Building a crypto strategy from A to Z

The thesis is the reason you think an asset will be worth more tomorrow. It is written in two or three sentences and the facts must be able to contradict it. “Bitcoin will go up” cannot be contradicted. “Bitcoin is establishing itself as a store of value among institutions, which ETF assets under management measure” can be.

Set the horizon too. A five-year thesis can withstand a 50% decline along the way; a six-month thesis cannot. The horizon decides almost everything else: how often you buy, how wide the stops are, how long the thesis is given to prove itself.

The allocation splits capital between assets: 50% bitcoin, 30% ether, 20% stablecoins, for example. It determines most of the portfolio’s risk.

Two safeguards complete it: a cap per asset, so that no position can take the portfolio with it, and an estimate of the decline you accept to go through. Bitcoin has already lost more than 70% from a high, several times. An allocation is built on the assumption that this will happen again.

Adding more holdings diversifies little: most cryptos follow bitcoin, especially on the way down.

The size of a position follows from the loss you accept. If you limit the loss to 1% of capital per idea and the stop is 20% from the entry price, the position is worth at most 5% of capital.

This rule keeps the portfolio alive through a run of losses. At 1% per position, ten mistakes in a row cost less than 10% of capital. At 10% per position, they cost about 65% of it.

Leverage increases the size without changing the capital and brings the liquidation price closer: with leverage of 5, a decline of about 20% is enough to lose the whole stake.

There are two families of rules. Scheduled buying, or DCA, invests a fixed amount at regular intervals and removes the question of timing. Conditional buying waits for a defined signal: a 20% pullback from the high, a return above a moving average.

The two can be combined, provided you write the full rule: the amount, the frequency or the signal, and what stops the purchases. A DCA with no stop condition keeps buying an asset whose thesis has fallen.

Three exits need to be planned: on the way up, through take-profit levels; on the way down, through a stop; and on the facts, through an invalidation condition that describes what would prove the thesis wrong.

Exits are written with numbers: which price, which share of the position. A rule such as “sell when it has gone up a lot” leaves the whole decision to the moment when it is hardest to make.

Before committing capital, a backtest applies the rules to the price history. It shows the maximum loss endured, the time spent below the capital invested and how sensitive the result is to the start date. Replaying the crises separately, March 2020 or November 2022, shows what the strategy would have done at the worst moment.

A backtest describes the past and guarantees nothing. Its main use is to rule out the rules you could not have borne to follow.

Then comes tracking. Regularly compare what you hold with what the strategy plans, and note each deviation. Review the strategy on a fixed date, with a cool head. In the middle of a decline, apply it.