Moon Phases and Sales

[LaRetta]

Not don't you all laugh ... but I would like to view our sales trends back through the past five years with the phases of the moon. I won't go into why - some of you understand while others think I'm nuts. But to most, it comes as no surprise - my being nuts, I mean.

Is there a database (with dates and percentages) or is there formulae? I suppose it would need to be based on __% of Full Moon. How would I approach such a task?

UPDATE: I did find this site which might give me the data I need. I believe I would need to base it on the Customer's location (we sell around the world) thus results would be based upon their location influence (more than ours).

LaRetta

Edited June 17, 2006 by Guest
Added update

[SteveB]

As an ex-market researcher, I've graphed loads of silly stuff. I assume you'll use Excel, or some other stat pak?

Try these links: http://scienceworld.wolfram.com/astronomy/LunarCycles.html or http://www.atimeforsuccess.com/articles/lunarcycles.html

You're only a 'little nuts' (is that like being a little preggie?)

Steve

[LaRetta]

I assume you'll use Excel, or some other stat pak

No, I hadn't even considered using any other program than FileMaker. I pictured using tables and even custom functions if necessary; joined on multi-predicate timestamp relationships and summing.

If I'm going to do it, I want to consider delta-T as well; I want it accurate and rounded to full percents. Thank you for the link!!

Edited June 17, 2006 by Guest

[LaRetta]

:jester: Alright. I confess. I am totally insane. But I'm not stupid. If I see trends and spikes, you can bet I'll suggest doubling our sales force during those periods!! :laugh2:

[comment]

Boy, you sure know how to pick'em. OK, this will return the lunar phase of Datefield as a fraction. It cycles from 0 to 1, with new moon being 0 and full moon returning 0.5. Due to natural variations of the lunar cycle, the calc can be off by as much as half a day.

Let ( [

epoch = 730124.597661 ;

synod = 29.5305888531

] ;

Mod ( Datefield - epoch ; synod ) / synod

)

Make sure you also read this.

Edited June 18, 2006 by Guest
Fixed the epoch constant

[Genx]

Hmmm, a real life use for regression analysis... Easy to do once you've got the data in excel but how on earth were you planning on doing it in FM?

I enjoyed the read btw comment .

[LaRetta]

Wow.

This is incredible! There are two reasons I posed this question:

1) I will do anything (as long as it is legal and moral) to increase sales. From personal experience, I believe the entire natural realm is influenced by moon, sun, season, electromagnetic and gravitational pulls. I am intrigued by these correlations. Dogs may not bite more on a full moon but did anyone ever consider that they may bite more on waning gibbous at 61% during August at the equator? The fact that results are unpredictable simply means we humans haven't devised the proper tests yet (in my opinion). I'm not implying that *I* could devise proper tests ... only that I refuse to discount possibilities.

2) I simply want to understand the mathematics and principles required to produce this type of analysis by doing it. What a world THIS has opened up for me. Hubble and Plate Tectonics move over!

I can find no function in Excel which even comes close but I think I've figured synod - "On average, the synodic period is 29 days, 12 hours, 44 minutes and 3 seconds long." But epoch is throwing me. The Unix epoch is 00:00:00 UTC on January 1, 1970 so you don't mean that. Julian epoch (current standard) is J2000.0 (noon on 1/1/2000). Your epoch produces approx. noon on 1/6/2000, right? Could you tell me more about your formula? I realize gravitational pull and geoid come into play ... is that what you are accounting for - those TT and TDT shifts? Or leap seconds or?

And thank you so much, Michael!!

LaRetta

Edited June 17, 2006 by Guest
Changed a few words

[LaRetta]

...no more significant tonight than on any other night of the year, according to scientific reviews of the theory that the full moon alters the way humans and wildlife behave.

It seems that we haven't told the oceans full of marine life which follow reproduction cycles dependent entirely upon the moon's cycle, ie, tidal action. Whether the behavior is inherited or learned, we are all creatures of nature. And maybe my horses run the fields during evening of full moon only because they can see better - but they nonetheless run!! Even wolves howl at full moons. Why? So they can see what they are howling at? Thanks for indugling my peculiarity with this request. :wink2:

Edited June 17, 2006 by Guest
Corrected quote

[Genx]

I wasn't implying for you to try and get the data in excel, i was implying that after you have extracted it you might consider using the regression data analysis within excel on the data that you calculate in FM to try and see how appropriate your model is by at least looking at the r square value - implying of course that was some sort of linear relationship between your two variables... and even if there wasn't you might be able to isolate some other type of non-linear function by looking at the regression plot.

[comment]

Your epoch produces approx. noon on 1/6/2000, right?

It was supposed to be Jan 6,2000 14:20:37.9104, but I've made a mistake. Fixed now.

As for your question, you've got to start the cycle somewhere. The moon does not run on Unix, so if you want the calc to roll back to 0 at new moon, you need to start your epoch at a new moon. From what I've found, that seems to be the agreed starting point for computing the AVERAGE synodic lunar period - even though the actual new moon on that very day occured some 4 hours later.

[Stuart Taylor]

just a little point ... would the 4 hours be accounted for if it was london and not the US/NY that the starting point of the full moon was based on. seems like the right time difference.

[comment]

No. The formula above is actually based on UTC (what used to be Greenwich time). Properly, it needs adjustment to local time, but since it is only an estimate anyway, it doesn't matter much.

The 4 hours are accounted for by the natural variations of the lunar cycle. The formula computes the position of the given date in an AVERAGE lunar cycle. Any PARTICULAR cycle may vary from the average by up to 14 hours. It just so happened that the cycle was off by 4 hours on Jan 6, 2000.