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

[Troy]

@MarcusMaximus, yes, that's exactly what happened. Here is how my Q2 2019 delivery estimate changed over time:

• 76K on 26 May 2019 here, 80% accurate
• 80K on 9 June 2019 here, 84% accurate
• 85K on 16 June 2019 here, 89% accurate
• 89K on 24 June 2019 here, 94% accurate
• 91K on 30 June 2019 here, 96% accurate

I kept increasing my delivery estimate mostly because of Canada and Europe deliveries. On 30 June 2019, my delivery estimate for Europe was 21,240 units here. Today, we have found out the actual number. It's 22,745 units. I was short by 1505 units (6.6%).

 

[Prime_Number]

I have thought a little more about deliveries in the next few quarters. Here are my current estimates:

Your basis premise seems to be that Tesla is limited by demand. Otherwise why would you not estimate 7,000 per week for 12 weeks - 84,000 for the Model 3 production, instead of your estimate of 69,000 ?

And yes, as someone mentioned above, your early in the quarter estimate was off some 24%.

84 + 19000 (your estimate for S & X) would be 103,000 .........roughly 17 % more than your current estimate.

 

[X Fan]

Is this per gaap or just industry standard ?

GAAP.....unless this is a complicated bonus payout approach. Attached article (albeit a little old) spells out an example in which bonuses could be taken in a subsequent period.

Bonus Deduction Timing: Finding the Correct Tax Year

 

[EVNow]

I believe this is the real situation. So no, this isn't a reason for Q3 deliveries to be lower than Q2; we still have reesrvation holders in Q3.

We know in Norway they haven't delivered any SR+ for eg. And they will have an explosion of orders for SR+ in next two quarters. So, I expect EU deliveries to go up in Q3/Q4.

They do have 14k Model 3s in inventory that can be delivered now - last quarter they had 19k. So, we should expect probably fewer deliveries in July.

But I'm afraid it will be one more quarter they have slog at the end to meet their guidance.

 

[M3Edm]

@MarcusMaximus, yes, that's exactly what happened. Here is how my Q2 2019 delivery estimate changed over time:

• 76K on 26 May 2019 here, 80% accurate
• 80K on 9 June 2019 here, 84% accurate
• 85K on 16 June 2019 here, 89% accurate
• 89K on 24 June 2019 here, 94% accurate
• 91K on 30 June 2019 here, 96% accurate

I kept increasing my delivery estimate mostly because of Canada and Europe deliveries. On 30 June 2019, my delivery estimate for Europe was 21,240 units here. Today, we have found out the actual number. It's 22,745 units. I was 1505 units (6.6%) short.

Based on this, I’ll wait for your Q3 estimate a week before quarter end.

 

[neroden]

Your basis premise seems to be that Tesla is limited by demand. Otherwise why would you not estimate 7,000 per week for 12 weeks - 84,000 for the Model 3 production, instead of your estimate of 69,000 ?

So, in the spirit of production based estimates... we've heard that Tesla hit 1000 cars a day "some days" over late June, and the the Gigafactory is reliably producing enough packs for more than 1000 cars a day. But Tesla is not up to 1000 cars a day reliably at Fremont. Q2 averaged only 797 cars/day.

So. My range of production estimates runs from an average of 800/day if they just can't get Fremont sped up yet (72800) up to the aforementioned 84000 if they immediately got to 1000/week consistently last week (very unlikely). Perhaps the best guess would be somewhere in between -- around 80,000 perhaps, as they ramp up from a reliable 800 to a reliable 1000 by quarter's end.

I'll go with that as a prediction.

I am not willing to predict deliveries because it depends on choices made by Tesla regarding "wave unwinding". It's manipulatable, so not predictable. I also don't care.

 

[Krugerrand]

The bonus would be paid in Q3. Can't be an expense from before it happened.

(Unless they earmarked funds for the bonus in Q2, but that would be outside the P&L, I think)

How are the bonuses going to be paid? Cash? Stock? Time off?

 

[EVNow]

Based on this, I’ll wait for your Q3 estimate a week before quarter end.

I was going to post something similar too

While one should update forecasts as more information is known - it would be useful to keep track of the earlier forecast and assign accuracy rates to them too.

For eg., an election poll done today for 2020 Nov will have much lower forecast accuracy than the one done in 1st week of November.

 

[KarenRei]

And yes, as someone mentioned above, your early in the quarter estimate was off some 24%.

