Deep Genesis AI is an XAUUSD Expert Advisor that uses DeepSeek to assess gold entries against H4, H1 and M15 market structure. Its distinctive controls include AI confidence gating, volatility-based trailing, pending-order routing and a daily loss threshold. The important limitation is that MT5 backtests do not fully reproduce the live AI decision layer.
Our assessment: suited to traders who want to evaluate a cloud-assisted gold workflow and are prepared to manage its API dependency. The documented entry and risk controls are more informative than the source’s broad institutional-trading claims. Public promotional graphics are not a substitute for a forward record with account history and drawdown.
Ready to Buy Deep Genesis AI?
$399.99 with crypto · $499.99 by card
Official version · Instant download · Free lifetime updates · 7-day money-back guarantee
How the trading decision is assembled
The developer describes a top-down process: H4 provides the structural direction, H1 checks alignment and momentum, and M15 supplies the pullback entry context. The EA is attached to H1; the other periods are read in the background.
Fair value gaps, order blocks and liquidity sweeps contribute to the information supplied to DeepSeek. Those terms describe the developer’s chosen analysis rules. They do not demonstrate that the program observes institutional order flow or knows where every trader’s stops are located.
The AI returns a confidence assessment. The overview cites 75, but the later release history describes adaptive thresholds and a conditional downgrade into a limit order. It is therefore too simple to treat 75 as an unconditional floor for every order path. Neither that number nor the illustrated scores is a statistically validated chance of profit. Preserve the original proposal, returned score, any routing change and final outcome in a forward log.
Market regimes and order routing
A Vortex-based filter is described as distinguishing trends from sideways conditions. In ranges the source favors limit orders at the edges rather than immediate market entries. Normal and intraday modes offer different intended holding and stop profiles, while the order setting can leave execution to AUTO or restrict it to market or pending orders.
The developer also describes converting a requested stop order to a limit order if a price move makes the original request invalid. That addresses a specific order-placement problem, but it changes the proposed entry’s relationship to the current price. Test order expiry, fill behavior and the handling of repeated proposals with your broker rather than assuming every routed order will obtain the original analysis price.
Release notes add important execution detail
The July 22, 2026 update describes expanded market context, additional stop/target rules, re-entry protection and stricter validation, alongside numerous parsing and risk-management fixes. It also changes the default mode to NORMAL. These are functional revisions, so an earlier screenshot or test cannot automatically represent the later decision process.
On August 11, the developer raised the MARKET validation requirement from six to seven of nine checks. The same update narrowed the confidence-downgrade window to three points below the adaptive threshold, with a floor of 62, and added BUY/SELL directional confluence before a rejected MARKET or STOP proposal could be repackaged as a LIMIT order. These details matter more than a single headline confidence percentage.
The September 25 release entry supplies no meaningful technical explanation. We cannot infer further behavior changes from that note. Confirm the supplied inputs and test the routing rules rather than treating the latest version label alone as performance evidence.
What the risk controls do
- Position sizing
- Fixed lots or a risk-percentage calculation. Check the actual stop distance and gold contract specification.
- ATR trailing
- A volatility-sensitive trailing distance; wider conditions can mean a wider stop, not automatically lower cash exposure.
- Daily loss threshold
- Described as stopping the EA until the next trading day once the configured threshold is reached. Confirm how it treats existing positions and the reset time.
- News windows
- Separate high- and medium-impact blocking windows around economic events.
- Support/resistance filter
- Checks major daily and weekly highs and lows before accepting a proposed entry.
These controls deserve testing individually. The overview’s daily-loss description does not fully specify whether the threshold uses balance or equity, how other EAs on the account are included, or whether existing positions are liquidated. Resolve those details from the supplied build and a demo test before treating the setting as a firm account-level loss cap.
Stops and daily thresholds can reduce exposure, but cannot promise a fixed final loss under gaps, slippage or disconnections. Similarly, news avoidance changes entry timing; it does not make gold insensitive to unscheduled events.
DeepSeek access and practical setup
The listed setup is XAUUSD on H1 with a low-spread broker. Enable Algo Trading and whitelist https://api.deepseek.com under MT5’s Expert Advisors settings. Use a valid personal API key and a provider account with available balance. AI requests require a working internet connection and separately billed provider usage.
The source’s small-account recommendation is not a risk guarantee. Broker minimum volume and contract size may make a desired percentage-risk setting impossible on a small balance. Check the calculated lot and stop risk before accepting it, and allow for other open trades.
During forward testing, record provider errors and response delays as well as trades. A good-looking historical chart does not show whether a live request arrived too late or whether an API balance or rate limit prevented analysis.
How much evidence do the screenshots provide?
The developer explicitly states that Strategy Tester cannot fully reproduce the AI selection process. Backtests are therefore partial technical references, even where execution logic is simulated successfully. They cannot validate the complete strategy that runs with live cloud responses.
The additional backtest-summary graphic claims a 2.15 profit factor, yet its displayed gross profit divided by absolute gross loss gives approximately 2.97. That arithmetic mismatch prevents treating the panel as a consistent test record. A raw report, input file, broker data and a clear account of how the AI component was handled would be needed to investigate it. The source’s own warning about incomplete AI backtesting still applies.
Evaluate a dated forward record with the exact mode, model and inputs you intend to use. Relevant observations include accepted versus skipped proposals, net returns after costs, floating equity loss, the number of simultaneous trades and how the system behaves when market conditions change. A single strong interval or promotional performance panel cannot establish repeatability.
No public customer review was present in the retrieved MQL5 product review section on September 30, 2026. The reviewed overview also did not link a verifiable live signal. That does not prove poor performance; it limits what can be concluded from the available public evidence. No third-party star score has been imported into CheaperForex ratings.
Deep Genesis AI or Unlocked AI?
Both use an external DeepSeek service for gold analysis, but their documented architecture is different. Unlocked AI explicitly separates a local H1 breakout engine from cloud decisions and selects one of three workflows from the chart timeframe. Deep Genesis emphasizes one top-down H4/H1/M15 decision framework with regime and key-level filters.
The choice is about the behavior you can evaluate: separate engine attribution is useful with Unlocked, while Deep Genesis needs close examination of proposal filtering, routed pending orders and its account limits. There is no controlled comparison here establishing that either produces better returns.
Sources and evidence scope
This is an editorial assessment of the developer’s public listing, update history and original gallery material checked September 30, 2026. CheaperForex has not independently verified a live performance record for this review. Strategy descriptions are attributed to the developer; practical testing observations above explain what remains to be established.
Deep Genesis AI developer listing · Release history
