Overview
Goldwave Ultra EA by Shengzu Zhong is a gold EA for MetaTrader 5 that, per the vendor, uses AI-assisted, adaptive logic to trade only selected setups, without grid and without martingale. As evidence the vendor links a single MQL5 signal. It has run for a long time and mostly closes its trades in profit, but open losses were at times clear. It is older than the product, and it made most of its growth before the market launch. On top of that come a price ladder with a scarcity note and buyer reports that differ from the signal. Our grade is provisional.
What the vendor sells
Per the product page (read 17 Nov 2026):
- Market and chart: XAUUSD only. The vendor recommends the M1 chart but writes that M1 only serves more precise monitoring: the strategy is “not simply an M1 scalping system”.
- Logic: The product page speaks of “AI-assisted adaptive decision logic” that judges whether the market environment suits trading, plus market and news filters that fetch data via WebRequest. What the AI does exactly, with which data and by which rules, the page does not describe. This is a vendor claim without verifiable evidence; we can neither confirm nor refute it.
- No grid, no martingale: In the FAQ the vendor writes: “GoldWave Ultra does not use Grid, Martingale, or averaging-down strategies.” Normally only one position is open, each with an adaptive take-profit and stop-loss. At the top of the page the wording is more cautious: “No excessive grid, no dangerous martingale, no averaging down.” We have not seen the source code.
- Platform: Per the FAQ, Goldwave Ultra supports MT4 and MT5. On 17 Nov 2026, however, only the MT5 version is on the MQL5 market; we found no MT4 version among the vendor's products.
- Price: $699, ten activations, free demo. Published on 1 Sep 2026, current version 5.9 of 15 Nov 2026 (new: a trading pause per weekday).
- Account: $500 or more recommended, an ECN or raw-spread broker and WebRequest enabled in the terminal. The vendor sends the manual and recommended settings only by private message after purchase.
Two statements on the page are mainly sales arguments. First the price ladder: the current price is an early tier, it is to rise step by step, “with a target upper price of USD 1899 during this year”. Second the scarcity: “This EA is currently released in strictly limited quantities — only 2 copies remain available at USD 699” (both read on 17 Nov 2026, vendor statements). Such notes say nothing about the quality of an EA. They add pressure to buy quickly instead of testing the demo first. The vendor's product image also advertises “MYFXBOOK VERIFIED” (read 17 Nov 2026); the product page links no Myfxbook account.
The product page is candid about the backtest: because parts of the logic use WebRequest, the strategy tester is not representative, and demo and backtest “should not be used as the primary benchmark” (read 17 Nov 2026). That also means: if you want to check before buying, you can hardly rely on the demo. That leaves the signal.
The signal
The vendor links one public MQL5 signal, “GoldWave signal” (2339082). Per the product page, Goldwave Ultra uses “the same core trading logic and execution framework as the verified live signal” (read 17 Nov 2026). The signal carries the name of the predecessor: one buyer writes that he had already bought the “original GoldWave EA” (review of 10 Oct 2026). The vendor no longer lists it on the market.
Real money or demo: The MQL5 rules only allow real-money accounts for signals, and MQL5 shows the “Real Account” icon on the signal. Which EA version and settings ran there and when is unknown. We do not run an account with Goldwave Ultra in our own EA park.
What we read on 17 November 2026, without figures, as in all our reviews (methodology):
- Older than the product: The signal started on 21 Oct 2025; per the monthly history the account was already trading from May 2025. Goldwave Ultra only came to the market on 1 Sep 2026.
- Growth mainly before launch: By far the largest part of the growth came before the market launch, a large share of it even before the official signal start, mainly in two months in autumn 2025. In that phase MQL5 twice set the warning that too much growth in the last month indicates high risk; both were removed again. Since the launch the account has grown much more slowly.
- Small account, fixed lot size: The account started with an amount in the double-digit dollar range and traded the smallest lot size for a long time; on 20 Oct 2026 the vendor raised it. With a fixed lot size on a growing account, the monthly growth in percent shrinks, and declines also look smaller in percent than they would with a lot size that grows with the account. A subscriber pointed this out in October 2026. There is also a deposit and a withdrawal of the same amount, which make the curve harder to read.
- Decline: What counts is the worst of the maximum and relative decline by balance and the relative decline by equity. Here it is the decline by equity, and it is clear. The closed declines by balance, by contrast, are moderate: open positions at times went noticeably into the red before they were closed.
- Many small wins, few larger losses: The win rate is very high, but an average loss costs several times an average win. Together with the deeper decline by equity this forms a familiar pattern: losing positions tend to be sat out rather than closed early at a stop, and so often still end in profit. Such systems look smooth on the balance curve for a long time; a few losses in a row can wipe out many wins.
