Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

[powertoold]

Tesla's cultural advantage

End of the day, all that matters for the SP is revenue and profits, so let's hope Tesla's many moats lead them to sky high revenues and profits

 

[Curt Renz]

more proof those guys are dinosaurs

minutes long bash segment

tim, carter, all of them, slimy

the same repeat nonsense OPINIONs portrayed casually and confidently as fact, in order to force an uninformed listener to feel stupid not to believe them. they’re trained to prey on the psychology of the viewer, and are pathological liars.

trash

rated helpful (thanks for the deep track curt) so the drones out there can stop believing everything the ‘professionals’ say on TV

The CNBC "Fast Money" panel appears to be a cadre of money managers hoping to promote their financial holdings. For all we know, they may not be paid. They would not have been paid on my show. If they had been on my show, they would have been required to state their holdings (long or short) in a stock being discussed. Particularly in the case of Tesla which pays no advertising dollars to CNBC (or anyone else), the producers and anchors apparently feel no need to intercede as devil's advocates when the panel is unanimously dissing Tesla.

Of course if anyone on the panel had recommended Tesla that day a little more than 5 months ago, they could gloat tomorrow about its 52% rise since then.

 

[TNEVol]

Anyone willing to FOIA UBS&Tesla from SEC?

SEC.gov | Request for Copies of Documents

If I recall correctly FOIA request will not work for current open investigation matters but hey maybe they would say "we cannot respond re matter which might be under investigation". It interesting that there were leaks aplenty on SEC v Musk but keep in mind that the folks at the SEC want to grab the gold ring into the private sector and likely curry favor.

 

[ReflexFunds]

Agree, just a couple numbers to tweak.
Tesla says one crash per 498,000 miles per NHTSA Tesla Vehicle Safety Report
4.34 million miles per accident while on Autopilot.

I really don't know why Tesla reports the 498,000 miles number, it is definitely the wrong NHTSA number to compare to.
498k miles is the number of miles driven per crash, it is not number of miles driven per crash per car.
The NHTSA reports there are 1.8 cars on average per crash. So in reality looking at it from the perspective of a single driver, a driver is likely to crash their car every 498,000/1.8 miles = every 277k miles.
This is the NHTSA number most comparable to the numbers Tesla report.

However, this doesn't account for the fact that NHTSA only tracks accidents which are reported to police or insurers and not accidents which aren't reported (so as to avoid higher insurance rates).
Surveys suggest in reality cars crash every 100-150k miles, but less than half of these accidents get reported.
Tesla's crash data is detected by Autopilot - so it doesn't matter if it gets reported, it should include everything. In reality I expect some minor low speed scrapes do not get detected by Autopilot. I would guess the accidents Tesla tracks are equivalent to somewhere between the estimated overall US average crash rate per car (every 150k miles) and NHTSA US reported crash rate per car (every 277k miles).

I don't think Tesla's crash rate on Autopilot disclosure is very easy to analyse because it is very hard to know how much it is driven by different general accident risks per mile for the road types/driving scenarios it is enabled. However it is very suggestive that AP + human supervision is safer than Human plus Autopilot Safety Features, which itself is far safer than a human driving a normal car.
The data Tesla discloses for crash rate for Teslas using Autopilot enabled features is clearer and easier to compare - in 3Q19 Tesla reported one crash per 2.7 million miles for cars without Autopilot but with Autopilot's free safety features. So it looks like around 10x less crashes per mile relative to the 277k mile NHTSA rate.

In terms of fatalities per mile, complete data on every accident is reported for the US each year by the NHTSA. While Tesla is not split out from a category called "other domestic brands", it is still possible to use this data to establish a worst possible case for Tesla fatalities - the result is that in 2018 at worst Tesla cars had 3.7x less fatal accidents per mile (relative to the US market average) and beat results from every other brand, even when limited to their 2015-2018 model cars.
This include benefits both from crash safety engineering and Autopilot. I would guess it is harder for Autopilot safety features to avoid high speed sudden accidents which would have resulted in fatalities (relative to avoiding lower speed more minor collisions), so it makes sense that Teslas reduce the fatality rate by a lower multiple than they reduce overall crash rates.

