### Modelling Q4 Model 3 Deliveries

***** NOTE: 12/25 *****

One more mistake to correct. I estimated 507 Dec deliveries in TMC, based on the number of undelivered configurations. However, it appears I drastically underestimated how many of those were "ghost" entries that were abandoned before delivery. It now appears that Dec deliveries in TMC will come in around 330 at most. That makes total TMC deliveries 934. NY sample rate was 2.38%, but assuming that's 20% higher than average, we get 1.9% sample rate. That implies

So my first estimate was high on Dec deliveries, but low on sample rate. The final result is about 2K higher.

***** NOTE: 12/16 1PM PST*****

After publishing this, I did some more analysis that makes me question my conclusions here.

1) It's become more clear that 20% of Sept VINS are simply missing from NMVTIS. This has implications both for the NMVTIS-based predictions and for the NY registrations (if we assume 20% of NY regs are simply missing as well)

2)…

One more mistake to correct. I estimated 507 Dec deliveries in TMC, based on the number of undelivered configurations. However, it appears I drastically underestimated how many of those were "ghost" entries that were abandoned before delivery. It now appears that Dec deliveries in TMC will come in around 330 at most. That makes total TMC deliveries 934. NY sample rate was 2.38%, but assuming that's 20% higher than average, we get 1.9% sample rate. That implies

**~50K deliveries in Q4**.So my first estimate was high on Dec deliveries, but low on sample rate. The final result is about 2K higher.

***** NOTE: 12/16 1PM PST*****

After publishing this, I did some more analysis that makes me question my conclusions here.

1) It's become more clear that 20% of Sept VINS are simply missing from NMVTIS. This has implications both for the NMVTIS-based predictions and for the NY registrations (if we assume 20% of NY regs are simply missing as well)

2)…