Why does it matter?

Bitcoin is Bitcoin, there is no other asset like it. Provably scarce, Bitcoin is in a league of its own with its characteristics around durability, portability, divisibility, fungibility, scarcity, and acceptability. It is perceived as the safest coin in crypto and most long-term-oriented crypto investors would argue that a portfolio is not complete without Bitcoin.

Historical performance

Here’s an overview of the past price ranges in which our Bitcoin Cycle Top Indicator would have bought and sold Bitcoin. Our model starts with the assumption that on the 1st of January 2013 we became active in the crypto market and held Bitcoin.

For a rough idea on the Bitcoin Cycle Top Indicator performance compared to the benchmark Bitcoin HODL strategy, take a look at the last column. Counting from the 1st of January 2013 until now, Ace BTC outperformed a simple Bitcoin HODL strategy by 35.3x!

What we measure and how

We’ve tested 50+ on-chain metrics to determine which ones make most sense to form the foundation of Ace BTC. The current model looks at 14 on-chain Bitcoin metrics and their ability to ‘predict’ past cycle bottoms and tops. Those metrics currently include:

• Pi-cycle

• Dormancy Flow

• Net Unrealized Profit/Loss (NUPL)

• Reserve Risk

• Realized HODL

• Liveness Ratio

• Price temperature

• Market Value to Realized Value

• Puell multiple

• Relative strength index (RSI)

• Exponential Moving Average (EMA)

• Stablecoin Supply Ratio (SSR) Oscillator

Note that we regularly evaluate and optimize our model, so these metrics might change over time.

Chainalytics Bitcoin Buy Score & Sell Score

We use extensive testing, weighing, and metric normalization to get to two easy to read scores on which we base our buy and sell decisions:

Some on-chain metrics are better indicators for a top or a bottom than others. In the two models, each metric has a specific weight depending on its predictive power; this allows the score to be more accurate. For example, if metric A has a high predictive power for tops but low for bottoms, it will respectively have a relatively high weight compared to other metrics in the sell model, and a relatively low weight in the buy model.

Taking action based on the BiTCoin cycle top indicator

As the model is designed for long-term investors (not for day-trading), several assumptions lay at the base:

A BUY or SELL signal is triggered when the respective sell or buy score has surpassed the threshold and stayed within range (red or green zone) for a minimum of 7 consecutive days.

At that moment the first sell or buy signal is triggered. That signal is the first of a series of 4 actions recommended. Its aim is not to time the exact top or bottom, but rather take action around the top and bottom. The idea is that by doing this, the average buy/sell price is less sensitive to the short-term volatility we usually see around the ultimate tops and bottoms. Nobody is able to time the top or bottom, and there is little need for long-term investors to perfectly time their actions.

• Actions to take in case of an actual SELL signal:

  • On the day of the SELL signal: sell 25% worth

  • After 4 days, sell 33.3% of what's left.

  • After 8 days, sell 50% of what's left.

  • After 12 days, sell 100% of what is left.

• Actions to take in case of an actual BUY signal:

  • On the day of the BUY signal: buy 25% worth

  • After 40 days, buy 33.3% of remaining.

  • After 80 days, buy 50% of remaining.

  • After 120 days, buy 100% of remaining.

Notice how the action intervals for selling and buying are different. This has to do with the fact that Bitcoin cycle tops tend to be short, and Bitcoin cycle bottoms tend to be relatively long. In other words, you have much more time to buy relatively close to the bottom, but you have little time to sell around the top.

Removing complexity from decision making

This Bitcoin Cycle Top Indicator provides comfort and peace of mind. It takes emotion out of decision making. We let the inputs be complex, and the output and actions be simple.

Models are never perfect, therefore we continuously aim to improve our model by evaluating and calibrating the model when necessary.