TSLA Market Action: 2018 Investor Roundtable

[mongo]

Is TSLA now likely to qualify for S&P500 inclusion after Q4 ER?

They need 4 quarters that added together have a net profit. That wont happen in Q4, It would need about $1.3b profit to counteract the first two quarters' loss... however, It's possible by 1Q19 IMHO, for 3Q18 (+312m), 4Q18, and 1Q19 to counteract the $747m loss in Q2.

Is the inclusion based on GAAP or non-GAAP profits? if it is based on GAAP and Q4 and 19Q1 are identical to Q3 then it would qualify for inclusion after 19Q1.

If it is based on non-GAAP and Q4 is better than Q3

, about 11%/$56M more profit, ​

then it could qualify for inclusion after Q4.

So 19Q1 is a safer bet.

The financial viability rules are:
It should be GAAP profitable the previous quarter
The sum of the previous 4 quarters should be GAAP positive.
S&P could add TSLA now, if they wanted.
https://us.spindices.com/documents/methodologies/methodology-sp-us-indices.pdf
Most web sites do not have the requirements correct. Above link is the official document.

 

[EVNow]

Expecting your $39.99 pair of jeans has comparable quality with your $399 or even $179 pair of designer jeans simply doesn't make sense at all.

I've bought exactly the same clothes in India (unbranded) for 1/10 the price of it in the US. The middle man eats a lot of that $399.

 

[S3XY]

My limited understanding of AI and Python is that you need to tell the computer what is a pass or fail which is called Supervised Learning. I think you're describing Unsupervised Learning which would result in the average Joe schmoe out there. I believe you just need to throw out the bad behaviors, and maybe there's an algorithm for that even... when the car wrecks

Well, one indication of a Joe Schmoe driver would be a sudden jerk on the steering wheel. But that could just be a good driver avoiding a squirrel. I would think they would use something along the lines of standard deviations from the mean path and throw out the extremes. In other words, the most likely correct path is the one that most people take. Or, if running on Autopilot, a bad result is indicated when someone intervenes and takes over the steering or brakes to disable Autopilot. Good outcomes are indicated by the "driver" letting Autopilot continue driving.

 

[neroden]

I interpreted it as a way to say Tesla is in a good position competitively, and those opportunities still exist for the other companies in the future.

Yeaaaah.... to me he sounded frankly irritated that none of the other companies had picked up on Tesla's invitation to make electric cars. Which makes sense if you think about the mission. Their refusal to make EVs is hampering the mission of accelerating sustainable transport. And he always thought they would get on board eventually, and sooner than *this*.

Tesla wants to help the other carmakers; they don't want to be helped. What Musk is emphasizing is that the other companies *not* taking that invitation is *why* Tesla has an lead which is unsurmountable for several years.

 

[StapleGun]

Another way to put it:
Tesla does not (yet) have appropriate supervisors for supervised learning, and "doesn't wreck" or even "doesn't wreck + ends up at the right place" is not a good enough standard for unsupervised learning.

I haven't read this entire side-discussion so apologies if this has been discussed already but I think you are missing a fundamental point. Most accidents are not caused by some situation which the average driver could not avoid. Most accidents are caused by a situation where either:

1. An average driver would avoid the situation 99/100 times but for some reason makes an error on the 100th time. Think not checking a blind spot, looking down for a brief second when the car in front slams on the brakes, etc...
2. A driver does something that 99/100 drivers would not do. Driving to aggressively around other cars, too fast for conditions, ignoring signs/traffic rules, etc...

In the first set of scenarios a neural network would not copy make the same mistake 1% of the time, it would be more comparable to doing the median action every single time. So if you ordered the drivers actions with 1 being the absolute best and 100 being the worst, and only action #100 is a crash, the NN would do action #50 every single time.

The same applies to the second set of scenarios. The NN would not drive too aggressively 1% of the time, it would drive average 100% of the time (although not the absolute best).

So even if the NN drives "average" the vast majority of incidents would be avoided. Additionally things like an instant reaction time, very precise speed/distance calculations, and never being distracted will reduce a huge number of accidents on their own.

 

[neroden]

Well, one indication of a Joe Schmoe driver would be a sudden jerk on the steering wheel. But that could just be a good driver avoiding a squirrel. I would think they would use something along the lines of standard deviations from the mean path and throw out the extremes. In other words, the most likely correct path is the one that most people take.

Guaranteed way to create a computer which is a bad driver. So I sure hope they're doing something smarter than that!

 

[Intl Professor]

I suspect that's Tesla investor mythology.

Evidence of myth (meaning made up)? Or, tradition commonly accepted as true, in which case, trivial comment. Please excite our braincells more.

 

[neroden]

The NN would not drive too aggressively 1% of the time, it would drive average 100% of the time (although not the absolute best).

My point is, that's not good enough. Not nearly good enough. Look, more than half the drivers on the road tailgate (follow dangerously close). Probably 2/3. This is actually a major cause of collisions. This is just one of the simpler examples where the majority *does it wrong*.

There are subtler examples: do you know what to do when you come to a blind curve? Less than 5% of drivers get that right.

