General Discussion: 2018 Investor Roundtable

[JRP3]

So apparently this happened https://pagesix.com/2018/03/25/elon-musks-father-had-a-baby-with-his-stepdaughter/

The dad of tech billionaire Elon Musk says it was “God’s plan’’ for him to father the child he had with his stepdaughter.

Errol Musk, 72, told The Sunday Times of London that 10-month-old Elliot is an “exquisite child.”

Yeah, I can see how Elon may have had some issues with his father. Something for those to consider when they try to suggest Elon was some sort of spoiled rich kid.

 

[anticitizen13.7]

So apparently this happened https://pagesix.com/2018/03/25/elon-musks-father-had-a-baby-with-his-stepdaughter/



Yeah, I can see how Elon may have had some issues with his father. Something for those to consider when they try to suggest Elon was some sort of spoiled rich kid.

 

[ValueAnalyst]

today,
3 days ago,
lols

Urban Dictionary: devil's advocate
Urban Dictionary: troll

 

[SteveG3]

Urban Dictionary: devil's advocate
Urban Dictionary: troll

I was literally thinking of ending that last post with, “VA, are you trolling me?” rather than “lols”, but I thought the former was too harsh, and the latter was more suggestive of- hey, let’s just take all this in good humor (honestly, exact thought process when I finished that post). Perhaps you are somewhat psychic, but just not quite fully tuned in, lols

 

[Bobfitz1]

JRP3 said: ↑
I wonder if Tesla will do a P Model 3. Might cut into Model S P sales if they enable the full potential of the 3.

The alternative is to cede this profit rich niche to BMW M3, MB AMG C, Lexus IS F Sport,Cadillac ATS V and Audi S4.

I believe Tesla can and will split the difference. At least for a few years. What's been learned about M3 LR battery pack supports the idea that the larger battery and reduced weight give 3 a great deal of performance potential.
Independent testing has shown greater than reported performance. 4.6 - 4.7 sec. In the past with MS and MX, when dual motor versions become available, they have been significantly quicker due to the additional power of the smaller front motor. My WAG is the AWD M3 will do 0 - 60 in 4.0 sec plus or minus .1 sec. That would be very quick. Only a few tenths more than a sport Roadster's 3.7.

I believe (but can't prove), there would still be SW limiting of M3 AWD acceleration going on. That would allow a P version to further reduce 0 - 60. Maybe close to 3.0? That level performance would not challenge the P100D with or without Ludicrous, therefore those wanting the fastest Tesla will still buy the top of the line offering.

 

[Esme Es Mejor]

JRP3 said: ↑
I wonder if Tesla will do a P Model 3. Might cut into Model S P sales if they enable the full potential of the 3.


I believe Tesla can and will split the difference. At least for a few years. What's been learned about M3 LR battery pack supports the idea that the larger battery and reduced weight give 3 a great deal of performance potential.
Independent testing has shown greater than reported performance. 4.6 - 4.7 sec. In the past with MS and MX, when dual motor versions become available, they have been significantly quicker due to the additional power of the smaller front motor. My WAG is the AWD M3 will do 0 - 60 in 4.0 sec plus or minus .1 sec. That would be very quick. Only a few tenths more than a sport Roadster's 3.7.

I believe (but can't prove), there would still be SW limiting of M3 AWD acceleration going on. That would allow a P version to further reduce 0 - 60. Maybe close to 3.0? That level performance would not challenge the P100D with or without Ludicrous, therefore those wanting the fastest Tesla will still buy the top of the line offering.

I don’t understand why people are debating whether they will do a performance Model 3:
Performance Tesla Model 3 confirmed by Elon Musk, via Twitter of course

 

[ValueAnalyst]

The following graph is based on the data submitted to the google spreadsheet. I seperated out VIN assignments from before the factory shutdown (blue) and after the factory shutdown (red). I took March 7th as the separation date. This is possibly a little arbitrary and I am open to rerun the data with a different date but I think it is a good fit. It's nearly two weeks after the shutdown to allow time for the production improvements to actually trickle down to VIN assignments. March 7 is also the first VIN assignment storm by Tesla. In addition to the two data sets, I also graphed the linear trendlines for both sets. Here is the result :

View attachment 288898

It's quite clear that there are two different trends. Production most certainly picked up. So far the good news. Now for the less good news : the slope of the two trendlines is the number of VINS assigned per day (assuming there are no gaps in the VIN assignment!) For the blue series this slope is 75 cars/day or 525 per week. The red slope is 135 cars/day or 945 per week. This is clearly much less than everyone is currently hoping (between 1000 and 1500; I've seen some here mentioning close to 2000/week after the NHTSA disclosures).

How is that possible? I think few here really realized how bad VIN progression actually was before the factory shutdown. Most still had that 1000/week from the end of year in mind, maybe discounted a little but certainly not by half. So even an almost doubling of the production rate still comes down to less than 1000 cars/week.