Which is not simply a little wrong, it's massively wrong. Halfway through the quarter and he thought Tesla's Q2 revenue was going to be something like 1 1/2 billion dollars less than it actually will be. Billion, with a B.

Rather than looking at the fundamental cause of why he got it so wrong (falsely believing in a "demand problem"), he's instead focusing on the fact that he highly belatedly started merging up toward the actual answer when numbers came out that greatly contradicted his expectations for the quarter. And yet is now back to trotting out the same "demand problem" line that caused him to get it so wrong in Q2.

Troy, I'm not saying this to be mean. I'm saying this in the hopes that you engage in some introspection.

 

[neroden]

Sooooo, I'm thinking about that leaked memo. How fast IS Shanghai going to come up? It is basically a copy-and-paste job (none of this "learning how to make a car production line" stuff which Fremont repeatedly dealt with). I could see the first cars coming out in early October, with volume production in January.

 

[Stretch2727]

I have thought a little more about deliveries in the next few quarters. Here are my current estimates:

Model Y is late 2020 not late 2019.

 

[Troy]

It's the last estimate before the end of the quarter that counts. I run the survey here where people submit their own estimates and everybody is allowed to change their estimates as many times as they want by the end of the quarter, Pacific Time. It has always worked like that.

In addition, changing your estimate doesn't mean you will improve it. For example, my Q4 2018 estimates here on 27 Nov 2018 (34 days in advance) were eerily accurate. However, I messed it up on 31 Dec 2018 and the messed up version was the one that counted. I was off by 12% in Q4 2018.

Screenshot source: Troy Teslike on Twitter

 

[EinSV]

@KarenRei summarized it very well, but I think for Troy and other pessimists/bears, the song is the same for Q3 as it was in Q2.

Expect estimates to be adjusted upward toward the end of the quarter just as they were in Q2, after much hand-wringing over supposedly falling demand when all signs are that Tesla is production constrained:

A quick clarification: Troy has now updated his estimate to 89,619 Troy Teslike on Twitter

He raised his forecast from 74K in early to mid-May, to 76K (5/23) to 80K (6/9) to 85K (6/16) to 89K on 6/23.

Side note: Troy doesn't seem to understand that Tesla is production constrained so is repeating the same mistake for Q3 as he did earlier in Q2. He initially suggested that Q3 deliveries might drop by as much as 15-20K from Q2, and is now settling on a 78K forecast for Q3 (down 11K from his Q2 forecast). Troy Teslike on Twitter

Watch for that forecast (and similar pessimistic forecasts from analysts and TSLAQ) to follow a similar pattern to Q2, increasing toward the end of Q3. Yes there will likely be a short-term dropoff in US orders with the tax cut phase down, but there are a large number RHD Model 3 in the pipeline (>5K orders already in the UK alone and only ~100-200 delivered in Q2), SR+ has barely been delivered outside the US, Tesla has the option of expanding the limited configuration options now being offered, Raven S/X will be produced for the full quarter, etc.

By the way, for folks concerned about order backlog, there were over 6000 Model 3 orders in the queue just in the UK on July 1, according to an owner spreadsheet. Tesla Model 3 UK Orders (Responses)

Add in ongoing orders plus the other RHD markets and RHD deliveries could easily exceed 10K in Q3 if the logistics work out.

 

[KarenRei]

It's the last estimate before the end of the quarter that counts

Teacher: "And Troy, what's your answer for problem 2?"
Troy: "74k"
Teacher: "Higher."
Troy: "Sorry, 76k".
Teacher: "The answer is higher than that, Troy."
Troy: "80k."
Teacher: "(sigh) Keep going...."
Troy: "85k."
Teacher: ".... and going...."
Troy: "89k."
Teacher: "Troy, time is running out, and you're still too low."
Troy: "91k."
Teacher: "The answer was 95k, Troy."
Troy: "Wow, I was only 4k off!"

----

You get no points for correcting a horribly wrong number once the data comes out that shows you were wrong. We can all look at teslastats.no, eu-evs.com, insideevs, etc, and make post-facto predictions by punching in actual numbers. Post-facto predictions are worthless. Before you had the actual data for the quarter, you thought they'd deliver 74k. Now you're making another forecast for before anyone has the actual data for the quarter. And you're basing it on the exact same assumptions that caused you to get it so wrong last time. Yet for some reason, you want people to trust you on it, when you won't even own up about the fact that said expectations gave you a horribly wrong forecast last time.

