You can install our site as a web app on your iOS device by utilizing the Add to Home Screen feature in Safari. Please see this thread for more details on this.
Note: This feature may not be available in some browsers.
No invites again? This sounds contradictory to the 2k cars that were supposed to be made this week.
This graph is not quite accurate because it does not take into account all the folks who are currently waiting for a VIN. I know theres a huge bunch of us who are 6 weeks deep who configured on February 22nd, that would skew the graph a bit.Based on spreadsheet data for the recent weeks, specifically VA seems to be on the longer end, currently 44-days for Config to VIN.
View attachment 291821
CJ, thanks for the hard work in putting the spreadsheet together. I used the VIN assignment data in the sheet, and did some modeling based on a hypothesis by @schonelucht that the increment of the max reported VIN could reflect line speed. I think I might have a method that can come close, some highlights of the model results:Now that we had a little time to digest the Q1'18 delivery/production report, I am re-evaluating how well the spreadsheet did in its projections. These are the projections of the various models before the report:
8,480 units based on the Buyer VINs Chart on the Invites Spreadsheet
9,126 units based on Bloomberg's Tesla Model 3 Tracker
9,609 units based on the NHTSA VINs Chart on the Invites Spreadsheet
Tesla reported Q1 production is 9766.
So,
The spreadsheet buyer VIN gets a grade of 86.8% correct.
Bloomberg gets a grade of 93.4% correct.
And the spreadsheet NHTSA VIN gets a grade of 98.4% correct.
The buyer VIN model as it is written gives too much value to historical numbers, so it is skewed by February bottle neck and is unable to cope with the burst rate of late March. The NHTSA VIN model is more forward looking and is more representative of Tesla intent on future production. We are working on a revision to shorten the time affecting the buyer VIN model to capture a more instantaneous picture of Tesla current production rate.
For now, I know which one I'm using next quarter, and yes, the VIN registration today matters a lot.
Last week, Tesla also threw us a curve ball with their batch of invites to Canada. The problem we have now is we don't have a good feel of the Canadian sample rate on our spreadsheet. I am not quite sure how a 100 reported invites in Canada compares to 100 in the US. There are, however, some information we can glean from the invited group. First, a large chunk of the Canadian reservations including most if not all of the Canadian owners have been invited in the last batch. Second, the deferral rate is much higher in Canada than it is in the US, probably because of the importance they place on AWD.
At first I thought the Canadian invites was Tesla's attempt at pushing 200K to Q3, but there is just not enough number in Canada to make that happen given the current projected production rate. So, unless Tesla starts with European delivery or they slow down production to retool for AWD, 200K will probably happen in Q2.
CJ
No invites again? This sounds contradictory to the 2k cars that were supposed to be made this week.
CJ, thanks for the hard work in putting the spreadsheet together. I used the VIN assignment data in the sheet, and did some modeling based on a hypothesis by @schonelucht that the increment of the max reported VIN could reflect line speed. I think I might have a method that can come close, some highlights of the model results:
I posted details in Investor General Discussion's thread, I hope it can be of some use.
- Daily rate ~250-280 for last week (since 3/30)
- 7 day total from 3/27-4/2 is 1803, 217 lower than actual, still lagging a little
- Total production in Q1 up to 3/31 is 10,105, ~340 off, or 1+ days of production at end of Q1 speed.
I agreed that max VIN is quite noisy, especially in Jan and Feb where the max VIN can jump up and then stay the same for weeks. Lately with higher production rate, the max VIN seem to move up more regularly, and may be more usable. I've also been thinking about avg VIN but it doesn't seem to work as well, see a long winded background below.@Waiting4M3,
Very nice how you made that work!
Just curious about your choice of the max reported VIN rather than the mean.
Seems to me that in a normal distribution, the average would better represent the sample than either extreme.
You would get less fluctuation in your trend line although at the expense of decrease sensitivity to sudden change.
CJ
Finally got call this afternoon, delivery scheduled 4/7, San Diego, Red.
CJ -Approximately 4000 invites went out last week mostly to 3/31 non-owners with a tiny portion to owners who reserved in February 2018.