Note : I had to lightly edit the graph to crop the trendlines to just the data range that was relevant for each.

This is what the post 3/19 period extrapolate to using current data (mine is slightly stale from last night). The linear fit is the maroon solid line, it's clearly not realistic since it predicts a 5k/wk rate. I also hand-drew in a 2k/wk line (green dash) just to see what it looks like. I think it's possible that it could fit. I agree that we need next week's data to be able to tell.

View attachment 288985

Gosh... everybody has a line these days. Well, here's mine FWIW.

 

[Mike Smith]

Great analysis by TMC moderator TEC shows that the Tesla crash and death is 100% the result of negligence by CalTrans, resulting from bad lane markings and a non-functioning safety barrier:

Model X Crash on US-101 (Mountain View, CA)
Model X Crash on US-101 (Mountain View, CA)

Here is a similar case, which NTSB determined was a result of bad lane markings:

San Jose: Fatal Greyhound bus crash blamed on poor road markings

Even if Tesla Autopilot was on it won't be found to have been the cause of the crash.

 

[mongo]

In the past with MS and MX, when dual motor versions become available, they have been significantly quicker due to the additional power of the smaller front motor.

The power helps, but the big advantage is the extra friction the car can use. With RWD and a 50/50 weight split, if you have performance tires with a coefficient of friction of 0.9 (standard tires are 0.7 ish), you can only pull 0.45G. Actual value is slightly more due to the CG causing increased loading of rear axle during acceleration, Tesla's CG is really low so less benefit.

With AWD, the full vehicle mass is available for friction, so 0.7G max on standard tires vs 0.35G plus weight shift for RWD.

0.7G gets you a 0-60 of 3.93, 0.35G gets you double that. A 3 second 0-60 requires a friction coefficient of 0.92. The 1.9 0-60 of Roadster 2020 needs sticky tires at 1.48. Some tires go all the way to 1.7 .

AWD will be good for stock price.

 

[ValueAnalyst]

This is what the post 3/19 period extrapolate to using current data (mine is slightly stale from last night). The linear fit is the maroon solid line, it's clearly not realistic since it predicts a 5k/wk rate. I also hand-drew in a 2k/wk line (green dash) just to see what it looks like. I think it's possible that it could fit. I agree that we need next week's data to be able to tell.

View attachment 288985

How do people see that Green Line (i.e. 2,500/wk) and say last week was at 1,500/wk?

Almost all dots at the beginning of the GL is below the line, and almost all dots at the end are above the line. Clearly an underestimation.

Especially excluding the bottom handful of data points, which I believe represent extreme QC-issues from months ago, so not representative.

What am I missing here? Is it just that we were burned by the end of December VIN registration? Lack of sufficient data points?

 

[mongo]

How do people see that Green Line (i.e. 2,500/wk) and say last week was at 1,500/wk?

Almost all dots at the beginning of the GL is below the line, and almost all dots at the end are above the line. Clearly an underestimation.

Especially excluding the bottom handful of data points, which I believe represent extreme QC-issues from months ago, so not representative.

What am I missing here? Is it just that we were burned by the end of December VIN registration? Lack of sufficient data points?

I'm getting 3,750 from that graph using the high, but not too high, VIN linear line fit.

 

[ValueAnalyst]

I'm getting 3,750 from that graph using the high, but not too high, VIN linear line fit.

yea.. same here. If that's the case, which can only happen if the Grohmann line is in, then we should see a flurry of new VINs in the next 48 hours.

 

[Mike Smith]

View attachment 289146

Gosh... everybody has a line these days. Well, here's mine FWIW.

To get a better sense of the ramp it would be helpful to remove the one week time period when production was shut down in February and compress the chart. It actually starts looking like a pretty smooth parabolic ramp.

 

[Waiting4M3]

How do people see that Green Line (i.e. 2,500/wk) and say last week was at 1,500/wk?

Almost all dots at the beginning of the GL is below the line, and almost all dots at the end are above the line. Clearly an underestimation.

Especially excluding the bottom handful of data points, which I believe represent extreme QC-issues from months ago, so not representative.

What am I missing here? Is it just that we were burned by the end of December VIN registration? Lack of sufficient data points?

The R^2 of the linear fit from 3/19 on is 0.25, which basically means that only ~ 25% of the spread of the VIN# can be explained (modeled) by a time trend model of 5k/wk, and there are other 75% influenced by other factors.

Another way to say this is, if you look at the VIN distribution in a 5 day time period (from 3/19-3/24), most of the spread comes from other factors instead of ramp progression. There are ~70 VIN assignments in these 5 days. If by random chance a few of the cars get assigned a few days earlier or a few days later, the linear fit of this dataset could yield a completely different slope. This randomness is part of the 75% variation that is not modeled by the 5k/wk time trend.

