Coinrule: When a Rule Does Exactly What You Wrote

A rule is a sentence the exchange will obey

As we understand it, Coinrule is a builder where you put a trading rule together from a condition and an action: when this is true, do that. The service then carries it out on your exchange account through API keys. Builders of this style are meant to be read back like a sentence, and how a particular product is put together is for the vendor to say.

That readability is the real attraction, because you can look at your own rule and see what it will do. It is also the trap. A sentence that sounds sensible, such as "buy when price has dropped two percent", leaves out everything it did not say: how many times, how much in total, and what to do when the condition goes on being true.

Four questions every rule must answer

Before a rule goes near an exchange account, write down the answer to each of these in a single line. If any line is blank, the rule is not finished.

Two more details sit underneath the table. A trigger based on a price or an indicator uses the data the platform feeds it, so a delay or a gap in that data changes when the rule fires. And an order can fill only in part, so it matters which order types a rule can place, market or limit, and what it does with a half-filled order.

Worked example: ten falling steps, with and without a limit

Everything here is made up, including the rule. It is not a Coinrule template. Price starts at 100 and falls 2 percent a step for 10 steps. The rule buys $50 each time price is 2 percent under the last buy. We run it twice, once with no limit and once allowed to fire only 3 times. This is the same exercise as asking what a rule does if price falls ten steps in a row.

At step 10 price is 81.71. The rule with no limit has spent $500 and holds coins bought at an average of 89.33, now worth $457.32. That is an open loss of $42.68. The capped rule stopped after 3 fires, spent $150 and holds coins worth $127.63, an open loss of $22.37.

Look at what the limit did and did not do. It capped the money at risk at $150 instead of $500. It did not make the position a winner, and the capped rule is still under water because it bought into a falling price. If the account in this example held $300, the uncapped rule could not even complete its ladder, which is a different failure again. See the DCA bot guide for the arithmetic of repeated buying.

What a no-code builder takes off your hands, and what stays yours

• Taken off you: turning your sentence into calls to the exchange, and keeping the rule running. As we understand it, you do not have to write that code.
• Still yours: the logic. A builder will run a bad idea as faithfully as a good one.
• Still yours: the size, and the total. No screen can know what amount would hurt you.
• Still yours: the limits of the plan you are on, such as how many rules it allows, how often a rule may run, and whether a group of rules can be paused at once. Those are vendor facts to read on the vendor's pages.

A cautious first rule, in order

1. Write the rule as one plain sentence, then answer the four questions above.
2. Try to break it on paper. What does it do if price falls ten steps in a row? What if it rises ten?
3. Put it on the smallest size the exchange allows, with the cap switched on.
4. Read the first few orders it creates against what you expected, line by line.
5. Write down the loss at which you will switch it off, and keep a key limited to trading only. Our paper trading page covers the option of trying a rule without money.

What the trace does not show

No claim is made about how well Coinrule works or how safe it is, and nothing is set beside it for comparison. The trace uses an invented rule and invented prices. Crypto-assets and leveraged products can lose all the money put into them, and past or simulated results are not a guide to future results.