Off the top of my head I'll go for March '09 the week ending the 20th.

edit: whoops brain fart on march <---> may, Corrected above.

April 2009 week ending 17th.

March 2009, week ending 27th

After some calculations, I ended up with the same answer as BengaBenga: week ending 27th of March 2009.

I'm looking forward to the "never" predictions....those always give me a giggle...you know..in a manly sort of way =P

Second half of 2009. Date I have no idea.

July 10th 2009.

Just finished running some quick numbers from the spreadsheet and it came up with this as well. I'll stick with my initial gut still but its interesting how close these predictions have been to each other thus far.

Two interesting things emerge from my analysis, first a strictly YoY approach yields a mid-April date that varies a bit depending on how you choose to handle the first few months of 2009. But remembering the strong historical evidence behind the "40% trend" for the holidays and comparing it to the strictly YoY analysis we find the methods disagree about the final holiday numbers by more than a million units with the YoY trend producing the lower result.

I'm inclined to go with the higher 40% trend for two reasons; First, it is a very strong trend that has proven reliable time and again, and second we know production ramp ups are still under way and its very likely that the strictly YoY analysis based on the first half of the year will produce underestimated results.

All in all if my assessment is accurate then the change would shift the date forward by aproximately a month yielding March 20th instead of April 17th.

*Note, I'm not trying to change your mind, just sort of thinking out loud about my process.

I calculated 27th March 2009 by assuming three things:

- Wii sales will be around 45 million by the end of 2008.

- production will still be at 2.4 million/month.

- in January-March 2009 sales will be at ~75% of production like in January-March 2008.

I think they're pretty reasonable assumptions. It's of course very hard to estimate the precise week when it happens, but I think I'll be pretty close.

@NJ5,

Yeah with these kind of predictions I wouldn't even consider a difference of a few weeks to be a real meaningful difference in method. I'm honestly kind of happy to see how clustered we are so far, particularly if we were all using different methods.

I'll have week ending march 13th 2009.