In contrast, if you look at Troy's chart over the entire time period from Dec till now, it has a R^2 of 0.85 now. This is the power of modeling time trend over a longer time period. Shifting a few cars earlier or later by a few days doesn't change the slope much.

 

[Mike Smith]

Like Schonelucht, I was also wondering if the spreadsheet was just getting more popular, but now it looks like assignments under 10xxx are already petering out and 15 of the last 17 assignments are above Troy's trend line. This seems statistically significant.

 

[ValueAnalyst]

The R^2 of the linear fit from 3/19 on is 0.25, which basically means that only ~ 25% of the spread of the VIN# can be explained (modeled) by a time trend model of 5k/wk, and there are other 75% influenced by other factors.

Another way to say this is, if you look at the VIN distribution in a 5 day time period (from 3/19-3/24), most of the spread comes from other factors instead of ramp progression. There are ~70 VIN assignments in these 5 days. If by random chance a few of the cars get assigned a few days earlier or a few days later, the linear fit of this dataset could yield a completely different slope. This randomness is part of the 75% variation that is not modeled by the 5k/wk time trend.

In contrast, if you look at Troy's chart over the entire time period from Dec till now, it has a R^2 of 0.85 now. This is the power of modeling time trend over a longer time period. Shifting a few cars earlier or later by a few days doesn't change the slope much.

Yes, I know what a coefficient of determination is, thank you. Would you like to discuss multivariable Calculus?

If you look at the last three days, and exclude the one outlier way below the line, which is most likely a major QC-lag, then R2 = 1 - 1 = 0

In English, that line explains *sugar*.

We need another best fit line that covers ONLY the period after the late-Feb production shutdown.

 

[mongo]

The R^2 of the linear fit from 3/19 on is 0.25, which basically means that only ~ 25% of the spread of the VIN# can be explained (modeled) by a time trend model of 5k/wk, and there are other 75% influenced by other factors.

Another way to say this is, if you look at the VIN distribution in a 5 day time period (from 3/19-3/24), most of the spread comes from other factors instead of ramp progression. There are ~70 VIN assignments in these 5 days. If by random chance a few of the cars get assigned a few days earlier or a few days later, the linear fit of this dataset could yield a completely different slope. This randomness is part of the 75% variation that is not modeled by the 5k/wk time trend.

In contrast, if you look at Troy's chart over the entire time period from Dec till now, it has a R^2 of 0.85 now. This is the power of modeling time trend over a longer time period. Shifting a few cars earlier or later by a few days doesn't change the slope much.

Non-statistician opinion:
I think r^2 It pretty pointless for this type of data. This is not a linear set/ function that is being modeled, it's multiple Y values for the same X. The bigger the VIN spread the worse the correlation will be. You can make lines that fit the error function better, but that is just curve fitting the days' VIN distribution. As long as there are low numbered outliers, the fit will always be less that the real, more so as time goes on. If we take the top 2-5 points for each day and fit those, it does tell how fast the max VIN is rising.

 

[Waiting4M3]

Like Schonelucht, I was also wondering if the spreadsheet was just getting more popular, but now it looks like assignments under 10xxx are already petering out and 15 of the last 17 assignments are above Troy's trend line. This seems statistically significant.

Yes. If we assume 50/50 chance of VINs being above and below the trend line, then this observation can be modeled by thinking of it as 15 heads coming up in 17 coin flips, a quick online search using coin flip simulator shows a 0.001 chance of this occurring randomly, so it does seem that it's significant.

 

[Mike Smith]

Yes. If we assume 50/50 chance of VINs being above and below the trend line, then this observation can be modeled by thinking of it as 15 heads coming up in 17 coin flips, a quick online search using coin flip simulator shows a 0.001 chance of this occurring randomly, so it does seem that it's significant.

It's also deviation in the direction you'd expect in an exponential ramp, adding confirmation.

 

[Waiting4M3]

Non-statistician opinion:
I think r^2 It pretty pointless for this type of data. This is not a linear set/ function that is being modeled, it's multiple Y values for the same X. The bigger the VIN spread the worse the correlation will be. You can make lines that fit the error function better, but that is just curve fitting the days' VIN distribution. As long as there are low numbered outliers, the fit will always be less that the real, more so as time goes on. If we take the top 2-5 points for each day and fit those, it does tell how fast the max VIN is rising.

A practical and easy way to check validity of a linear fit, is to artificially "mess up" the data slightly and see if the trend still holds. For example I took 2-3 of the lowest VINs reported from 3/20-22, and moved them to 3/25, which caused the trend to look less "rampy".This basically simulate some randomness in when VIN assignments may occur or reported on Troy's sheet. The resulting linear fit slope is reduced from ~5k/wk to ~2.5k-3k/wk. The conclusion I draw from this exercise is that it's really difficult to say it's 5k/wk with any accuracy, but it does look like it's 2k+/wk.