Diligence is the mother of good luck.
Two algorithms trade by different entry rules on thirteen instruments. Risk per trade is fixed, and everything is measured in percent of the deposit. Trade decisions are made by the code — they cannot be overridden by hand. Why it is built this way, I explain on the Author page.
The money stays in your account. You hand it to no one and transfer it nowhere: a copy subscription repeats the trades on your own account at the broker. You can switch it off, change the copy size or close positions yourself, at any moment. The fee is a share of the profit and only of the profit — no profit, no fee.
This is a calculation over five years of history, not the result of live trading. Live trading starts on 7 September 2026. How the calculation was made and what survives of it under the worst run of luck is on the Backtests page; what is happening on the live account is on the Results page.
The difference is in the signal that opens a trade. Their risk mechanics are the same and are described in the next section.
| Algorithm 1 | Trades against a false breakout. Price breaks yesterday’s high or low, fails to hold beyond it and comes back — the entry goes with the return. At most two trades a day per instrument, and the position is force-closed by the end of the session. |
|---|---|
| Algorithm 2 | Trades against an exhausted move. After an impulse the price pulls back and shows a reversal sign — the entry goes against the impulse. |
Both are counter-trend logics, and that is worth knowing: by their nature they do not insure each other — a market is conceivable where both end up on the wrong side. Their independence does not follow from the design, it is measured: correlation of results −0.13, losing days coincide no more often than chance would give, and in five years there was not a single month out of 61 where both algorithms were down.
The mechanics are identical for both algorithms and never change from trade to trade.
| parameter | how it works |
|---|---|
| Stop-loss | Placed on every trade at opening, calculated from the instrument’s current volatility. There are no trades without a stop. |
| Exiting a position | The take-profit is fixed, set in advance for each instrument as a multiple of risk: the position closes when it is reached, and the target does not trail the price. In addition, Algorithm 1 has a time-based exit — the position is force-closed by the end of the trading day, or before the instrument’s session ends if that comes earlier, while there is still liquidity in the market. That is why it never holds trades overnight. Algorithm 2 has no time exit: a position lives until its stop or take-profit. |
| Position size | 1% of the deposit per trade. Lot size is fitted to the width of the stop: the further the stop, the smaller the volume. So a loss at the stop is always 1% of the deposit, however wide that stop turns out to be. |
| Averaging down, grids, martingale | Not used. A losing position is never added to or reversed — it is closed at the stop. |
| Positions per instrument | One for each algorithm: while its trade is open, it makes no new entries on that instrument. The algorithms are independent, so one instrument can carry two positions at once, including opposing ones. |
| Holding time | Algorithm 1 — 3.6 hours on average, and never holds a trade longer than a day. Algorithm 2 — 10 hours on average, with 29% of trades carried overnight. |
A fixed stop means the loss on a single trade is capped at 1% of the deposit. The exception is a price gap at the market open: if price jumps over the stop level, the loss will be larger than calculated. Algorithm 2, which carries positions overnight, is exposed to this risk; Algorithm 1 almost never is.
This is the strategy’s debut on the platform. The historical calculation exists; live statistics are only beginning (Results) — and accumulating them is the main task right now. The goal for the coming year: to build up a full body of live trading statistics that can be presented not as words but as an export from the platform.
For the first year the fee is below market — on request. The usual rate for copy trading is 25–30% of profit, and the public offer is priced at 30%. But while the live record is only being built, charging everyone the full rate does not feel right: for the first year after launch, until 7 September 2027, I give a 20% rate on request — as a separate subscription link. The terms of an existing subscription do not change once it is set up: whoever joins at 20% stays at 20% after the year is over.
Work on the system continues. The set of instruments is not frozen: each new one goes through the same selection — a stability check across the two halves of history, a test against randomness, an assessment of execution costs. Whatever fails is not added, even if it looks profitable. A third algorithm with independent entry logic is in development.
Every expansion of the set has reduced the portfolio’s drawdown rather than increasing it: instruments are chosen so that they lose in different periods. That is the point of adding new ones — not so much a rise in return as a smoothing of the dips.