TSLA Market Action: 2018 Investor Roundtable

[Fact Checking]

2. Carsonight commented twice on Electrek just 2 days ago that "Tesla is making in excess of 6k battery packs per week. " Carsonight is known to have good inside info on GF1.

To further support your argument, two days ago carsonight wrote:

"I have friends and family working at GF1 and it's running 24/7. It ran right through Thanksgiving and until yesterday they did not even know if they're going it be getting Christmas off."​

Which suggests they are supply limited.

One detail I haven't seen @HG Wells mention is that the reduction in VIN reporting has three independent driving forces:

• Lower new owner enthusiasm,
• faster delivery times due to improved logistics, which gives even enthusiastic owners less time to register the VIN (fewer of them will bother after the wait is over, so "days waiting with VIN" matters to the survey sample rate),
• higher rate manufacturing.

VIN to delivery days were cut in half since the summer. This number is mostly demand invariant, because it is measured roughly from start of production to taking delivery. I.e. this number dropping significantly suggests much smoother logistics - Tesla is out of "logistics hell".

So I have the impression that @Troy's estimates for Q4 Model 3 production of 53k are perhaps a bit too conservative this time around. Note that I estimated 56k Model 3 deliveries in Q3 when some people here were were predicting 65k-75k deliveries, so I don't have a tendency to be overly bullish.

(So far at least. My Q4 expectations could be way off.)

But the @Troy tracker was remarkably accurate in Q1, Q2 and Q3: 98%, 96% and 98% accurate, which gives it a heck of a track record that cannot be ignored. Also, sample rate only matters in terms of getting the highest VIN ASAP, and with several VIN reports per day even the current sampling rate is more than enough.

So maybe ~53k (production) it is in Q4.

 

[Krugerrand]

BTW, an addendum on use of derivatives:

in my academic and professional career I have worked with and/or for/regarding two Nobel Prize winners, large numbers of Ph.D. quants from 'the Chicago school' and countless people much smarter than am I. All of them, without exception, thought they could somehow use quantitative models to predict market movements. Anybody who's ever studied the academic foundations of statistics in general, stochastic models specifically and VaR(or any of the more recent concepts) specifically understands that the very nature of the subject depends on the future being closely analogous to the past.

My most famous example of that of Long Term Capital Management (look it up). What happens when the "new Russia" does nto behave like the old one? What happens if a few huge banks discover that creditworthiness does not matter and documentation is for wimps? Well, taht produces simultaneous failure of everything from Merrill Lynch to Bank of America to Wells Fargo plus dozens of others. Everyone knows that, I hope.

Today we have Trump, Erdogan, Netanyahu, ISIL and others, destabilized Saudi Arabia, internecine conflict between Qatar and Saudi Arabia, Iran feeling for power, Italy/UK/EU all in disarray plus every global/multilateral institution/agreement under attack (UN, NATO, NAFTA...). Anybody tell me if the market stochastic models have their 'boundary conditions' intact. Obviously nobody can say that because none of those models can recognize current global conditions.

None of this implies that we're headed for global depression. It does clearly show tact the global balance of power is shifting 'Eastward' at breakneck pace.

All of this is why Steve Mnuchin made his panicky call to the six big banks. Almost everyone says he overreacted. The poor reception to recent treasury auctions foretold the impending doom. I agree; he NEVER should have made a press release. Mr. Powell understood something almost nobody wants to admit, some sectors fo the economy (agricultural commodities and basic metals, for example, tourism, construction among others) are suffering both decreased demand and rising prices due to erratic trade barriers and large scale cessation of seasonal worker visa programs. Those factors combine to put the Fed in an impossible situation, partly because of the drop-off in demand coupled with rising prices. These factors are emerging now, but mostly were not in the historical statistics. Thus, I think the Treasury Secretary is wise to be terrified.

The Republican Party of the US is emulating Neville Chamberlain very, very well. They mostly know it too. Nothing will stop this until Republican donors rebel. When they do the crisis will become much, much worse before it can get better.. The global situation is painfully similar to cancer. If there is not surgery before metastasis sets in... Problem: many are refusing to accept that the present situation is cancerous.

Well, Merry Christmas to you too!

 

[Navin]

I agree with these thoughts. Troy obviously needs to stay relevant and applaud his prior and current contributions. Just don’t think a guy who has been surveying enthusiasts is equipped to survey a market that is now more mainstream. Cannot interpolate a consumer response that has shifted away from the early adopters and enthusiasts without a wide area of margin.

