Thread by @DeFi_initiate

The Dynamic Power Cycle (DPC) model of #Bitcoin uses the power law as its base, and adds periodic behaviour resulting from the coin's 4-year block subsidy (miner reward) halving cycle. The implementation of the model now reflects my original vision (conceived in Oct 2021). 🧵 17

1. The red line on the chart above shows how the median price of bitcoin is expected to unfold over the next 3 years. Today's closing price is projected between $30K and $73K (median: $47K). Time uncertainty is ±34 days, and price tolerance is -36%/+56% (at 95% confidence).

2. The cycle trough is expected between Dec 2, 2022, and Feb 8, 2023 (midpoint: Jan 5), and the price at that time is projected at $11-27K (median: $17K). The next halving is expected to occur on Apr 24, 2024, and the price at that time is projected at $29-70K (median: $45K).

3. The current cycle peak was projected for Nov 9, 2021, at $79K ($51-124K). Actual peak was on Nov 8 at $68K. Timing was super accurate but price was off by 18%. Four years ago, the same projection was for Sep 29, 2021, at $90K: timing was off by 40 days (3%) and price by 33%.

4. The next cycle peak is expected between Aug 30 and Nov 6, 2025, with a projected price range of $230-560K (median: $360K). The chart below shows the entire history and includes this peak. The light blue line at the bottom (DPC Multiple) shows projection accuracy (right axis).

5. The DPC model is far from perfect but it's better than any other long-term model I'm aware of. I compared the candidates currently in the running three weeks ago:

https://twitter.com/DeFi_initiate/status/1479897909890105346

6. How well will it hold up? This is difficult to answer, but we can get a feel for it by employing a time machine. The chart below shows an overlay of the model today and 4 years ago. The solid red line was the projection on Jan 1, 2018, and the dotted red is the current one.

7. While price action is very different form this past projection, the model did get the general trends fairly accurately. The cycle trough was projected for Oct 6, 2018, at a price of $1.8K. Actual trough was on Dec 15 at $3.2K. Timing was off by 70 days (20%) and price by 78%.

8. Keep in mind that this was at the height of euphoria and yet the model showed a sudden and significant drop. Did anyone expect this at the time?

Recall that the model doesn't have information from the future and is not optimized on backtested data. What's shown is real.

9. The next halving was expected on Apr 26, 2020, at a price of $8.2K. Halving took place on May 11 (15 days or 2% later) at a price of $8.6K (5% higher). This prediction was spot on.

10. For the year 2022, the projection 4 years ago showed a steeper decline into the cycle trough than what the current (dotted) curve indicates. This is because the model has since took into consideration the slower and gentler 2018 decline, and adjusted itself automatically.

11. The model consists of the following 4 components:
- Power law for long-term growth
- Periodic behaviour due to block subsidy halving every ~4 years
- Adjustment for daily price action
- Adjustment for diminishing effect of halving cycle

12. The power law in this context is a growth pattern for the long term using logarithmic regression. For #bitcoin, it was first proposed by "Trololo" in Oct 2014: bitcointalk.org/index.php?topi…. The regression parameters were very different at the time due to limited data.

13. A precursor to this was a modified logistic regression model by @gsantostasi in Jul 2014: reddit.com/r/Bitcoin/comm…. This included an adjustment for the periodic bubbles observed to date. This as an important contribution that could be revived using current data.

14. @hcburger1 published a full treatment of the power law for #bitcoin in Sep 2019: medium.com/quantodian-pub…. A live chart of this model is here: stats.buybitcoinworldwide.com/long-term-powe…. The Bitcoin Power Law Corridor by capriole_charles is the model's implementation as a TradingView indicator.

15. The various power law models were reviewed in Sep 2019 by InTheLoop here: medium.com/quantodian-pub…. All these models are static, meaning that they are based on a snapshot at the time of creation. To keep the models alive, they would need to be updated from time to time.

16. DPC adds to the long-term growth arc the periodic behaviour driven by the halving cycle. It also adjusts itself dynamically with more data. Due to these characteristics, it has a better chance of survival - projections will be valid as long as the principal assumptions hold.

17. While the power law is currently the best model for overall long-term behaviour, this can change in the future. Will we have to adopt logistic regression for implementing the S-curve at some point? Time will tell. ⏹️

@hcburger1 @caprioleio Did you create the Bitcoin Power Law Corridor TradingView indicator?

@InTheLoopBTC Did you write the "Power-law price models of bitcoin" Medium article?