TWENTY-SIX GUARDS.
ONE JOB: KEEP YOU
IN THE GAME.
Most trading products sell you the upside and leave the downside as your problem. We built SentryQ the other way round — the guards came first, the strategies were fitted inside them. Every guard on this page applies to automated strategy trades and to signals you enter yourself.
DRAWDOWN IS WHAT
ACTUALLY ENDS ACCOUNTS.
A losing streak doesn't just cost you money — it costs you the capital base you need to earn it back. The deeper the hole, the more disproportionate the climb. This is why our engineering effort went into the floor, not the ceiling.
| If you lose… | You must then gain… | Reality check |
|---|---|---|
| 10% | 11.1% | Recoverable in a normal quarter |
| 20% | 25% | A whole good year, just to get level |
| 35% | 53.8% | Beyond most validated strategies' annual return |
| 50% | 100% | You must double the account to break even |
| 65% | 185.7% | Statistically, the account is finished |
Because a 50% loss demands a 100% gain, the single most valuable thing trading software can do is refuse to let you get there. Every guard below exists to keep the account inside the shallow, recoverable part of that table — even when a strategy is wrong, a market gaps, or you are asleep.
EVERY TRADE SIZED
FROM ITS OWN STOP.
Fixed lot sizes are the most common way retail accounts die. A 0.5-lot position on a 12-pip stop and a 0.5-lot position on a 90-pip stop are not the same trade — they are wildly different bets wearing the same clothes.
SentryQ never works in fixed lots by default. It takes the stop distance the strategy (or your signal) specifies, takes your configured risk percentage, converts through the pair's pip value into your account's own currency, and solves for the position size that risks exactly that much.
The consequence: a volatile gold trade with a wide stop automatically gets a smaller position than a tight EURUSD scalp. Your risk per trade stays constant even as the market's character changes underneath it.
FIVE TRADES,
ONE BET.
The most under-appreciated risk in FX is correlation. Long EURUSD, long EURGBP, long EURJPY and short USDCHF looks like four diversified positions. It is one enormous euro-versus-dollar bet — and when it goes wrong, all four go wrong together.
THE PORTFOLIO
EXPOSURE GUARD.
Before any entry, the engine decomposes every open position into its constituent currencies and computes the net lots you would hold after the new trade. If that number breaches your cap for any currency, the trade is refused — regardless of how good the setup looks.
Alongside it, the per-symbol cap limits simultaneous positions in the same pair. A three-target signal counts as one position group, so multi-TP signals aren't punished — the cap guards genuine concentration, not tranche structure.
These two guards routinely turn down more trades than any other. That is the point: the trades they refuse are precisely the ones that turn a bad day into a catastrophic one.
FOUR WAYS TO STOP
A BAD DAY SPREADING.
| Guard | What it watches | What it does | Why it exists |
|---|---|---|---|
| Equity Guardian backstop | Live equity against its all-time peak, continuously | Halts trading and flattens every position the moment drawdown crosses your limit (default 8%) | The structural floor. Whatever else fails — a strategy, a market, a stop — this one ends the session. |
| Daily loss limit | Realised + floating P&L since the day's start | Stops all new entries for the rest of the trading day once your threshold is hit | Stops a bad morning becoming a bad week. Automates the discipline humans lose exactly when they need it. |
| Drawdown risk throttle | Current drawdown depth | Automatically scales risk-per-trade down while the account is underwater, restoring it as equity recovers | Flattens the decline instead of compounding it. You bet smaller when you're losing, not bigger. |
| Cooldown guards | Consecutive losses, and recent stop-outs per pair | Stands the engine down on that pair (and globally after a streak) for a configurable period | Markets go through phases that suit a strategy badly. Cooling off beats re-entering the same bad regime. |
Every one of these rules is one a good discretionary trader already knows. The difference is that software does not get frustrated, does not need to win it back before dinner, and does not make an exception "just this once". The rules execute identically on your best day and your worst.
GUARDS THAT MEASURE
WHAT'S ACTUALLY
HAPPENING.
The newest layer doesn't work from rules of thumb. Each of these reads what your account and your broker are really doing — live prices, real fills, the moves that followed past releases — and acts on the measurement, not on an assumption. When conditions change, they change with them.
