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

[CorneliusXX]

'No one' is two words. In other off-topic news, no one cares about filling time for any vehicle on the investors roundtable thread.

Sparky 4 Life is approx 2.5 words.

 

[MP3Mike]

I don't think so. Did you see this hacked Tesla ? It would have hit the divider when turning and had to be disengaged. And is not even SanFrancisco - but Bay area

May be they have better version on HW3 hidden somewhere, if so, I'd like to see it and change my stance.

That may be a hacked version of the production build with some extra features turned on, but it isn't one of their development builds.

 

[CorneliusXX]

I get the point that people are trying to make... and I don't disagree that calculating time needed to gain range is useful.

The point I'm making is that it's not the ideal metric to compare "charging capability" (which is where this discussion descended from: "The S needs a refresh to be as advanced as the 3").

As an example: Aero wheels can make 10% difference in energy usage efficiency. Does a Model 3 now charge 10% faster when the owner installs his aero caps?

Answer is "No", despite his drivable miles-per-minute charge rate has increased. Same as if one driver sticks to the speed limit and the other drives 15MPH over and uses more energy.

So while the "range gained" metric is helpful, when taken in a vacuum doesn't really compare vehicle charging capability well.

Ultimately, as higher-power charging continues to take hold, much of this will be moot..the differences between charge sessions will become a smaller fraction of the overall time. If one power/efficiency combo is only 2-3 minutes difference out of a 20 minute session, who really cares?

Fair call.

I do think that over time the "range gained" will become the default metric for EVs rather than charging speed as it provides a better shorthand than just charging speed.

 

[Sparky4life]

I don't think so. Did you see this hacked Tesla ? It would have hit the divider when turning and had to be disengaged. And is not even SanFrancisco - but Bay area

May be they have better version on HW3 hidden somewhere, if so, I'd like to see it and change my stance.

ps : BTW, I've heard this before - that waymo/cruise are always in "dev" mode. That is not true the right way to look at it. They are only interested in robotaxi scenario. So, as such we won't see their software in retail cars - even when they have solved the FSD problem. Add to this fact that Tesla says their AP/FSD software is Beta i.e. it is also in "dev".

This is not dev mode. I'm talking about what Elon calls dev mode; HW3 with an expanded NN like we saw in the Autonomy day video. It's the ability to turn on features only available to developers, like interacting with 4-way intersections. The point is that comparing what is available to end-users (even in 'hacked' mode) in Teslas to what's always turned on in Waymo/Cruise does not tell the whole story.

 

[InTheShadows]

Disclosure: Long Tesla since 2016 so my only interest is in stock appreciation.

You make good points average daily volume is a misleading indicator of short covering capacity. From reports I have seen HFT makes up more like 50-60% of the market volume in 2018. The accumulation / distribution power is driven by true sellers or net new buyers of a stock and I agree is the primary factor of price movement from day to day. 80% correlated to HTF seems high although this I would imagine is very stock and time depended. HFT drives down the spread of bid/ask and provides liquidity to both buyers and sellers. Maybe not net liquidity for the day but intraday they contribute and act like market makers.

There are so many dynamics in stock pricing I don't think it's fair to characteristic price movements in June and October with short squeezes although short covering no doubt plays a role. VW case was a clear short squeeze and the type of mythical event that most people think of in a short squeeze. The stock going up 30%, 40 or even 50% doesn't seem like a short squeeze to me and can be traced to new new buyers, reduction of sellers just as easily. With 46m shares short and a float of about 120m. This stock should skyrocket in the conditions of a short squeeze were met where the stock doubles, triples in months. Instead the short balance over the last several years just ebs and flows between 25m and 45m more or less in line with general price movement of the stock.

I would characterize previous short related price increases in TSLA as gentle embrace rather then a short squeeze. I guess all I'm saying those hoping for a 100%, 200 or 300% price increase in relation to the epic "short squeeze" are likely going to be disappointed.

I agree completely. This feels like 2013 all over again when everyone thought a VW squeeze was coming. There was a nice multi week gain (perhaps months) that was probably in part due to shorts covering their positions but it wasn’t the squeeze everyone thought was going to happen. Because the “squeeze” never happened many people were disappointed because they didn’t get the gains they were hoping for, particularly those playing with options.

Even though the squeeze didn’t happen I feel it was still one of the best times to be investing in TSLA. Just like now is one of the best times to be investing in TSLA. The strength of the company then and now are like night and day. That we got another opportunity to buy at sub $200 is mind baffling to me.

 

[EVNow]

That may be a hacked version of the production build with some extra features turned on, but it isn't one of their development builds.

This has been an ongoing debate for sometime. Note that this hacked software shows ALL the features we saw on autonomy demo day. It is exceedingly unlikely they would have put some of the features into production but not others i.e. they put a feature that lets the car turn - but withheld the feature that turns it so it doesn't hit the divider

I think they have been regularly merging from their dev branch to production branch. It is likely they have only one dev branch - and they merge it with production branch to push out features that are mature (while turning off ones that are not, like this City NOA). So, their dev branch could be a few weeks ahead of the production branch (2019.20.x).

 

[LN1_Casey]

Welp. I'm both happy and unhappy about today's stocks. As I am traveling, I couldn't watch it to buy as it developed, but I had placed an order to buy TSLA while I was out. I put in a buy price that was, at the opening, too low and it only got worse (better?) from there.