My full model on the fatalities data is below.

I don't know how IIHS arrives at its data exactly and why it takes so long to compile. Much of the key data must come from the NHTSA's database though which is much more up to date and where you can access the raw numbers directly. Secondly, I don't know where you get the Tesla death rate from. Other brands are recorded separately by NHTSA but not Tesla, if this death rate has been complied/extrapolated from a Tesla short death list i'd be highly skeptical of the accuracy.

Anyway, the data we do have from Tesla and NHTSA is much more up to date and much more comprehensive than these IIHS and Tesla death list numbers, so I don't know why you wouldn't give them more weight.

You can pull full information on all 2018 fatal traffic accidents from NHTSA. ftp://ftp.nhtsa.dot.gov/fars/2018/National/
Filtering this data, you can see in 2018 there were 51.9k vehicles involved in fatal accidents, 36.6k fatalities, 10.5k cars with model year > 2014 involved in fatal accidents and 102 BMW cars with model year > 2014 involved in fatal accidents.
Separately, you can add up from BMW reports and government data all BMW's sold in 2015-2018 and all US vehicles sold in 2015-2018. Weighting 2018 sales at 50% (to account for average on the road during 2018), this totals 1.120 million average fleet size of "New" BMWs and 60.6 million average fleet size of "New" US vehicles.
Multiplying these numbers by 13.5k average miles driven per year in the US in 2018 you can get to an estimate of total miles driven by these vehicles in 2018. Following this, just divide by the number of fatal accidents to arrive at "Number of miles per fatal accident per vehicle".

These numbers arrive at:
Total US fleet (taking 276m registered US vehicles): One fatal crash per 72 million miles.
Total US cars with Model year 2015-2018: One fatal crash per 78 million miles.
Total BMW cars with model year 2015-2018: One fatal crash per 148 million miles.

Just a quick note here - this is fatal crashes per vehicle per mile, not fatalities per mile. There are on average 0.7 fatalities per fatal crash per car (less than 1 because often 2 cars are involved in an accident with both cars counted separately in the fatal crashes statistics).

These numbers make sense. Newer cars are safer than the fleet average (but not dramatically). Luxury cars like BMW which can afford more accident avoidance technology and higher crash safety engineering are 2.1x safer than the overall US fleet.
But remember, the best estimate we have using the comprehensive and up to date data from Tesla is that Tesla's currently crash 18x less frequently than the US fleet average. This does not even include crash safety engineering benefits which should reduce the probability of death when in an accident.

So for new BMW cars we have: Accident Avoidance features plus Crash Safety Engineering = 2.1x safer than the US fleet.
For Tesla cars we have: Accident avoidance alone = 18x safer than the US fleet average (this isn't entirely fair because it's possible AP safety features find it easier to avoid small accidents than serious ones, but likely in the right ballpark).

I find this very difficult to bridge to your claim that Tesla's crash more often than other modern luxury cars.
Even using the (I presume) Tesla short death list sourced and exaggerated Tesla death rates which also don't account for Tesla's current rate of Autopilot safety, Teslas historic deaths per mile look far lower than modern BMWs.

Edit:
For the sake of completeness. 2015-2018 models miles per fatal accident vs the US fleet average:
BMW 2.1x
Jaguar 2.4x
Porsche 3.6x
Lexus 2.7x
Audi 3.0x
Mercedes Benz 2.6x

Edit again:
Actually, it looks like you can more or less get Tesla's numbers from the NHTSA data. Tesla is included in a category called "other domestic brands", but Tesla must be the majority of these brands sales. Taking the most conservative assumption that Tesla is 100% of this category - vehicles in this category built after 2014 were involved in 11 fatal accidents in 2018.

So using the most conservative assumptions, the worst case from NHTSA data is that Tesla sits at 3.7x less fatal accidents per mile than the market average and beats every other brand.