| Guard | What it measures | What it does |
|---|---|---|
| Open-risk budget | The total money at stake across every open trade at once — if every stop hit together | Refuses an entry whose own stop-risk would push the book past your budget. Sees that six tight-stopped trades can carry less risk than two wide ones — real risk, not a position count. |
| Correlation-aware exposure | How much of the open book actually moves with a new trade, from live prices | Blocks an entry that would quietly turn one bet into several — even across pairs that share no currency. Loosens on its own when pairs genuinely decouple. |
| Event-spike protection | How far each pair has moved in the hour after past releases of each event — learned from your own feed | Holds entries around the releases it has measured to whip that pair, catching the medium-rated event a generic impact tier underweights. |
| Weekly & monthly loss tiers | Realised losses over the week and the month, scaled to your balance | Backs the daily limit with a softer weekly brake (smaller size) and a harder monthly halt — so a bad run is caught over weeks, not only within a day. |
| Small-account mode | Your balance against the broker's minimum trade size | On a small account, tightens the whole envelope — fewer positions at once, lower open-risk ceiling — so a cluster of forced minimum-lot trades can't come to dominate it. |
| Live-book stress test | What a single sharp move against your whole book at once would cost | Prices that shock every cycle and warns when it crosses a share of your balance — so a book that looks calm but is one headline from a bad day is visible before the headline. |
| Broker execution quality | The broker's recent fill-versus-reject rate, per pair | Tells you plainly when a broker starts refusing orders — and why — and holds new entries there until fills resume. No more silent idleness when a broker won't accept your orders. |
| Profit protector | Each open trade's peak profit against where it is now | A winner that runs past its own risked distance is watched; give back half of that peak while still green and half the position is banked at market — the rest rides your existing stops and targets. A good trade can't quietly round-trip to nothing. |
Alongside the guards that block, two measured advisers now report. The per-strategy equity-curve throttle turns the old performance circuit-breaker into a dial — a strategy whose live results slip is traded at half size before it's ever paused, and restored automatically on recovery, so one strategy's cold streak no longer drags on the others. And the stop audit measures how far each winning trade actually went underwater before it won, then tells you in plain words whether a strategy's stops are well-placed, wider than they need to be, or tight enough that ordinary noise is a real risk.
THE TRADES THAT LOOK
FINE ON A CHART.
Some losses have nothing to do with whether your analysis was right. They come from trading into a scheduled news spike, a spread that quietly tripled, a weekend gap your stop can't cover, or a margin level that was one bad tick from a broker stop-out. These guards handle that whole category.
| Guard | Trigger | Action |
|---|---|---|
| News Guard | High-impact calendar events (NFP, CPI, rate decisions) from a live economic calendar | Suspends new entries in a configurable window either side of the release |
| Spread guard | Spread on a pair exceeds its normal band | Refuses entries and stop moves on that pair until liquidity normalises |
| Rollover guard | The daily liquidity gap around broker rollover | Pauses all new entries through the window |
| Weekend gap guard | Approach of Friday close | No new positions into the weekend — because stop-losses do not protect across a gap |
| Margin floor | Account margin level falling toward the broker's stop-out | Pauses entries below your floor; flattens if it becomes critical — on your terms, before the broker's |
| Session time filters | Per-strategy trading windows | Each strategy trades only the sessions it was validated in |
| Cancel-on-reversal optional | The setup behind a resting order disappears | Withdraws the pending order rather than letting it fill into a dead idea |
EVERY GUARD, ONE
PANEL, YOUR CALL.
Protection you can't see isn't reassuring — it's just opacity. The safety panel lists every guard, its current state, and exactly what it does in plain English. Each is yours to tune or switch off, and the console tells you how many are active at a glance. Turning one off now snoozes it for 24 hours and re-arms it automatically — so an account is never left unprotected indefinitely by a toggle flicked once and forgotten.
All twenty-six, in the engine's own words
Copied from the running engine rather than rewritten for the website, so what you read here is exactly what the safety panel tells you. Five ship off by default because they trade protection for opportunity in a way that should be your decision, not ours.
| Guard | What it does |
|---|---|
| Equity Guardian | Halts trading and closes everything if equity falls 8% below its peak — the hard account backstop. |
| News guard | Holds new entries from 30m before to 15m after high-impact economic releases. |
| Pre-news protection off by default | Just before a high-impact release, locks in profit or moves stops to breakeven on exposed positions instead of letting them ride the spike. |
| Spread guard | Skips entries when a pair's spread is abnormally wide (>3× its normal) — thin liquidity, rollover, news gaps. |
| Rollover window off by default | Pauses ALL new entries in the daily liquidity vacuum around the 5pm New York close (20:55–22:30 UTC). |
| Currency exposure caps | Blocks entries that would concentrate more than 3 lots in one currency (8 gross across the book). |
| Profit protector | A winner that runs past 0.5× its own stop distance is watched; if it then gives back 50% of that peak while still in profit, 50% is banked at market and the rest rides your existing stops and targets — so a good trade can't quietly round-trip to nothing. |
| Look after my own trades | Positions you open yourself are counted in the exposure limits, and once one is ahead by 1x what it was risking its stop moves to breakeven. Only pairs SentryQ trades, only trades that already have a stop, and it never closes, adds to, or loosens anything. |
| Close my own trades before the weekend off by default | Also flattens positions you opened yourself ahead of the Friday close. A stop cannot protect across the weekend gap — it fills on the far side of it — but ending a trade you meant to hold is your call, so this is separate. |