On one hand, I didn't get my shiny new stocks to add to my slowly growing hoard.

On the other, TSLA went up $10 today, nicely rounding my stash almost back to the black.

 

[mongo]

This has been an ongoing debate for sometime. Note that this hacked software shows ALL the features we saw on autonomy demo day. It is exceedingly unlikely they would have put some of the features into production but not others i.e. they put a feature that lets the car turn - but withheld the feature that turns it so it doesn't hit the divider

I think they have been regularly merging from their dev branch to production branch. It is likely they have only one dev branch - and they merge it with production branch to push out features that are mature (while turning off ones that are not, like this City NOA). So, their dev branch could be a few weeks ahead of the production branch (2019.20.x).

Is it possible to merge two NN BLOBs? It seems like it is a one or the other choice for each distinct NN.

 

[ggr]

Mod: OK, guys, the charging speed/range gain/kW discussion is officially over, for this thread anyway. --ggr

 

[MP3Mike]

I think they have been regularly merging from their dev branch to production branch. It is likely they have only one dev branch - and they merge it with production branch to push out features that are mature (while turning off ones that are not, like this City NOA). So, their dev branch could be a few weeks ahead of the production branch (2019.20.x).

Elon specifically said that he had two different dev builds on his two cars. One was a FSD branch and the other was an Enhanced Summon branch. So no, they have not put all of their development code into production, and no they don't have a single dev branch.

 

[EVNow]

This is not dev mode. I'm talking about what Elon calls dev mode; HW3 with an expanded NN like we saw in the Autonomy day video. It's the ability to turn on features only available to developers, like interacting with 4-way intersections. The point is that comparing what is available to end-users (even in 'hacked' mode) in Teslas to what's always turned on in Waymo/Cruise does not tell the whole story.

I actually thought that the fact that these features are available to anyone (who wants to hack) is actually very bullish ! That means, the autonomy day wasn't a "hacked" version that worked just on those roads or route.

We'll not obviously know the "whole story" since all companies hold a lot of information closely. But, to me the argument that waymo/cruise are further along than Tesla when it comes to city NOA at this point is fairly compelling. If Tesla could do those things with the quality we see in those companies, there was little reason to not have released City NOA already (i.e. Tesla would have been "feature complete").

 

[tander]

...May be they have better version on HW3 hidden somewhere, if so, I'd like to see it and change my stance.

Isn't that whole strategy with autopilot, let it learn in ghost mode and then release it/update computer when it's ready for primetime?

 

[EVNow]

Elon specifically said that he had two different dev builds on his two cars. One was a FSD branch and the other was an Enhanced Summon branch. So no, they have not put all of their development code into production, and no they don't have a single dev branch.

Link ?

I'm going by this.

green on Twitter

 

[MarcusMaximus]

I don't think so. Did you see this hacked Tesla ? It would have hit the divider when turning and had to be disengaged. And is not even SanFrancisco - but Bay area

May be they have better version on HW3 hidden somewhere, if so, I'd like to see it and change my stance.

ps : BTW, I've heard this before - that waymo/cruise are always in "dev" mode. That is not true the right way to look at it. They are only interested in robotaxi scenario. So, as such we won't see their software in retail cars - even when they have solved the FSD problem. Add to this fact that Tesla says their AP/FSD software is Beta i.e. it is also in "dev".

That’s still the production software, just hacked to follow nav on surface streets. The point of comparison would be the actual FSD networks that aren’t in any production firmware(the kind Musk *might* be using, but is definitely in use by Tesla’s internal FSD testing team).

 

[Silentk]

Spotted on Twitter...

https://www.sec.gov/Archives/edgar/data/1446440/000117266119001322/0001172661-19-001322-index.htm

Value: 420

 

[EVNow]

Isn't that whole strategy with autopilot, let it learn in ghost mode and then release it/update computer when it's ready for primetime?

Right - in which case - the version in production has to be fairly new. I don't know how or how much they are using shadow mode, but to me this is the strongest argument for having the latest FSD build available in production but with features turned off.

Another explanation for having all the FSD dev code fleetwide but disabled is to collect data in the "shadow mode". While shadow mode potential is huge, not sure what they actually do.

​

 

[Cherry Wine]

Spotted on Twitter...

View attachment 420368 https://www.sec.gov/Archives/edgar/data/1446440/000117266119001322/0001172661-19-001322-index.htm

That value column. Lol!

 

[urarthur]

So this is interesting. Chanos' Kynikos fund is long $420k Tesla call options according to the latest 13F SEC filing.

Short positions are not shown (not required by SEC)

SEC FORM 13-F Information Table

Edit: my bad, didn't notice it was posted earlier.

 

[EVNow]

That’s still the production software, just hacked to follow nav on surface streets. The point of comparison would be the actual FSD networks that aren’t in any production firmware(the kind Musk *might* be using, but is definitely in use by Tesla’s internal FSD testing team).

Not really - it actually does a lot of things - like recognize stop signs/traffic lights, turn etc. Why would all those features be in production ?

 

[EVNow]

So this is interesting. Chanos' Kynikos fund is long $420k Tesla call options according to the latest 13F SEC filing.

Short positions are not shown (not required by SEC)

SEC FORM 13-F Information Table

They really need to change that ...