 

[PhaseWhite]

The CNBC "Fast Money" panel appears to be a cadre of money managers hoping to promote their financial holdings. For all we know, they may not be paid. They would not have been paid on my show. If they had been on my show, they would have been required to state their holdings (long or short) in a stock being discussed. Particularly in the case of Tesla which pays no advertising dollars to CNBC (or anyone else), the producers and anchors apparently feel no need to intercede as devil's advocates when the panel is unanimously dissing Tesla.

Of course if anyone on the panel had recommended Tesla that day nearly six months ago, they could gloat tomorrow about its 52% rise since then.

Agreed, CNBC should at a minimum be disclosing the hosts positions when they are talking about a stock on the program. They used to be much better at this, and I specifically recall you could see it posted in articles along with FastMoney clips posted on CNBC's website.
Here's an old example from 2012 which is what I was thinking about: Your First Move For Thursday January 26th

And some discussion of how the disclosures had gone down hill at that point:
CNBC Fast Money disclosures incomplete at best, misleading at worst

Perhaps they got push back from their star hosts on this and decided it wasn't *legally* required or the new host Melissa Lee has just dropped the ball.

 

[Thekiwi]

I am not so sure Cathie Woods would agree.

At 4m30s, Woods says that Ark believes that in 2024 (i.e. for their 5-year forecast) total EV sales will be 37M units. She elaborates that this would be more than a third of total auto sales - which is consistent with current, actual annual total auto sales on the order of 100M, so I think she did not misspeak.

So 17% of that would imply the belief that in 2024 Tesla sells 6.3M cars. Starting from 360k in 2019 Tesla would need an annual growth in deliveries of 77% to deliver that many cars in 2024.

Her old bull case of 11% would imply the belief that in 2024 Tesla sells 4.1M cars. Starting from 360k in 2019 Tesla would need an annual growth in deliveries of just over 62% to deliver that many cars in 2024.

Her bear case at 6% would imply the belief that in 2024 Tesla sells 2.2M cars, which Tesla would achieve with an annual delivery growth of close to 44%.

There has been assumptions elsewhere of an annual delivery growth of 50%. Over 5 years, and starting from 360k in 2019 Tesla would in 2024 sell 2.74M cars, corresponding to 7.5% of a 37M unit market.

I would say that her bear and bull cases of Tesla's market share in 5 years are very optimistic, but not entirely unrealistic.

If the 37M forecast is correct, then an annual growth of more than 77% would be required for the next 5 years in order for Tesla to capture more than its current market share, something I find hard to imagine.

I find it hard to imagine that in 48 months time (beginning of 2024) that the auto industry will be anywhere near a 37 million annual run rate for the simple fact that there will be nowhere near that much battery supply available.

Thats not to say I'm in general agreement that even at the base 6% marketshare case of the overall auto market once at 100% EV, that Tesla wouldn't be doing extremely well if margins are similar to today.

 

[mongo]

I really don't know why Tesla reports the 498,000 miles number, it is definitely the wrong NHTSA number to compare to.
498k miles is the number of miles driven per crash, it is not number of miles driven per crash per car.
The NHTSA reports there are 1.8 cars on average per crash. So in reality looking at it from the perspective of a single driver, a driver is likely to crash their car every 498,000/1.8 miles = every 277k miles.
This is the NHTSA number most comparable to the numbers Tesla report.

Unless the accident is really strange, only one of the 1.8 vehicles (drivers) caused the crash. So from an insurance point of view, 277k is the important number (car rate of collision), but from a driver reliability pov, 498k is the proper one.

 

[dc_h]

@PBRStreetGang7

From the Tweet:

"North America: Thousands of cars are leaving the intermodal facility in Birmingham, Alabama, for destinations on the east coast of the United States. $TSLA"

I didn't know about the intermodal in Birmingham. Can anyone provide more information about that. Are cars being transported by rail to that location? In the past three weeks I seen trucks loaded with Teslas on I-85 going north from Atlanta. That didn't make sense to me then, but now it does if east coast deliveries go through Alabama.