| Broker execution quality | Watches the broker's recent fill-vs-reject rate per symbol, warns when a pair starts being refused (e.g. invalid volume, off-quote), and holds new entries there until fills resume — so a rejecting broker is a plain signal, not silent idleness. |
| Live-book stress test | Each cycle models a single 60-pip adverse shock hitting every open position at once and warns when the loss would exceed 15% of balance — the joint move the correlation guard says is plausible, priced before the headline. |
| Event-spike protection | Learns from your own feed how each pair reacts to each event, and holds entries for 15m before to 30m after any release measured to move the pair a median 25+ pips — catching the ones a generic impact rating underweights. |
| Small-account mode | On a balance under $2,000, caps concurrent positions at 3 and the open-risk budget at 6% — so a cluster of forced minimum-lot trades can't come to dominate a small account. |
| Correlation-aware exposure | Measures how much of the open book moves WITH a new entry (|r| ≥ 0.7 from live prices) and blocks it once that correlated cluster reaches 3 lots — catches one bet worn as several pairs, even when they share no currency. |
| Medium-impact news blocking off by default | Also holds new entries around MEDIUM-impact releases (regional CPI, retail sales etc.) — the high-impact-only guard ignores these. |
| Pair losing-streak cooldown | After 3 consecutive losing trades on one pair, no new entries on it for 24h — a pair that keeps failing gets benched. |
| Re-entry cooldown | After a losing trade, the same strategy waits 60m before re-entering that pair — stops rapid clip-and-re-enter churn. |
| Symbol loss breaker | Benches a pair once it has cost more than 2% of balance over 7 days. Counts money rather than consecutive losers, so one small win can't clear it. |
| Per-pair position cap | Caps how much can sit on one pair across ALL strategies, so several strategies can't stack the same trade. An open position costs a full slot; a resting order that has not filled costs half, because several limits genuinely can fill together but an unfilled one is not a trade. |
| Drawdown risk throttle | Automatically trades smaller while the account is in a drawdown (ramping down from 3% below peak), restoring full size on recovery. |
| Weekend gap guard — no new entries off by default | Stops new trades opening from Friday evening until the Sunday reopen, so nothing new is carried into the gap. Measured cost: under 2% of the flagship's profit, with no change to its worst drawdown. |
| …and close open strategy positions too off by default | Also closes positions the strategies already hold before the Friday close. ⚠ The expensive half: about 10% of the flagship's profit over three and a half years, and it made the worst drawdown slightly worse. Insurance against a gap severe enough that it is not in that history. |
| Margin guard | Refuses new entries when the margin level drops below 200% — keeps the account clear of the broker's stop-out. |
| Performance circuit breaker | Auto-pauses any strategy whose live results degrade well below its validated track record, and notifies you. Open positions stay managed. |
| Open-risk budget | Caps the total money at stake across all open trades — if every stop hit at once — at 6% of balance. Sees real risk, not just position count. |
| Weekly & monthly loss tiers | A slow bleed a daily limit misses: past 12% down on the week, new-trade size halves; past 25% down on the month, trading halts. |
ONE TAP
TO FLAT.
The kill-switch is the guard of last resort, and it is deliberately available from everywhere: the desktop console, the mobile app, and your web account page. One action cancels every SentryQ order, closes every SentryQ position, and pauses the engine.
The remote path matters more than it sounds. If you are away from your desk and something happens — a geopolitical shock, a broker outage, a change of heart — you do not need to reach your PC. The command relays through your account and your engine executes it locally within seconds.
And because everything the engine opened carries a SentryQ label, the kill-switch touches only its own trades. Positions you placed yourself in your broker's platform are left completely untouched.
THE GUARDS REPORT
WHAT THEY REFUSED.
Any system can claim it has protections. The test is whether it will show you the trades it turned down — including the ones that would have won. SentryQ writes that list every single day.
A reason is not a measurement. So each entry a guard refused is replayed against the market that followed, with the strategy's own stop, and the outcome is kept for thirty days: which guard blocked it, how many times, and the pips that block avoided or cost. The safety panel shows the running total — in the screenshot above, refused entries would have lost 28 pips net, so the guards were earning their keep that month.
They do not always earn their keep, and we publish it when they don't. On the account we run this on, the last thirty days read: open risk guard −27.6 pips, spread guard −21.9, symbol loss guard −15.1 — all money saved — and the per-pair cap +26.5, money it cost you. That last line is the reason the cap changed in v0.9.168. The full working, and what we did about it →
It is deliberately honest in both directions. Blocks that cost you profit appear with a positive number beside them, in the same table, un-highlighted and un-explained-away. If your guards are set too tight, this is the number that tells you.
It has already changed the product once. In July 2026 the per-pair cap showed up as the only guard on an account costing money rather than saving it. Reading the individual blocks showed why: the cap counted an unfilled limit order exactly like an open trade, so on an account resting orders across twelve pairs, eleven of fourteen occupied slots held nothing at risk and three pairs were closed to every other strategy. Orders now cost half a slot. Whether that was the right call is a question the same scorecard will answer.
Some of those eleven refused trades would have been winners. We show them anyway, because a protection system you only hear about when it's convenient is a marketing claim, not a risk control. Seeing the cost of your own safety settings is the only way to judge whether they're tuned correctly — and to trust them when they matter.
Risk controls reduce risk — they cannot eliminate it. Gaps, slippage, broker outages and extreme events can still produce worse fills than intended, and no guard can protect against every scenario. Trading FX on margin carries a high level of risk. Never trade with money you cannot afford to lose. Read the full risk disclosure →