Also - don’t think he’s qualified to make forward prognostications and predictions on future demand. He doesn’t grasp price elasticity of demand as we move down the price curve as well as demand in Europe and China. This is just my conjecture and opinions but so are his.

 

[mongo]

2. Carsonight commented twice on Electrek just 2 days ago that "Tesla is making in excess of 6k battery packs per week. " Carsonight is known to have good inside info on GF1.

In fact two days ago carsonight wrote:

"I have friends and family working at GF1 and it's running 24/7. It ran right through Thanksgiving and until yesterday they did not even know if they're going it be getting Christmas off."
Which suggests they are supply limited.

I can't tell what you mean by "in fact", whether you were supporting or pushing back against the prior post. In case it was pushing back:
They can be both pack limited and producing >6k a week.

 

[Fact Checking]

I can't tell what you mean by "in fact", whether you were supporting or pushing back against the prior post.

I was supporting @PANN's point. I clarified that in a fresh edit.

The point I tried to make is that "24/7 utilization" is an even stronger data point than his 6k/week comment. It suggests GF1 is making packs as fast as possible - and Fremont is likely not stockpiling it.

 

[Hock1]

To further support your argument, two days ago carsonight wrote:

"I have friends and family working at GF1 and it's running 24/7. It ran right through Thanksgiving and until yesterday they did not even know if they're going it be getting Christmas off."​

Which suggests they are supply limited.

One detail I haven't seen @HG Wells mention is that the reduction in VIN reporting has three independent driving forces:

• Lower new owner enthusiasm,
• faster delivery times due to improved logistics, which gives even enthusiastic owners less time to register the VIN (fewer of them will bother after the wait is over, so "days waiting with VIN" matters to the survey sample rate),
• higher rate manufacturing.

VIN to delivery days were cut in half since the summer. This number is mostly demand invariant, because it is measured roughly from start of production to taking delivery. I.e. this number dropping significantly suggests much smoother logistics - Tesla is out of "logistics hell".

So I have the impression that @Troy's estimates for Q4 Model 3 production of 53k are perhaps a bit too conservative this time around. Note that I estimated 56k Model 3 deliveries in Q3 when some people here were were predicting 65k-75k deliveries, so I don't have a tendency to be overly bullish.

(So far at least. My Q4 expectations could be way off.)

But the @Troy tracker was remarkably accurate in Q1, Q2 and Q3: 98%, 96% and 98% accurate, which gives it a heck of a track record that cannot be ignored. Also, sample rate only matters in terms of getting the highest VIN ASAP, and with several VIN reports per day even the current sampling rate is more than enough.

So maybe ~53k it is in Q4.

56,065 Model 3s were delivered in Q3. So you're saying deliveries are DOWN from Q3. I think not.

 

[Fact Checking]

Just don’t think a guy who has been surveying enthusiasts is equipped to survey a market that is now more mainstream.

This is IMO a really unfair characterization of @Troy's methodology: it's very important to note that his production estimates do not rely on the sample rate nearly as much as say the ASP estimates.

To estimate production the main thing that matters is:

• The highest VIN reported,
• the "density" of VIN allocation by Tesla. (Tesla only uses about 80% of all registered VINs).

So even with 1% sampling rate, if the highest VIN reported is 185,000, it means there's 148,000 Model 3's made so far, which is 52,000 more than the 96,000 at the end of Q3.

The 1% sampling rate means there's a statistical uncertainty of about 100-200 VINs only - which is less than 1%.

So this is a pretty robust methodology, if we assume the 80% allocation density of VINs stays constant.

The main uncertainty comes from end of quarter effects: how strong is the push, does Tesla ramp down the factory, etc.

 

[Alphacrux]

56,065 Model 3s were delivered in Q3. So you're saying deliveries are DOWN from Q3. I think not.

Deliveries down but ramping up production for Europe?

 

[Fact Checking]

56,065 Model 3s were delivered in Q3. So you're saying deliveries are DOWN from Q3. I think not.

Sorry, the 53k is the current @Troy estimate for Q4 Model 3 production.

Deliveries should be several thousand above that. Since Q4 is special, deliveries in the 56k-60k range are possible as well, even if production is 53k.

Personally I think production will be higher than 53k, but I have no good data to support that.

 

[Troy]

Also - don’t think he’s qualified to make forward prognostications and predictions on future demand. He doesn’t grasp price elasticity of demand as we move down the price curve as well as demand in Europe and China. This is just my conjecture and opinions but so are his.