I read that also and wasn't sure it made sense. Maybe cutting the Chicago bottleneck out of east coast shipments and sending trains south to LA and then across the southern corridor. It doesn't seem to make sense from a geographic perspective, but perhaps there is some logistics magic I don't get. I could see someone running some logistics algorithms to look for optimizations.

*In fact, if anyone needs a college junior to intern this summer, with advanced algorithm and Python skills, DM me. I have a college sophmore (Junior by hours) doing some cool math and machine learning. Completely serious.

 

[ReflexFunds]

Unless the accident is really strange, only one of the 1.8 vehicles (drivers) caused the crash. So from an insurance point of view, 277k is the important number (car rate of collision), but from a driver reliability pov, 498k is the proper one.

It doesn't matter who caused the crash. It is just as important, if not more important to be able to avoid accidents from other road user's errors as it is to avoid making errors yourself.
Tesla Autopilot detects a crash, it doesn't know who caused it. And if it does, it doesn't then remove this crash from its data just because it determines it was someone else's fault.

I'm pretty sure Tesla only reports the 498k number because someone rushed putting together the safety report and didn't spend time thinking what number they should use from NHTSA.

Just think of it this way. Person A drives 250k miles over 10 years before crashing into a wall. Person B does exactly the same. This is 2 crashes per 500k miles or one crash per 250k miles.
At the same time Person C and Person D both drive 250k miles in their separate cars over 10 years before crashing into each other. This is one crash per 500k miles in the NHTSA data. But If both of these cars just happened to be Teslas, Tesla would have detected two crashes here, or one crash per 250k miles. Also, Person C caused the crash, but if Person D had had an Autopilot enabled car, his accident avoidance features would have dodged the accident.

 

[Darkfox021]

I find it hard to imagine that in 48 months time (beginning of 2024) that the auto industry will be anywhere near a 37 million annual run rate for the simple fact that there will be nowhere near that much battery supply available.

Thats not to say I'm in general agreement that even at the base 6% marketshare case of the overall auto market once at 100% EV, that Tesla wouldn't be doing extremely well if margins are similar to today.

I can't say if the rest of the industry can keep up, but Battery Day is supposed to show a path to 2TW of battery production per year. Even with EM-time, 48 months seems like a long way out for Musk, but let's assume it's the time frame for the plan. 37 million cars with a ~75KW pack is 2.775 TW of batteries.

Tesla may own 72% of the market?

 

[Todd Burch]

Regarding the UBS rumors: I recommend everyone ignore this rumor for now unless we get some more substantive independent confirmation.

Tesmanian is the company run by "Vincent", well-known Tesla uber-bull on Twitter. Not to say it isn't true, but we'd be best not to take and run with rumors with no substantiated facts when we are often critical of the anti-Tesla crowd who do the same thing.

 

[mongo]

It doesn't matter who caused the crash. It is just as important, if not more important to be able to avoid accidents from other road user's errors as it is to avoid making errors yourself.
Tesla Autopilot detects a crash, it doesn't know who caused it. And if it does, it doesn't then remove this crash from its data just because it determines it was someone else's fault.

I'm pretty sure Tesla only reports the 498k number because someone rushed putting together the safety report and didn't spend time thinking what number they should use from NHTSA.

It does matter if you are saying the 498k is high. If a driver screws up every 200k miles and only crashes their car, that is a crash rate of 200k. If they always hit someone else, that would be a 100k crash rate using your approach. The driver is (arguably) equally bad in each case.

The defensive aspect is difficult to factor in since it is situation dependent. Was the accident avoidable?

Whether the Tesla was at fault is an important factor indeed, and skews the Tesla number to a higher accident rate than AP alone causes.

But I'm way OT.

 

[StealthP3D]

Yea but his bottom line after all his verbal nonsense is

" You, by contrast, should avoid shorting this stock in any form or fashion"

Taken in context then, that means TSLA is currently a "screaming buy", LOL!

