EdgeDriven Gold Portfolio
- Experts
-
Dan Mishima
Edge, Backed by Evidence.
~ 근거에 뒷받침된 우위를 ~
EdgeDriven Algo는 통계 검증을 통해
견고성·재현성을 중시한 알고리즘 거래 시스템을 개발하는 EA 브랜드입니다.
우리가 중시하는 것은 단기간의 높은 승률이나
보기에 좋은 백테스트만이 아닙니다.
장기 데이터, 미지 기간, 스트레스 환경 등 여러 각도에서 로직을 검증하고,
특정 시세에만 의존하기 어려운 EA의 개발을 목표로 하고 있습니다.
또한 단일 EA에 의존하는 것이 아니라, - 버전: 1.5
- 활성화: 5
EdgeDriven Gold Portfolio is a fixed, multi-strategy XAUUSD Expert Advisor for MetaTrader 5 hedging accounts. One installation manages twenty strategy modules with shared risk and operating controls.
EdgeDriven Algo. Edge, Backed by Evidence. The figures below are historical simulations, not live or Forward returns. Leveraged Gold trading can cause substantial losses.
Why a portfolio?
Different modules can contribute at different times, while one installation keeps their ownership and operating controls together. Twenty modules do not mean twenty independent sources of future profit. They share Gold-market exposure, margin and the possibility of simultaneous losses. This product is for buyers willing to examine both the evidence and the operating requirements before choosing their risk settings.
What is inside
The portfolio and intended relative allocation are fixed. Modules have separate state and order ownership; individual strategy recipes are not tuning inputs. Risk profiles scale the same intended relative allocation rather than select different strategies.
Position size rounds downward to the broker's volume step. If its minimum lot would exceed the permitted sizing budget, the trade is skipped, not rounded upward or reallocated to another module. Small accounts can therefore participate differently. Increasing risk merely to force a trade is not a solution.
Historical assessment process
The Gold portfolio research used Development candidates and Development plus Validation priority setting. The selected twenty-module portfolio retained its priority fixed before the Final period; the Final results were not used to rerank the candidates. Combined research history was descriptive after selection.
Later data had already been used in component qualification or earlier diagnostics. Development, Validation and Final labels here denote historical assessment periods, not completely unseen individual-strategy tests or live/forward performance. The published production results and their modeling limitations remain the evidence for this release.
Historical evidence: Standard profile
Reference configuration: USD50,000 initial capital, USD hedging account, leverage 1:100, common scalar 3, PortfolioRiskScale 100%, no compounding. One continuous native MT5 test covers January 2014 through June 2026. The original statistics are linked to R08R3, version 1.001. The identical full-history Standard test was repeated with the R08R7 version 1.005 compatibility candidate: all report tables, order rows and deal rows matched, apart from the Expert filename.
| Official MT5 statistic | Historical result |
|---|---|
| Net profit | USD65,090.09 |
| Profit factor | 1.45 |
| Trades | 7,512 |
| Maximum modeled equity drawdown | USD4,281.13 |
| That drawdown relative to its equity peak / initial capital | 4.52% / 8.56% |
| Recovery factor | 15.20 |
Drawdown percentages have different denominators. They are not interchangeable and neither is a future loss limit. Floating drawdown is measured at the tester's modeled control points, not from real ticks.
How the historical periods differ
| Period and purpose | Net USD | Net-position PF |
|---|---|---|
| 2014–2022: Development | 47,331.92 | 1.425 |
| 2023–2024: Validation | 9,510.59 | 1.371 |
| 2025–2026H1: Final Historical Confirmation | 8,247.58 | 1.892 |
| 2014–2026H1: Combined descriptive history | 65,090.09 | 1.445 |
These are sections of one continuous run, not separately reset accounts. Net-position PF is calculated from realized position cash after costs; the official MT5 result above is displayed at its report precision. Later periods had previously been observed during component development. Final Historical Confirmation is not untouched OOS, never-seen data, or Forward trading. Combined history is descriptive, not an additional independent validation sample. The 2026 bucket contains H1 only.
Losses and time without a new high
The same 150 calendar months include 92 positive and 58 negative months. The worst calendar month was January 2023: −USD1,215.32. The worst month-end rolling 12-month window was July 2021 through June 2022: −USD2,140.97. Positive yearly totals did not prevent losing months or a losing 12-month window.
Balance stagnation reached 362.92 days. Hourly-observed equity stagnation reached 441.04 days; this is not an intratick maximum. These are elapsed historical periods without a new high, not forecasts of recovery time. Future losses or stagnation can be worse.
