It was a very arbitrary choice, and one which I see after review actually weakened my thesis.
I had tried several possible combos of how to measure short-term stock price movement when I did this analysis back in August and found that it doesn't really affect the overall average results much. This is due to the fact that differences between the open-to-close % change and the open-to-next-day-open % change average out to approximately zero, and my overall sample size for stock sales is 20, which is reasonably large. I presented tables and graphs for only one of them for brevity's sake.
Here is the scatter plot again with the requested adjustment to use the difference between opening price on days Elon has sold and the opening price the next day.
View attachment 870256
As we can see, there is clearly no negative effect and if anything the TSLA price tends to be slightly
higher on the morning after Elon's selling days, which happened in 13 out of 20 instances (65%).
The average increase after Elon's sales measured this way was +1.3% after adjusting for TSLA's beta and what the S&P 500 did that day, compared to a baseline average of +0.1%.
In fact, that adjustment appears to make the case stronger that even on the days Elon sold more than 5% of the float there was zero measurable impact by opening bell the following day.
After looking at all this data I think we can conclude that the only reason to care about exactly when Elon sells stock and exactly how much he's selling is if:
- You're day trading TSLA
- It happens to be one of the rare days in which Elon is going to sell 5% or more of the volume
- You somehow know this in advance
- You have no ethical or legal qualms about insider trading
Therefore, I wish this issue would stop garnering so much unwarranted attention to make room for information and discussion that's actually useful.