Ihor reported TSLA shares shorted as 39m yesterday but today had to revise to close to 46m... based on yesterday's short figures released. That's a 7m error. I'm not sure if we can really trust his TSLA shares short figures anymore. 7 million share error is really big and for him to claim he's an expert using "algos" to calculate this, I'm not so sure.
I think Ihor is just guessing, I probably can guess better.
Ihor's been wrong multiple times when the official numbers come out. In fact, he's usually wrong on a monthly basis....and he's always wrong on the low end. When the official numbers come out, he's always underestimating the number of shares shorted and then he immediately puts out a tweet updating his numbers.
Either he's just clueless or he's purposely downplaying short activity to their benefit. His comments on naked shorting told me all I needed to know about how his perception of wall st shenanigans compares to real life.
Or the shorts are good at hiding their activity from the data that he has access to, and he hasn't figured out how to account for that yet...
Here's pure speculation on what he's doing, which may or may not be entirely wrong:
He likely has a model that can, for most stocks, estimate short interest based on trading patterns, and whatever data he can see.
He then extrapolates that model to TSLA (and other popular short targets), which appears to have unusual shorting activity (either stuff he can't see, or unusual patterns throwing off his model), not fitting the model (which I'm guessing is an educated guess), and he has to revise with what actually happened. However, I'd also guess that the model isn't being tuned specifically for TSLA, and therefore isn't getting any better.
If I wanted to stretch my guesses further, if it's not an insufficient data but rather a pure unusual pattern situation, I wonder if his models are underweighting the possibility of shorts selling to other shorts - that may show up in a model as covering and underestimate the count of shorted shares somehow, if he only sees part of the transaction?
The S/X rear AC motors do not spin freely; the controller is set to "torque sleep". This zero net thrust mode has lower losses than the drag of an unpowered axle pulled by a powered axle. In torque sleep, most work is done by the more efficient front motor.
The AC motor, when active, continually wastes power in the rotor. The PMSR is much higher on the efficiency curve in cruise than the AC at near idle is. So it is more efficient to drive the gearing and rotor with the front motor than to drive the rear motor electrically.
See also
@wk057's investigation showing the motors turn off:
Let the hacking begin... (Model S parts on the bench)
Yeah, I would be
very surprised by torque sleep being a powered coasting mode, especially given the descriptions of behavior coming out of torque sleep. On an AC induction motor, you'd just cut the field current, and the motor will coast with only bearing/gear drag.
(There
is an electric bicycle vendor, Grin Technologies,
playing with powered coasting (with the addition of regenning while powering the same motor). However, they're using a surface permanent magnet motor, which has extremely high cogging torque due to the magnet configuration. It'll be interesting to see how well that works in practice, but it's essentially irrelevant to anything Tesla is doing.)
The Roadster2 likely won't have a AC motor in the front. It achieves both a 1.9 sec 0-60 performance and a 250+ mph top speed. AC motors don't allow this range of performance with a single speed drive.
Further, we saw the Roadster do 0-70mph 'wind-sprints' for an hour at the reveal event. An AC motor would melt doing this demonstration. The simpliest explanation for all these capabilities is that there are three Model 3 SRPM motors in the Roadster2.
Although part of the advantage of the AC induction machine is that at higher speeds, it can freewheel, as the PMSRMs get into their stride at higher speeds (and traction becomes less of an issue, so AWD is less necessary). Cooling is an issue, but...
Rimac uses four permanent magnet motors, but they need a 2-speed gearbox on the rears...
Three motors is still clearly an interim step. As Rimac knew, there will eventually be one motor per wheel. It makes the suspension more complicated, and at the moment that means more expensive. I wonder when Tesla will develop a four-motor design with low unsprung weight which they consider cheap enough to mass produce -- I'm sure every engineer working on the drivetrain wants to do that.
I do wonder how long before someone clones Honda's Sport Hybrid SH-AWD system, for things like that, if it's decided that a 2-speed is needed:
Sport Hybrid SH-AWD System Animation
The shared shifting mechanism seems like a rather clever cost/weight optimization to get a 2-speed reduction box for very high speed operation. Rimac uses two independent 2-speed reduction boxes (one per rear wheel), for comparison, although they advertise the ability for each to be shifted independently based on demand.