I have a survey about production and delivery estimates HERE. Recently, I added these two questions:

1. Model 3 production in the next quarter (Q1 2019 in this case)
2. Model 3 production in the next 4 quarters (Q1-Q4 2019 in this case)

So far, 30 people have answered these questions. Answers can be found in column J and K if you scroll down. Feel free to add your estimates so we can find out people who are good at long-term estimates.

 

[dc_h]

To further support your argument, two days ago carsonight wrote:

"I have friends and family working at GF1 and it's running 24/7. It ran right through Thanksgiving and until yesterday they did not even know if they're going it be getting Christmas off."​

Which suggests they are supply limited.

One detail I haven't seen @HG Wells mention is that the reduction in VIN reporting has three independent driving forces:

• Lower new owner enthusiasm,
• faster delivery times due to improved logistics, which gives even enthusiastic owners less time to register the VIN (fewer of them will bother after the wait is over, so "days waiting with VIN" matters to the survey sample rate),
• higher rate manufacturing.

VIN to delivery days were cut in half since the summer. This number is mostly demand invariant, because it is measured roughly from start of production to taking delivery. I.e. this number dropping significantly suggests much smoother logistics - Tesla is out of "logistics hell".

So I have the impression that @Troy's estimates for Q4 Model 3 production of 53k are perhaps a bit too conservative this time around. Note that I estimated 56k Model 3 deliveries in Q3 when some people here were were predicting 65k-75k deliveries, so I don't have a tendency to be overly bullish.

(So far at least. My Q4 expectations could be way off.)

But the @Troy tracker was remarkably accurate in Q1, Q2 and Q3: 98%, 96% and 98% accurate, which gives it a heck of a track record that cannot be ignored. Also, sample rate only matters in terms of getting the highest VIN ASAP, and with several VIN reports per day even the current sampling rate is more than enough.

So maybe ~53k it is in Q4.

I think on deliveries there is a wildcard, which is apparent inventory regionally delivered for an all out end of quarter sales push. Those VINs won’t show up at all, or perhaps end of quarter. I think that will add about 5000 to Troy’s model. For production, where are the cars going? Have they started shipping to China or Europe yet? We are getting feedback that Fremont is producing 7000 cars a week (3SX), but Troy is showing about 60% of that number, which I don’t think has ever diverged so much.
I’ve looked on the building Tesla site and have not seen any updates in the last month. There seem to be an extra 10,000 plus cars unaccounted for based on local production reports and Troys tracker. Looking forward to some insight on the 3rd.

 

[PANN]

But the @Troy tracker was remarkably accurate in Q1, Q2 and Q3: 98%, 96% and 98% accurate, which gives it a heck of a track record that cannot be ignored. Also, sample rate only matters in terms of getting the highest VIN ASAP, and with several VIN reports per day even the current sampling rate is more than enough.

So maybe ~53k it is in Q4.

Am not an expert with estimates, just a thought, did your estimate or Troy's estimate consider the possibility that Tesla batch build Lemurs before actual orders and then try to match them to orders? IIRC it's reported by Fred of Electrek.

 

[mongo]

56,065 Model 3s were delivered in Q3. So you're saying deliveries are DOWN from Q3. I think not.

Q3 had the inventory overhang from Q2 not going over 200k. So deliveries (58,840 from inital letter 56,065 from Q3 report (this discrepancy is likely the financing delay mentioned)) could be much higher than production (53,239). Q4 deliveries will likely closer track production. However, the 2,775 car difference in reports will also get added in to boost Q4 deliveries.

Edit: plus 8,048 in transit 3s from Q3.

 

[Electroman]

2. But guess what's the 1-year return% after those worst quarters, mostly positive!

But those are years that follow the normal pattern of bust followed by boom cycles. But these are not normal times, these are Trump times

 

[svp6]

I find current TSLA valuation very tempting to buy - but again Greenspan's "run for cover" comes to mind. I cannot figure the fundamentals of DJIA going from ~7000 to ~17000 in 8 years (2008-2016), then from ~17000 to 27000 in less than 2 years. Now that we are below 22,000, I wonder how much lower it will get. I think I will wait a little longer before buying any stock.

 

[Starno]

According to Troy's sheet, Tesla still didn't achieve steady 5k/w Model 3 production ?

 

[Right_Said_Fred]

This is IMO a really unfair characterization of @Troy's methodology: it's very important to note that his production estimates do not rely on the sample rate nearly as much as say the ASP estimates.