 

[bdy0627]

@GerberKawasaki
Now that Tesla has an positive eps estimate. Maybe $6-$8 per share over the next year. Every analyst MUST raise their estimates Jonas at MS got his note out first. Titled. “I’m just so wrong about Tesla”

 

[neroden]

The NN doesn't need to learn how to drive, that problem is already solved.

Seriously, it's not solved.

 

[X Fan]

Any view on impact of TSLA announcement to VW & BMW stock? Their down but I'm thinking there is more slide to go....

 

[neroden]

They need 4 quarters that added together have a net profit. That wont happen in Q4, It would need about $1.3b profit to counteract the first two quarters' loss... however, It's possible by 1Q19 IMHO, for 3Q18 (+312m), 4Q18, and 1Q19 to counteract the $747m loss in Q2.

You're right, and further I'd say it's almost certain that's what will happen now. Q1 earnings, which will be announced near the end of April/start of May, so S&P addition should be in May or June. Mark your calendars.

 

[tivoboy]

Come on, you'd be mad as hell at yourself if it goes to $420 tomorrow. I don't think that's going to happen...tomorrow. Not with the market in trouble now. Patience. Nice out at $331.

Frankly. No I wouldn’t be mad as hell at myself. I have a process and it serves me well. Sometimes it means I don’t get the top tick or that I leave money on the table but I know it leads me to make money 85% of the time and only lose 5% relative to the gains the rest of the time

I learned long long ago. Have a thesis. Do your research. Set a plan and stick to it. Now the plan can have some options and permutations but sticking to it has served me well.

I really DO hope it goes to 420$ tomorrow. Or soon. For all the people hoping for that. I’ll be long again sometime soon.

 

[tivoboy]

Any view on impact of TSLA announcement to VW & BMW stock? Their down but I'm thinking there is more slide to go....

Totally irrelevant. They are dealing with their own issues. There is enough market for all at the moment if the other OEMS could just get their s..t together.

 

[neroden]

It really boils down to the consumer. If they begin to push for more EVs the OEMs have no choice but to go along. Buyers rule.

Donn:

If? The buyers already have pushed for more EVs. We knew that when the EV-1 came out.

The legacy OEMs are moving as sluggishly as humanly possible, for the most part. This is why Tesla is taking over. This is why the big auto companies in a decade will be Tesla, BYD, Geely, etc. and not the legacy OEMs.

This is how capitalism works. The OEMs can, in fact, choose not to go along; they just go bankrupt and get replaced by new companies. Most seem to have made this choice. :shrug: Weird on their part, but hey, it's why Tesla's a great investment.

 

[neroden]

Oh there are lots of solutions, but with no financial incentive they don't just happen. You can't make clean concrete or steel sexy like a model 3. You have to make unclean uncool by including the carbon price (tax).

Novacem, Ferrock. They have advantages but they're slightly more expensive. Someone needs to fund their commercialization.

Clean steel's actually happening, actual steel mills are shifting to the clean processes now. Interesting processes. Requires renewably-generated hydrogen, so there's your use for renewable hydrogen...

 

[StapleGun]

My point is, that's not good enough. Not nearly good enough. Look, more than half the drivers on the road tailgate (follow dangerously close). Probably 2/3. This is actually a major cause of collisions. This is just one of the simpler examples where the majority *does it wrong*.

There are subtler examples: do you know what to do when you come to a blind curve? Less than 5% of drivers get that right.

Well tailgating is a solved problem (and a very easy one at that). No NN required. All you need to know is the distance of the car in front, your current speed, and some safety factor.

My point however is that most accidents from tailgating are not caused by people tailgating at whatever the average tailgating distance from a car is. The majority of accidents due to following too close are probably from the worst 10% or 1% of tailgaters. If every driver today started following at exactly the same distance as the median driver does now that would cause a huge reduction in the number of accidents caused by following too closely.

I'm not sure I know exactly what you mean about a blind curve so I won't comment on that.

 

[neroden]

The financial viability rules are:
It should be GAAP profitable the previous quarter
The sum of the previous 4 quarters should be GAAP positive.
S&P could add TSLA now, if they wanted.
https://us.spindices.com/documents/methodologies/methodology-sp-us-indices.pdf
Most web sites do not have the requirements correct. Above link is the official document.

Yes, it's always up to the S&P committee. It's going to be embarassing to them if they don't add Tesla next May/June, though. They might add it earlier, but I think they would not delay adding it longer than that.

 

[neroden]

I'm not sure I know exactly what you mean about a blind curve so I won't comment on that.

Hint, there's a piece of equipment on your car which has been mandated by law for a long time specifically to deal with blind curves.

 

[S3XY]

At the end of last year Uber claimed to have 750k drivers in the U.S. and 2 million globally. With a parttime take rate of 10% of Tesla Model 3 owners, it could take a decade or more to be able to support a service. Tesla drivers are better off just signing up for Uber or Lyft as they do now.

No one is going to use a service where the low numbers of cars make it a toss-up if there is even a car available when they need a ride. Riders are just not motivated to wait. At this time Uber and Lyft do not even have a way for someone to request an EV. That would be an important first step to see just how many riders would be willing to wait an extra 10-15 mins for an EV.

Yeah, who would want to take an $8 ride in a Tesla when they can take that same ride in a Ford Focus?