Risk profiles: reference configurations, not return projections
| Profile | Initial USD / scalar | Net USD | Equity DD USD | DD / initial capital |
|---|---|---|---|---|
| Conservative | 100,000 / 1.5 | 65,090.09 | 4,281.13 | 4.28% |
| Standard | 50,000 / 3 | 65,090.09 | 4,281.13 | 8.56% |
| High | 50,000 / 4.5 | 102,050.87 | 6,449.43 | 12.90% |
All rows use native 2014–2026H1, M1 OHLC, non-compounding reference tests. Standard is bound to the original public EX5 and the exact R08R7 full-history regression; Conservative and High endpoint confirmations use the prior same-core, QA-instrumented build. They are not separately claimed public-EX5 endpoint reruns. Conservative at USD100,000 and Standard at USD50,000 have the same absolute sizing basis; equal dollar results do not mean equal capital-relative risk. Rounding and skipped trades can change participation, so scaling is not a linear return projection. “Conservative” is a profile name, not a safety guarantee.
These deposits are verified reference configurations, not universal broker minimums or personalized capital recommendations. Other balances, compounding, contract specifications and execution conditions can produce different behavior and results.
Risk Controls
Operational controls manage order handling; they do not prevent losses or guarantee execution.
Friday cleanup and weekend pause
Designed to avoid routine weekend position carry through scheduled Friday cleanup and a weekend trading pause. Execution remains subject to broker availability, terminal connectivity and market conditions.
New entries are blocked from Friday 20:00 UTC until Sunday 23:00 UTC; the fixed weekday session rules can delay the next entry further. From Friday 20:00 UTC, the EA requests closure of its module positions and removal of its pending orders. Unacknowledged requests remain eligible for retry in the Friday/Saturday cleanup window. These are UTC rules, not a universal 20:00 broker-server setting. Live time conversion requires a correct computer UTC clock and a valid observed server offset; the historical reference uses NY+7/EET-US timestamps.
At Sunday 23:00 UTC the weekend gate ends, but the fixed weekday-only module sessions still exclude Sunday entries. Subsequent entries require an eligible weekday session, a valid signal, quotes and broker trading availability.
Reading weekday and hourly statistics
MT5 Entries by weekdays/hours counts entry deals. Profits and losses by weekdays/hours groups exit deals by their recorded closing time. Scheduled pre-weekend cleanup can therefore concentrate realized profit or loss in the Friday bucket, or the corresponding server-time hour/day. This is not evidence of a Friday-specific entry edge. The report clock and cleanup clock may differ, especially for the UTC-based Gold rule.
Protective stops and order checks
Module-specific protective stops and exits are preserved. Accepted hard stop-loss/take-profit orders remain broker-side, while software exits and pending management require the terminal. Sizing checks broker volume steps and limits, stop-based nominal risk and available margin; below-minimum volumes are skipped without increasing lots. Order handling checks broker stop-distance and pending-order capacity constraints. These checks do not guarantee a fill, exit price or maximum loss.
Restart and ownership
Owned positions and orders are identified by symbol and module Magic range. Persisted state and recovery logic restore module management, and a terminal-local ownership lease checks overlapping active Magic ranges. This does not coordinate separate terminals or VPS instances. Preserve terminal state and prevent duplicate management during migration. A controlled terminal-process restart was tested on R08R2; that exercise was not repeated on this R08R7 release and is not an OS/VPS power-loss test.
Economic news
The embedded legacy News Guard is disabled in all twenty modules and is not an active feature of this release. There is no active event, importance, currency or pre/post-event window, and no news-triggered entry pause or position closure is claimed. Published tester results do not verify an enabled news filter; no live economic-news protection is provided.
Friday cleanup and software management depend on incoming quotes, connectivity, trading permissions and broker acceptance. Pending orders can fill while the terminal is offline, and gaps or slippage can increase losses.
Operating requirements
- MetaTrader 5 hedging account; the disclosed reference is USD with leverage 1:100. Netting is not supported.
- Attach one instance to a broker XAUUSD H1 chart with synchronized H1/H4 history. Version 1.005 evaluates the current chart symbol without an instrument-name refusal. It can place orders on other instruments when the same signal and risk checks pass. The commercial evidence and recommended use remain XAUUSD only; other instruments and arbitrary broker specifications are not performance-validated. Check the chart symbol before enabling entries.
- Check volume steps, minimum lot, margin, quotes and stop constraints. A correct operating-system UTC clock is required. Historical testing uses the disclosed NY+7/EET-US time convention; arbitrary broker-history timezones are not automatically inferred.
- Keep the terminal or VPS connected and trading permissions available for EA-managed actions. Use an exclusive Magic range and do not run a second instance in the same account/Magic context.
- Dependencies are embedded; no customer-side indicator installation or DLL is required.