 

[tinm]

Hurt me beacuse my margin requirement was raised to 75% and so I had to cut back slightly for safety. Still, if my merger arb closes while TSLA is still cheap...

I just went and looked and sure enough, the Schwab site says "special margin maintenance for TSLA" is 75%. They sure never told me. Did you get an email or any notice from them or did they just up and change the percentage without notice?

 

[Thekiwi]

I have thought a little more about deliveries in the next few quarters. Here are my current estimates:

Some people will probably say 69K Model 3 deliveries in Q3 looks low. Why would it be less than Q2 2019? I can think of two reasons:

• In Europe, there are no more reservation holders left. In Q2 2019, 18,279 Model 3s were sold in Europe and some of them were purchased by reservation holders. Let's say if 7K out of 18K were reservation holders, the demand in Europe going forward will be 11K, not 18K because you can't count reservation holders in Q3 anymore because there are no more reservation holders left. Tesla cleared the reservation queue in North America in Q4 2018 and in Europe in Q2 2019. Of course, demand keeps increasing. If sustainable demand was 11K in Europe, it might increase to 14K but I think it's still likely to be less than Q2. However, deliveries to the UK will make a difference. Therefore I will keep updating my Q3 delivery estimate. I will make another calculation around 10 Aug 2019 when we have the Eu numbers for July.
• In the US, some buyers who didn't need a new car yet still bought it in Q2 because they didn't want to lose out on federal tax credits which dropped by $1875 for deliveries after 30 June 2019.

By the way, here is my accuracy in the last two quarters. Q1 was a difficult quarter. Model 3 deliveries were 10K lower than most people expected and there was a big drop in S/X deliveries. You can see the accuracy of all my estimates since Q2 2018 here.

Others have already made the same points I would make about the above.

But I really struggle to see why you would make a prediction so low when Tesla itself said in its Production and Delivery report that they expect an increase in production and deliveries in Q3.

“We believe we are well positioned to continue growing total production and deliveries in Q3.”

 

[KarenRei]

And now for something completely different...

Neuralink on Twitter

 

[Troy]

Model Y is late 2020 not late 2019.

I expect them to move it sooner because the new Model 3 production line Tesla is developing for the Shanghai factory and the Model Y production line they will build at the Fremont factory is identical. The same production line will produce both Model 3 and Model Y. Eventually, all production lines at the Shanghai factory and all Model 3/Y production lines at Fremont will be the same.

It makes sense for them to build one of these lines at Fremont in Q4 2018 so they can fine tune and sort out problems before building multiple versions of it at Shanghai. They also need to switch all existing Model 3 lines at Fremont to this new 3/Y line that is more automated. Model Y production at Fremont in Q4 2018 is an old prediction of mine since late 2018. At the time, Fremont looked unlikely too. See the reaction here.

 

[2virgule5]

Sooooo, I'm thinking about that leaked memo. How fast IS Shanghai going to come up? It is basically a copy-and-paste job (none of this "learning how to make a car production line" stuff which Fremont repeatedly dealt with). I could see the first cars coming out in early October, with volume production in January.

I’ll call you ‘optimist’ on this one
For one, copy & paste seems too simple for Tesla ethos, I’m betting on Elon & co wanting to make improvements and experiments there and there... More importantly it is still not clear to me where these 2170 cells are going to come from. At 50Kwh per car it is a lot a GWh that are needed, with the right chemistry, and fast.

 

[Antares Nebula]

@Troy

Thanks for the estimates. However, I think the tax credit effects are minimal at this point, and sales are being driven by media and word of mouth, as many new customers discover and realize how great the Model 3 is (though anecdotal, I've seen this first hand). That is, Tesla is experiencing geometric growth now.

Thus, sales are generally going to increase every quarter going forward, save for seasonal (e.g., Q1 dip) or macro changes (e.g., recession). And in fact, Q3 and Q4 are seasonally strong for car sales. In short, I think it's fairly simple and not in need of any detailed analysis.

Adding it all up: stronger upcoming sales seasons, geometric demand growth, and a 95k Q2 print, one can reasonably estimate that Q3 will be 95 - 105k, and Q4 105 - 115k. One can be somewhat conservative or aggressive by subtracting or adding, respectively, 5k to these numbers. JM2C