One grid per pair, one direction chosen by the swap, no stop-loss per trade, and five defenses that freeze but never close. Scroll down: depth here is drawdown, and we publish how deep it went.
SurfBot does not try to guess where price goes. For each pair it picks the side that pays the better swap, lays orders one spacing apart, and each layer closes on its own take-profit. The edge is the swing plus the swap of the chosen side. Nothing else.
Long or short, never both. AUTO reads the swap of the pair and takes the side that gets paid to wait. You can force it. You cannot run both sides: there is no hedge in this engine.
Spacing comes from a built-in table per pair, as a percent of price. Every layer carries a take-profit one spacing away. When price breathes back, the layer closes and is re-armed.
With an empty grid, the pending orders trail the market so the first layer is never far away. Four pendings on screen, re-centered as price moves.
Drag the price against the grid and watch what accumulates. After the 4th layer the spacing widens by 1.35× each step, so the deep layers are further apart. The cap is 12 layers per pair: beyond that, the grid freezes.
Simplified illustration: uniform spacing for the first 4 layers, widening 1.35× per layer from the 5th, one lot per layer. Real spacing is per pair (percent of price) and the account freezes may stop adds earlier.
Every defense in SurfBot stops new entries and cancels pendings. None of them closes a position. That is the whole point and the whole risk: a frozen grid still floats with the market until price comes back, or until your broker's stop-out.
MaxTotal12 layers per pair, positions plus pendings. The 13th is never placed.
PairLossFreezePctFloating loss of one pair beyond N% of the starting balance freezes that pair and cancels its pendings.
MaxNetLayersPerCcyNet layers per currency across the whole account. Seven USD pairs cannot all pile on the dollar.
EquityFloorPctEquity below N% of the starting balance freezes every grid on the account.
MarginFreezePctFree margin below the threshold or spread above the filter: no new layer. The red ring is the last one before the broker's.
MT5 Strategy Tester, the 7 live pairs at 0.04 lots on USD 16,000, the exact engine that runs the live account. Five crisis windows between 2021 and 2025, portfolio equity summed across pairs. The curves behind this panel are those runs.
Drawdown in a grid scales with lot per capital. Put your numbers in and see what the same five windows would have done. This is a linear extrapolation of the published test, not a promise: a worse year than 2025 is always possible.
Rule of thumb from the test: 0.01 lot per USD 10,000 keeps the five windows under 25% drawdown. Margin level scales the same way. Hedging account, 1:100 or higher, low-swap broker.
No lifetime purchase while the live track record is this young. Rent for a month, run the free demo in your tester, read the stress test above, and only then decide.
No, and distrust anyone who does. SurfBot is risk engineering applied to a grid. Past results, even the ones MQL5 verifies, do not guarantee future results; the stress test above shows how deep it has already gone.
You need to understand what the robot does and what it risks; the inputs are documented for that. Run the free demo in the MT5 Strategy Tester and size your capital with the calculator before any real account.
On signal 2383326, the author's own account, audited by MQL5 and shown live at the bottom of this page. Not a screenshot, not a backtest.
A grid has no stop per trade, so a long one-way trend piles up losing layers. That is exactly why the five rings exist, and why sizing the capital matters more than any input.
Manta is the one gliding beside you on this page. A manta does not fight the current; it rides the layers of water and filters what passes. That is the whole philosophy of this engine, and why she, not a surfer, keeps you company below the surface.
The author's own account, the one behind signal 2383326. The robot on it posts equity, balance and margin every 10 seconds; this page keeps one point per day. Up to 9 Oct 2026 the curve is rebuilt from the balance and equity samples MQL5 publishes on the signal page; from 10 Oct 2026 it is our own daily record.
Loading the live account…