To estimate production the main thing that matters is:

• The highest VIN reported,
• the "density" of VIN allocation by Tesla. (Tesla only uses about 80% of all registered VINs).

So even with 1% sampling rate, if the highest VIN reported is 185,000, it means there's 148,000 Model 3's made so far, which is 52,000 more than the 96,000 at the end of Q3.

The 1% sampling rate means there's a statistical uncertainty of about 100-200 VINs only - which is less than 1%.

So this is a pretty robust methodology, if we assume the 80% allocation density of VINs stays constant.

The main uncertainty comes from end of quarter effects: how strong is the push, does Tesla ramp down the factory, etc.

Assuming that Tesla still allocates only 80% of VINs is a tricky assumption, with a big impact on production/delivery estimates. What if they’ve gone to allocating 90% of VINs? Is it known why they allocate only part of the VINs?

 

[wgood33]

I now personally don’t think Tesla will ever see $400. The bears will play the same games over and over again until Tesla has lost his advantage and the competition (VW in particular) takes over volume EV sales. In about 2-3 years would be my guess.

Yes I agree VW is going to be the leader (and this is awesome) but the TSLA price is not reflecting earnings on future Tesla products. You are ignoring the following stock price drivers:
Asian, European and Domestic Model Y sales
Asian and European Model 3
Have you noticed how many Model 3's are showing up on the roads, this is increasing exponentially in my area (this car is a total hit)
Tesla Semi
Tesla Pickup
Tesla Autonomous Product
this is Elon Musk we are talking about. He is the King of Innovation.

 

[ZeApelido]

Assuming that Tesla still allocates only 80% of VINs is a tricky assumption, with a big impact on production/delivery estimates. What if they’ve gone to allocating 90% of VINs? Is it known why they allocate only part of the VINs?

Exactly. Assuming 80% with high confidence is like assuming voter turnout in 2016 would be like previous elections. Then you end up like Nate Silver

 

[avoigt]

Yes I agree VW is going to be the leader (and this is awesome) but the TSLA price is not reflecting earnings on future Tesla products. You are ignoring the following stock price drivers:
Asian, European and Domestic Model Y sales
Asian and European Model 3
Have you noticed how many Model 3's are showing up on the roads, this is increasing exponentially in my area (this car is a total hit)
Tesla Semi
Tesla Pickup
Tesla Autonomous Product
this is Elon Musk we are talking about. He is the King of Innovation.

„VW to be the leader“ ... of what and when? More important why?

I did not see a single VW product yet be it in real or a concept or on paper that is anywhere near to compete with a Tesla regardless which one. Actually being German I would welcome that product as the unions just announced 200 k jobs are at risk in Germany to be lost in OEM forever and not to return! That’s the reality and quite down to earth.

VW has a gigantic challenge in front of them and need to reinvent the entire company. This is not done with money and the €40 bn they want to invest do not make the difference. They don’t impress me at all as this is not about investment but about change which you cannot mandate top down from management to the line workers. It does not work like that.

Just for couriousity why do you believe they will be a leader in a new technology they did not bring one compelling product out yet?

With a dropping SP the sentiment here is very changing too. Once again a good signal that the stock price may go up again. Right now I believe it’s just driven by an overly negative Marco sentiment caused by many politicians polarizing but really not by fundamentals. The latter prevail and make the prices on the long run and that’s what counts. Markets are sometimes stupid.

SP may fall further, so what? Numbers will be solid to good to surprise and all will jump in again. Don‘t be fooled and look at facts not fiction.

I scratch my head around the 3 demand discussion here and on twitter and can only say people either have not been in Europe or China yet at all or just talk about US only.

A flat demand for the 3 once deliveries go to overseas is a dream of shorts but detached from reality unless the demand in the US drops dramatically something I don‘t see.

So quite a negative sentiment and I include Troys 2019 predictions in that statements knowing his superior accuracy from the past but we talk 2019 and we talk first 3s delivered to market larger than the US with a car that’s a much better fit than in the US. No VINs for all of 2019 so no model from him and no better prediction that other people around. Again great modeling in the past but you cannot use VINs to calculate what will be produced and delivered in 2019 in 3 continents by looking at VIN count and reports. Happy to learn about the model that can do that. So let’s not get confused about prediction Modells and accurately.

To finish I like the negative sentiment including some shorts and trolls appearing again which is what we need for a good SP developement on Q1.....