I built Scalperology in response to a specific problem with discretionary XAUUSD scalping: the cognitive load of identifying valid scalp setups on M1 across an 18-hour gold session is simply unsustainable for a human trader. By hour 6 you are taking lower-quality entries; by hour 10 you are revenge-trading the last loss. The neural-network filter inside Scalperology exists to do exactly the pattern-matching work that exhausts a manual trader, while leaving the heavy risk-management decisions (position size, stop placement, hard stop on the account) to deterministic rules I can inspect and explain.
The core strategy is what I call liquidity-aware momentum scalping: the EA looks for M1 candles that break above or below the previous fractal high or low, confirmed by M5 momentum, but only when the current spread is below a dynamic threshold (typically 1.0–1.5 pips depending on session). The neural-network filter is trained on roughly 8 years of XAUUSD tick data and inspects features from the last 50 M1 candles before issuing a yes/no on the entry. In practice it rejects about 60% of rule-qualifying setups, which is the entire point — the rules-only version of this strategy is profitable but noisy, and the filter strips out the noise.
Exits are deliberately simple: a fixed stop-loss (volatility-scaled by the ATR of the last 30 M1 candles), a fixed take-profit at 1.5× the stop distance, and a time-based exit if the trade hasn’t reached either after 12 M1 candles. The trailing stop activates only after the trade has moved 1× the stop in our favour. I deliberately resisted adding complex break-even or partial-close mechanics because every additional exit rule increases the surface area for broker-execution quirks to interfere.
News behaviour: Scalperology pauses 15 minutes before any high-impact news release on the embedded calendar and resumes 30 minutes after. NFP, FOMC, ECB decisions, and major geopolitical announcements always fall into this category; the news filter has prevented the EA from getting caught on the wrong side of every major XAUUSD spike since the live reference account began publishing.
Why M1 and not M5? Faster timeframes produce more samples and let the filter ML model converge on a real edge with statistical confidence. The trade-off is execution sensitivity — at M1 every pip of spread matters. That is why I will only recommend Scalperology on an ECN broker with sub-1.5-pip XAUUSD spread. Without that, the math does not work.
Recommended setup
Three things move Scalperology's realised performance away from the published reference account: broker spread, execution latency, and risk-per-trade configuration. Here is what I recommend for each:
Broker
Execution model: ECN or raw-spread account with explicit commission ($3–7 per round-turn lot). STP/market-maker accounts work but cost you 30–50% of the published edge in spread overhead.
Server location: London (Equinix LD4) for the European session bias. Several US-server brokers also work for the NY session, but LD4 covers the high-liquidity window better.
Scalping policy: confirmed in writing. A handful of retail brokers either prohibit scalping outright or quietly throttle scalpers via slippage; verify before depositing.
VPS
Strongly recommended — see the FAQ below for the full reasoning. Specifics: 2 vCPU / 4 GB RAM / 60 GB SSD on Windows Server 2019 or 2022, located in the same data centre as your broker's MT5 server. Sub-20 ms ping is the target.
Risk per trade
Default setting: 1% risk per trade. For accounts under $1,000 I usually recommend dialling down to 0.5% to absorb a longer losing streak without forcing oversized lot rounding. For prop-firm accounts the same applies even more strongly — see the prop-firm FAQ below.
Presets (.set files)
Five preset risk profiles ship with the license: Conservative, Defensive, Balanced, Aggressive, Maximum. Default install applies Balanced; switch presets via the EA's input dialog without restarting MT5. On the Cloud subscription tier the active preset is auto-updated by the .SetAnalyze pipeline based on rolling out-of-sample tests.
Backtest results and caveats
Scalperology has been backtested against multi-year XAUUSD tick data across multiple brokers (IC Markets, Pepperstone, RoboForex tick exports). Across the full sample the strategy is profitable with a profit factor in the 1.3–1.6 range and a maximum drawdown in the 8–15% range depending on the preset. I deliberately publish the live reference account on /live-trading rather than the backtest equity curve because backtests on tick data still under-represent the realistic spread widening during news, and we have all seen marketplaces that publish backtests that look like financial perpetual-motion machines.
Use the MT5 Strategy Tester (Modeling: "Every tick based on real ticks", 90%+ modelling quality) to reproduce the backtest on your own broker's historical data — see the Strategy Tester guide for the procedure. Your numbers will differ from mine by the magnitude of the spread difference between your broker and the brokers I backtested on.
Risk profile
Even on the conservative preset, Scalperology can produce losing weeks. The typical pattern when it loses is a sequence of 3–6 stops in a row during a period of low gold volatility or unusual session behaviour (extended Asia session, holiday-thinned liquidity). Expected ranges for the Balanced preset on a properly-sized account:
Typical max drawdown
8–15% of account equity, historical.