Inputs and operational limits
| Input | Default | Operation and limits |
|---|---|---|
| RiskProfile | Standard | Conservative, Standard, High, or Custom; scales the fixed construction. |
| PortfolioRiskScale | 100 | Percentage multiplier of the chosen profile; increasing it can increase losses and margin needs. |
| CustomPortfolioScalar | 3 | Common scalar used only with Custom. It is not a per-trade risk percentage. |
| UseCompounding | false | False uses the recorded starting-capital basis; true uses current account equity. |
| TradingEnabled | true | Permits new entries when checks pass. False does not cancel existing pending orders or disable owned exit management. |
| MagicBase | 2026090800 | Reserves MagicBase+1 through MagicBase+20 exclusively. Do not change it with open exposure. |
| EmergencyStop | false | Blocks entries and requests owned pending cancellation. Open positions retain their frozen exits; it is not a flatten-all command. |
Accepted pending orders and hard stop-loss/take-profit orders remain broker-side outside new-entry windows. Friday cleanup depends on connectivity, trading permission and broker acceptance. Stop losses do not guarantee exit price or maximum realized loss. Turning terminal Algo Trading off can prevent EA-generated requests while accepted broker-side orders remain active.
What the evidence does not establish
The model is 1 Minute OHLC, reported history quality 78%, not real ticks. Intraminute fill chronology and floating drawdown are model-dependent. The nominal spread specification is 0.250 in USD price units, commission is zero, and native tester swap/fills apply. Fixed adverse slippage of 0.020 per side and exact research 23:00 rollover parity are not execution-proven. A cash-cost overlay is not proof of worse live fills.
No Forward return is claimed. Historical testing does not establish future profitability, independence between modules, broker universality, or a maximum future loss. A controlled terminal-process restart test on R08R2 recovered representative exposure. R08R3 removed only non-trading history-export calls. R08R6 removed instrument-name refusals while retaining the same risk and module logic on the current chart symbol. R08R7 adds a broker pending-order capacity check before new pending-order submissions; exit management, risk sizing and minimum-lot skip are unchanged. That demo restart test was not rerun on R08R3 or R08R7 and was not an OS/VPS power-loss test.
Installation and support
Install the purchased Market version through MetaTrader's Market workflow. Attach it with TradingEnabled=false while checking the account, symbol, history, clock, profile and Magic range. Enable entries only when you intend to allow orders and the Journal shows readiness. A demo setup check does not promise live results.
Retain terminal state across supported restarts. Before updates or VPS migration, inspect open exposure and ensure the previous instance will not manage the same account/Magic context. Follow release-specific migration instructions rather than assuming every update can occur with positions open.
Technical support is available through this product's comments or MQL5 messaging. Provide the product version, sanitized Journal messages and relevant symbol specifications. Never share credentials or unnecessary account identifiers. Support is not personalized investment advice.
Risk disclosure: leveraged trading can cause substantial losses. Historical simulated performance does not guarantee future results. Only select settings after deciding what loss you can afford; stop orders and historical drawdown are not guaranteed loss limits.
How to read the evidence gallery
The first image identifies the product; it is illustration, not a trading result. Images 3 and 4 contain actual English MT5 Strategy Tester screenshots: the Combined Backtest report and the native Balance / Equity graph. Only the surrounding brand frame and explanatory captions were added; report values and graph pixels were not redrawn. Period and risk panels are explanatory summaries. These are historical simulations, not live or forward results.
No product-linked live or demo-forward performance is presented here. A technical demo setup or restart test is not a Forward return record. Evaluate historical results together with modeling, costs and operating limits.
Frequently asked questions
Is this the Dollar Yen product with a different symbol?
No. Gold uses twenty fixed modules with H1/H4 history; Dollar Yen uses nine fixed H2 modules and its own production evidence. Their joint account performance has not been validated.
Does this use Martingale or grid recovery?
Position size comes from the configured capital basis, fixed module allocation and stop distance, not a ladder that increases lots to recover previous losses. Modules submit their own signal-driven trades; simultaneous module positions are possible and should not be mistaken for independent account risk.
Why is a hedging account required?
Separate module positions and their exits must retain separate ownership. Hedged account accounting does not itself promise a hedge against losses. Netting is not supported.
What should I do before enabling entries?
Install through MT5 Market, select a broker XAUUSD H1 chart, synchronize H1/H4 history, check the UTC clock and broker specifications, and reserve MagicBase+1 through MagicBase+20. Attach with TradingEnabled=false; confirm the intended risk profile and Journal readiness before enabling new orders.
Does disabling TradingEnabled cancel orders?
No. Existing pending orders can fill and owned exit management continues. EmergencyStop blocks entries and requests owned pending cancellation; open positions retain their frozen exits. It is not a flatten-all command.
Is USD 50,000 a minimum deposit?
No. It is the Standard historical reference. Broker minimum lots and contract specifications affect participation. Do not raise risk just to force a trade that the sizing checks skip.
Why might my backtest differ?
The reference uses custom Gold history, 1 Minute OHLC, NY+7/EET-US time conventions and zero commission. Ordinary broker history, spreads, swaps and intrabar modeling can differ. The reported 78% history quality must not be represented as a real-tick test.
Where is the live track record?
No verified product-linked live or demo-forward return record is presented. These historical results do not establish future returns. A future monitoring record must disclose account type, start date, version, settings and all losses.
Version notes
Trading version 1.005 is displayed by MQL5 as 1.5. This page revision changes presentation only. The released EX5, strategy membership, construction, parameters and risk design are unchanged.
