Bloomberg Model 3 tracker shows production crashing below 1000/week

Just for fun, I decided to give a wack at trying to figure out Bloomberg's VIN tracker formula, and failed miserably. I took the Model 3 VIN data, top graph, each data bar represents one week and then I tried to average the sum of registrations to produce the graph below, thinking I would come up with something similar to Bloomberg's. In order to try to make the graph smoother I averaged 8 weeks up to the point plotted in the graph. I tried various other from 4 weeks to 13 weeks, but settled on 8 as being closest. What this does show is that most of the last 6 weeks was over 5k production. I guess my main finding is that you can play with the data any way you want and come to greatly different conclusions.
index.php
 

Attachments

  • VINtracker.webp
    VINtracker.webp
    21.4 KB · Views: 70
Last edited:
One thing you should keep in mind is that VIN registration shows orders they have decided to allow customers submit their desired configuration. As visually apparent a vast majority in the last 2 months have been for AWD and in regards to the VIN gaps around 80,000 you can see they are almost all AWD. Well, I guess it is not obvious unless you realize these are on August 4th and 5th Which is the highest bar and the one immediately preceding. As speculated earlier these gaps are now being filled in as shown by the latest reported VINs by Bloomberg.
 
Last edited:
@Roy_H, Thanks for trying to interpret the data. I am not sure if you calibrated your plot based on past quarters. The trend appears similar but your absolute numbers are quite different. Bloomberg looks at registered VINS and reported VINs by people taking delivery. May be they offset it too.
Anyway, their number isn't that accurate as we all now agree. The total may be somewhat closer to relaity than the weekly numbers.

BTW, it has dropped down to 2001 now. Even though the trajectory is headed south, Tom keeps predicting a flat lining of the plot. We will see. Possible that after the delivery rush, production picks up next quarter.
 
Last edited:
The raw VIN registrations in the first graph is from the same source Bloomberg uses and I believe this is very accurate. I made no attempt to incorporate any data from the volunteer VIN reporting by customers. If I read the Bloomberg site correctly, their first graph does not use any customer reported data. I used a running average of the previous 8 weeks to produce my graph. I think we can be confident that production is close to 5k/week and the 2k/wk on Bloomberg is totally unrealistic.
 
LOL. Tom is totally lost now. Moves to a 13 weeks rolling average! So he finally admits the weekly production rate is totally useless.
"The most notable change is that we’re switching to a 13-week trailing average (rolling quarter) for our weekly estimates.".

I guess it's time to close this thread for good.
 
Latest update. I have now gone to 13 week rolling average to be the same as Bloomberg, but still get a much different graphModel3VIN.webp
As before I believe we are working from exactly the same raw data shown in the top graph. I have also gone to a smooth line to get rid of the sharp peaks, the line goes through the same data points, just rounded.
 
Hmm, not sure what;s the difference.

So, it seems Tom has been reading our criticism here. and notes this (what I conspired):
This should also address concerns by some on social media that Tesla could be “gaming” our model, by over-registering VINs or skipping groups of VINs that it deploys when building cars, to appear strong during periods when the company is particularly vulnerable. We’ve never ascribed much value to these theories, but the worry should nonetheless be mitigated by switching to a trailing average that stretches over an entire quarter.
Click to expand...

Tom stil claims that his total is accurate. So I looked at the total only. It shows 95952 at th emoment ( 11.30 p.m. PST).
Here are the quarterly numbers ublished by Tesla since Q3 2017.


Tesla production Bloomberg estimate Difference BBG
2017 Q3 260
2017 Q4 2,425
2018 Q1 9,766 9,285 -5%
2018 Q2 28,578 27,957 -2%
2018 Q3 53,239 53,457 +0.4%

Total 94,268 reported by tesla till Q3 2018)

That mean, in Oct Tesla has produced 1684 cars this week in ~6 days.
 
TeslaInvestors said:
Hmm, not sure what;s the difference.
Click to expand...

Are you referring to the difference between my model and Toms? My model shows production rate of this past week of 4,732, some 600 more than Tom's at 4,104. And I would take 6 days of 4732 = 4056 and add that to the declared 94,269 and say production is 98,325. I have no idea how Tom can declare 6/7 * 4104 = 3,518 + 94,269 = 97,787 is 95,952. So the difference is 98,325 - 95,952 = 2,373.
 
There is a 13-to-1 scale ratio between quarterly and weekly numbers. As a retired engineer, changing the units needs to be done very carefully. Having a reported 8,000 units being in transit after the Q3 press release suggests a ‘short’ week gives time to tweak both production and delivery systems.