Win rate
Roughly 55–65% across recent samples; exact number depends on broker.
Profit factor
1.3–1.6 across tested samples.
Trades per active session
8–20 typical, more during high-volatility sessions, zero during the news filter window.
Recovery factor
Above 2 on the published reference account; lower on accounts with wider spreads.
These figures are descriptive of historical and reference-account performance. Live performance on your own account will differ — sometimes materially — based on broker execution, your chosen preset, and account size relative to the EA's tested operating range.
Pros & cons
Pros
Verified live performance on a public broker-API-connected account.
Neural-network filter is a real edge, not marketing language — it materially improves on the rules-only equity curve.
Deterministic, inspectable exit rules. No martingale, no grid, no doubling.
Built-in news filter prevents the worst-case slippage during scheduled events.
Five risk presets cover everything from prop-firm-friendly to scaling-account aggressive.
Cons
High broker-execution sensitivity — needs an ECN account to perform to spec.
Active strategy. Without a VPS your laptop's reliability becomes the bottleneck.
$500 minimum deposit is a hard math floor, not a marketing floor — smaller accounts will under-perform.
Most demanding in our line. If you want lower-attention trading, look at Trendopedia AI.
Scalperology AI runs best on ECN brokers with raw-spread XAUUSD pricing under 1.5 pips during liquid hours and sub-100 ms latency to the broker's server. IC Markets (LD4 server), Pepperstone (LD4), and RoboForex (LD4) cover most of our reference accounts. Any regulated broker with those specs works — see /brokers for current EA-friendliness reviews.
Because Scalperology trades dozens of times per session on M1, even a 0.3-pip difference in average XAUUSD spread compounds to a meaningful number across a month. We have tested the EA on roughly 20 brokers; the published profit-factor figure assumes an ECN account with raw-spread pricing and 0.5-pip-or-better headline XAUUSD spread during liquid London/NY hours.
If you trade with a market-maker broker (wider STP spread, no explicit commission), the EA still works but the realised PF is meaningfully lower than the reference account.
Do I need a VPS for Scalperology AI?
Yes — a Forex VPS is strongly recommended. Scalperology operates on M1 with 5-second-or-faster decision windows. Without a VPS in the broker's data centre your laptop's latency, sleep cycles, and intermittent connectivity will materially diverge from the reference account. Pick a Windows VPS in London (LD4) or near your broker's server with sub-20 ms ping.
Specific recommendation: 2 vCPU / 4 GB RAM / 60 GB SSD Windows Server 2019 or 2022, ideally at the same data centre as your broker's MT5 server. BeeksFX, FXVM, NYC Servers, and Vultr all have offerings in the £15–25/month range that work fine.
If you cannot use a VPS, run Scalperology on a desktop you keep awake during trading hours and accept that intermittent disconnections will cost trades and skew your live results away from the published reference.
What is the minimum deposit for Scalperology AI?
The recommended minimum is $500. The EA technically scales down to 0.01-lot positions and can run on smaller accounts, but at $200 or less the commission overhead is a large share of P&L and your realised performance will materially diverge from the published reference. Above $500 the EA hits its intended operating range.
The math behind $500: at 1% risk per trade with a typical 25-pip stop on XAUUSD, the minimum lot size that respects the risk envelope is around 0.01–0.02 lots on a standard account. Below $500 you cannot consistently honour the risk-per-trade rule without rounding to 0.01 and effectively risking more than 1%.
This is not a license restriction — the EA runs on smaller accounts. It is a math restriction: small accounts trade outside the EA's tested envelope and produce noisier results.
Is Scalperology AI suitable for prop firm challenges (FTMO, FundedNext)?
Scalperology can be used on prop-firm accounts but with caution: many prop firms impose daily loss limits that an active scalping system can breach during high-volatility sessions. Run it with risk-per-trade dialled down to 0.3–0.5% and an explicit daily-loss circuit breaker. For prop firms specifically, Trendopedia AI is usually the lower-stress choice.
Prop firms typically enforce three rules that conflict with default scalper behaviour:
1. Daily loss limit (often 5%). A bad XAUUSD news event can push a 1%-per-trade scalper close to this in one session.
2. Total drawdown limit (often 10%). Easier to manage, but still tighter than retail.
3. Trading-day minimum. Scalping naturally satisfies this — not a problem.
If you do run Scalperology on a prop account, reduce `RiskPercent` from 1% (default) to 0.3–0.5%, enable the news filter, and set a daily P&L circuit breaker outside the EA (most prop firms provide one). See the planned /prop-firm/best-ea-for-prop-firms-2026 page for a fuller comparison.
Try Scalperology AI
Free demo + 30-day money-back guarantee on every license tier.