What a Bitcoin robot at sixty trades a year can and cannot show you

Bitcoin Interval trades one instrument, BTCUSD, about sixty times a year. That one number decides how every other figure on its page should be read, and it decides what you can and cannot learn about the robot in the time most people give a robot to prove itself, which is a month.

So this is a post about sixty. What sixty a year can show you in an afternoon, what it cannot show you in a month, and what it took six years to show at all.

Five decisions a month

Sixty a year is five a month. That is the listing's own figure divided by twelve, and it is the whole problem in one line.

Real months are lumpier than the drawing. The robot reads calendar windows, a session window inside the day and the weekend, and most windows close with nothing happening. A month with two trades and a month with nine both exist in the test. But the average is five, and five is what a month of watching it on your account will give you.

What a month can show you

Quite a lot, as long as you ask it the right questions. A month shows you the plumbing.

It shows whether your broker offers Bitcoin against the dollar under a name the robot finds, whether the stop goes out with the entry on your account the way it does in the test, whether the time limit closes a position that went nowhere, and whether the panel tells you what it is doing. Every one of those is a yes or a no, and none of them needs a large sample. They need one trade each.

The other thing you can check in an afternoon is the test itself. The free demo runs in your own Strategy Tester on your own broker's data. Symbol BTCUSD, any timeframe, because the chart's symbol and timeframe are ignored. One-minute bars or the tick model, both are valid and they agree, so the result does not depend on optimistic fills. A 3000 USD account at the default package over 2020.01 to 2026.08 reproduces the row on the listing on our feed, +412.7% at a 16.77% worst equity drawdown, profit factor 1.50, 397 trades. On your feed the number will differ, and that difference is information: across five brokers we measured, the profit factor ran from 1.44 to 1.51 and the return landed 12 to 34 percent below the reference feed. One of those five is worth a look on its own, because its equity drawdown read 29.45% against a balance drawdown of 17.45%, which is open-position float rather than realised loss, and you would want to know that about your broker before your money is on it.

That is what an afternoon buys: the plumbing on your account, and the test on your broker's prices.

What a month cannot show you

It cannot show you whether the edge is real. Not because the edge is hidden, but because five decisions are not enough to separate a real edge from an ordinary run of luck, and this is true of every robot at this trade count, ours included.

To make that concrete I built a model from two figures on the listing and nothing else: 47.1% of trades closed in profit, and a profit factor of 1.50. Those two together mean the average winner is 1.68 times the average loser. Then I let a computer play that coin, with those odds and that payoff, twenty thousand times, and looked at where the paths stood after five trades, fifteen, thirty, sixty, and so on.

After five trades, a month, the median path is barely above zero, the middle of the band runs from about two average losers down to nearly six up, and 23% of all paths are under water. That is a robot with a real, measured edge, and one month in four it looks like a robot with none. After thirty trades, half a year, 17% of the paths are still under water. After sixty, a year, 7%. After a hundred and twenty, one in a hundred. After two hundred and forty the share rounds to zero, and after the full 397 the whole band sits above zero, from about sixty average losers at the pessimistic end to nearly a hundred and fifty at the optimistic one.

This is a model, not a result. It assumes every trade is independent and that the two listing figures hold, which is exactly what a backtest cannot promise about the future. But it is the honest shape of the question, and the shape does not change with the assumptions: a month cannot show what a year can, and a year cannot show what the test can.

It also applies to our own live account. TechnoTrader EA Showcase runs this robot at factory defaults, alongside Gold Tempo and Index Cadence, on one small real account. It is three weeks old and currently in the red. Read it for the plumbing, not for a verdict: three robots share that account, its curve belongs to the three together, and even the share of it that is Bitcoin Interval's is a handful of trades. The figure above says what a handful of trades can tell you.

What six years can show you

The test window runs from January 2020 to August 2026, and it was split in half before it was measured, not after. The first half, 2020.01 to 2023.04, returned +132.5% at a profit factor of 1.52 over 195 trades. The second, 2023.05 to 2026.07, returned +110.0% at 1.47 over 202 trades. Per year that is 29.4% against 26.0%. The halves differ in drawdown, 11.52% against 16.42%, which is what a changing market does to a fixed set of rules.

Every calendar year in the window is positive, and the thin one is on the slide exactly like the good ones: 2025 returned +9.0% at a profit factor of 1.17. A year-by-year table is the cheapest way to see whether a system is one lucky stretch or a repeated one, and it is the first thing to look for on any listing that does not print it.

Account size barely changes the result. From 500 to 10 000 USD the return per year moves between 24.9% and 29.1% at the same profit factor. Fewer than half the trades are profitable, by design, which is why the profit factor and the payoff ratio are the numbers to judge by rather than the win rate.

None of this is a promise. A backtest describes how fixed rules behaved on prices that have already happened. What six years can show, and a month cannot, is that the rules behaved the same way in two halves that were separated before anyone looked.

What that means for the price

The listing prices the robot by what can be shown, and the same arithmetic explains why. Sixty trades a year cannot prove themselves live in a month, or in three. So the price today reflects a backtest window and a live account that is weeks old, and the listing says where it goes as the live record grows: 99 USD until the end of Sunday 13 September 2026, 199 USD from Monday 14 September, and an expected settled price of 799 USD as the live record fills in. Buyers keep all future versions at the price paid. The date will not be extended.

Waiting is a reasonable choice. It buys certainty, at a price that rises with it. Buying now is a reasonable choice for someone who has read the two halves, the year-by-year table and the cost row, and is willing to be paid for carrying the part of the uncertainty that only time removes.

What to do this weekend

Run the free demo in your Strategy Tester on your broker's data, at the defaults, over the full window. Compare your profit factor and drawdown to the listing's, not your return, because the return depends on your feed and the profit factor tells you whether the mechanism survived it. Read the year-by-year table and decide whether +9.0% in a year is a year you can sit through. Pick the risk package by the drawdown you can hold, not by the return column. Then decide.

And if you buy it, judge it the way the arithmetic says it can be judged: the plumbing in the first month, the rules over the first quarter, the edge over a year and more. Anyone who tells you otherwise about a robot at this trade count is describing a different robot, or a coin.

The listing, with every figure above and the full list of what was measured and rejected, is Bitcoin Interval TechnoTrader.