 

[jerry33]

The defensive aspect is difficult to factor in since it is situation dependent. Was the accident avoidable?

Whether the Tesla was at fault is an important factor indeed, and skews the Tesla number to a higher accident rate than AP alone causes.

Regardless of which car was at fault, if there are many fewer accidents in total, that says a lot about the number of accidents that were avoided. And if you only count the ones where the Tesla was at fault, doesn't that actually lower the number of accidents per mile? If there are 400 accidents total, but you only count 200 because those where where the Tesla was at fault, that makes a 200/X miles vs. 400/X miles.

 

[StealthP3D]

I hope UBS can be found guilty and fined, followed by a class action lawsuit from shareholders.

I like to think bigger.

My hope, but of course not my expectation, is that the SEC figures they can damage their tainted reputation with an investigation that doesn't stop at USB. That they follow-up on all the TSLA related complaints filed and follow the information wherever it takes them. That by taking down the biggest network of anti-American, anti-capitalism, anti-free-market TSLAQ scoundrels, liars and connivers, they can patch up their reputation which has been in shambles since sometime around the time Bernie Madoff started growing his pyramid scheme.

A guy can dream, can't he?

 

[Boomer19]

or the new host Melissa Lee has just dropped the ball.

she’s in a difficult spot considering she’s married to tesla bull ben kallo. so maybe she gets a pass.
it would be super messed up if she was married to tamberrino or langan

 

[StealthP3D]

There may be more, but both Senator Ed Markey (Massachusetts, Democrat) and Senator Richard Blumenthal (Connecticut, Democrat) today have used the complete non event of a Tesla crash into a police cruiser to argue to ban Autopilot.

Well then, that pretty much makes it fair game to use non-Autopilot crashes into police cruisers to argue that cars without Autopilot should be banned.

I'm not joking here.

 

[Curt Renz]

Well then, that pretty much makes it fair game to use non-Autopilot crashes into police cruisers to argue that cars without Autopilot should be banned.

I'm not joking here.

In the not too distant future some folks may whine, "What? I'm not getting into a car driven by a human!"

 

[Thekiwi]

I can't say if the rest of the industry can keep up, but Battery Day is supposed to show a path to 2TW of battery production per year. Even with EM-time, 48 months seems like a long way out for Musk, but let's assume it's the time frame for the plan. 37 million cars with a ~75KW pack is 2.775 TW of batteries.

Tesla may own 72% of the market?

It would be nice - but even if Tesla could make 2TW of batteries in 2024, I think they might have a little bit of a difficult time producing 26.6 million vehicles (unless they have another 10 or 20 unannounced Gigifactories they plan to build in the next 48 months.)

 

[Crowded Mind]

It does matter if you are saying the 498k is high. If a driver screws up every 200k miles and only crashes their car, that is a crash rate of 200k. If they always hit someone else, that would be a 100k crash rate using your approach. The driver is (arguably) equally bad in each case.

The defensive aspect is difficult to factor in since it is situation dependent. Was the accident avoidable?

Whether the Tesla was at fault is an important factor indeed, and skews the Tesla number to a higher accident rate than AP alone causes.

But I'm way OT.

The point is that if someone hits a Tesla, that would be included in Tesla's crash rate, regardless of fault. So without @ReflexFunds proposed adjustment, it may not be apples-to-apples.

Example:
Driver Ford drives 198k miles
Driver Tesla drives 300k miles

Driver Ford hits Driver Tesla. 498k total miles driven with 1 crash, but two cars. Tesla would report 300k miles driven per crash even if not at fault. NHTSA would report (I think) 498k miles per crash. This makes Tesla look worse than the total population (of 2, here), when in reality it went more miles without crashing on a per car basis. If NHTSA number is adjusted for per car, it makes more sense. Miles driven per car per crash is 249k miles in this case, so Tesla's 300k accurately reflects as superior.