to refactor it to make precitions.