Bob Wilson
 
TeslaInvestors said:
So he finally admits the weekly production rate is totally useless.
Click to expand...

If you want to see someone who is late to the party on admitting Bloomberg's Model 3 Production Guess-O-Meter is useless, then try using a mirror.

You trying to accuse someone else of that isn't a matter of the pot calling the kettle black, it's a matter of a black hole calling the kettle black!
:p
 
bwilson4web said:
There is a 13-to-1 scale ratio between quarterly and weekly numbers. As a retired engineer, changing the units needs to be done very carefully. Having a reported 8,000 units being in transit after the Q3 press release suggests a ‘short’ week gives time to tweak both production and delivery systems.

Bob Wilson
Click to expand...
This is all heavily averaged data, and from a factory that usually runs 24hr, 7days/wk. It is definitely not precise, my argument is that both Tom at Bloomberg and I start from the same data, and use the same 13 week rolling average. Tom must be supplementing his data with something else I am unaware of. Number of unit in transit is a non-issue as we are talking about production, not deliveries.
 
An update showing this weeks VINs. Tesla is going to have to aggressively increase their ramp rate if they expect to get to 10k/wk by the end of the year.

Model3VIN_2018-10-13.webp
 
Roy_H said:
An update showing this weeks VINs. Tesla is going to have to aggressively increase their ramp rate if they expect to get to 10k/wk by the end of the year.
Click to expand...

Elon certainly gave no indication during the last investor conference call that they're aiming for production of 10k per week by the end of the year. Looks like they have scaled back to more modest ambitions, in light of what appears to be a limit to demand for the currently available options for the Model 3, including the lack of any lease option.

I find it curious that Tesla has elected to offer the Mid Range Model 3 to sustain a very high domestic demand level without offering leases for Model 3 purchasers, and without putting the Standard Range TM3 into production. Would offering leases impact Tesla's income that significantly?

At any rate, it seems that Tesla is now shifting its focus to overseas markets; with gearing up to export the TM3 to Europe and other overseas regions, as well as building the Shanghai Gigafactory at an accelerated pace. My guess is that we won't see Tesla trying for the 10k per week production level until sometime in 2019, and perhaps not even the first quarter.

As a reminder, only a few short years ago (ancient history in the Internet Age!), the plan was for Tesla to ramp up to 500k cars per year by 2020, which would have been ~100k Model S/X and ~400k Model 3's. I'd like to point out that 10k TM3's per week would come to 520,000 per year... which is substantially more than Tesla planned for 2020!

So, anyone expressing disappointment that Tesla (likely) won't reach 10k+ per week on producing Model 3's by the end of 2018... those people need to step back and look at how far Tesla has come in just the first three quarters this year. If Tesla delivers 100k cars globally in Q4 (and that's not only possible, but likely), that will represent a 250.9% increase over 2017!

Go Tesla! Keep going Tesla!
:cool:
 
I am making a number of assumptions here. One, that all registered VINs represent pre-sold cars, not 100% true but should be close enough. Two, October production equals October sales, based on the assumption that number of vehicles in transit is the same this month end as last month end. Highest VIN is currently 179,851 and last quarter production totaled 94,269 so that means 85,582 vehicles pre-sold (most not produced or delivered) since beginning of this quarter. 85582/13wks = average of 6,583/wk and end of last quarter was 5,300/wk. Therefore if production ramp is linear the year would end at (6583-5300)+6583 = 7,866/wk. Although this figure is not 10k, as Pushmi-Pullyu has pointed out is a far more likely figure and probably still high. The fact that Tesla has collected enough sales to carry them through to the end of the year is supported by Elon's statement last week that there was only 3 more days to get your order in to qualify for full federal tax credit. I would argue that my numbers here are high, that Tesla will not produce more than 85k Model 3 cars this quarter.

So, what is this all leading up to? If the current ramp-up (dashed line on my graph) is accurate, Tesla will sell about 25k Model 3 cars this month. Bloomberg, OTOH has maintained that production is nearer flat at 4,200 to 4,600 cars/wk and that means October sales will be about 21k. I am anxiously waiting for InsideEVs Model 3 sales figures to see who is closer to the mark.
 
gooki said:
My understanding the goal was always 10k cars per week by end of 2018 (S, 3, X), not 10k Model 3 alone.
Click to expand...

Clearly claims made over a year ago, before "production hell" were more optimistic than now. Musk stated then Model 3 production would be 10k/week by the end of 2018, now he is suggesting 7k.
 
You must log in or register to reply here.
Back
Top