- Leverage and margin: The account runs at 1:500 leverage, although the signal description itself states “Leverage: 1:200 recommended” (read 17 Nov 2026). The maximum deposit load is moderate. With EU leverage of at most 1:20 on gold, the same position ties up many times the margin. A position of the same size relative to the account would not be possible; anyone copying needs much more capital or smaller lots.
- Holding time: Positions are open for hours on average, and the account trades on only a small share of days. That is not a scalper in the narrow sense. Twice MQL5 reported several days without trading.
- Not only gold: Next to the gold deals there are a few deals on a Chinese stock index. Whether they come from the EA, the page does not show.
On the signal page itself, some subscribers report problems with copying, for example when the broker lists gold with a suffix, and slippage. One subscriber wrote on 25 Jul 2026 that a buy and a sell position were open at the same time at the weekend, although the signal describes one position at a time. Another reported an EA error on 17 Jul 2026; the developer explained it and announced stricter daily limits (review of 18 Jul 2026). Another subscriber reported on 31 Jul 2026 a net loss in his first month of copying the signal at a different broker, a large open position in the red and a very wide stop-loss. We cannot verify these reports. How to read signal metrics and why we use the worst value from balance and equity for the grade is explained in the guide Trading metrics.
What buyers report
On the MQL5 market Goldwave Ultra stands at 4.14 of 5 from 9 reviews, plus 2 comments (as of 17 Nov 2026). Buyer reviews do not count towards our grade. We report them because they show what buyers should watch for.
The positive reviews mainly praise that the EA works without grid and martingale, waits patiently for setups, and the developer's support. One buyer writes that the EA is profitable for him at all brokers, including prop-firm accounts (comment of 8 Nov 2026); we cannot verify this.
Critical reports, without judgement on our part:
- Few trades, losses. A buyer wrote on 26 Oct 2026 that he had set up the EA as the vendor instructed. It trades about once a week, often not at all, and the trades it opened lost money. The developer tried to help, but nothing changed about the rare and losing trades. He advises against buying. The developer replied on 14 Nov 2026 that he had helped with the setup; after a fresh installation the EA opened a trade, and the buyer himself reported a profit to him. The review was therefore misleading. The EA trades selectively on purpose, losses are always possible, and the logic matches the signal.
- No trade live. A buyer wrote on 20 Sep 2026 that the backtest showed trades for the previous week, his live account not a single one. The developer replied the same day: he does not know who the buyers are unless they get in touch, so he could not send the manual and settings; differences in price feed, spread, execution or settings could explain the gap. He asked for the review to be removed or changed temporarily until this was checked.
How far backtest, signal and your own account can diverge on gold is shown in Gold EA: backtest vs. live.
Relation to other gold EAs
Goldwave Ultra trades only gold. If you already run another gold EA, such as The Gold Reaper, you stack the same bet. How that affects the combined drawdown is worked out in the guide EA portfolio management. Why the design matters so much on gold is explained in Gold EA risk: grid, martingale and stop-loss. An overview of reviewed gold EAs is in Best gold EA 2026.
How the provisional grade comes about
We grade Goldwave Ultra on the same six criteria as every EA (calculation) and deliberately go lower where the data formula would be too kind:
- Risk control: What counts is the worst value from the declines by balance and by equity, here the decline by equity; it sits in a middle band of the table in the calculation. We deduct further because an average loss costs several times an average win, open losses are sat out, the stop size is not disclosed and the fixed lot size on the grown small account makes the decline look smaller in percent.
- Track record: A public signal of more than a year would score highly under the formula. We lower it clearly: a single account chosen by the vendor, older than the product, with an undocumented EA version, very small starting capital, deposits and withdrawals and deals outside gold.
- Transparency: The public signal counts positively. Deductions because the logic is only described as AI-assisted and not disclosed, the backtest is not meaningful per the vendor, settings only come after purchase and the MT4 and Myfxbook statements are not backed on the market.
- Price and value: $699 sits in the band between $500 and $1,000. If the price rises as announced, we will re-grade.
- Usability: neutral value; we have not tested the installation ourselves.
- Broker independence: documented at one broker, through a single account. The vendor himself names broker differences as a possible reason for differing results, and buyer reports differ from the signal.
As with every EA, an AI from a different model family also grades Goldwave Ultra blind; where it differs strongly, the mean applies until a human has re-checked. For Goldwave Ultra that is the case for risk control, transparency and broker independence: the second review was stricter each time, and the mean of our sub-grade and its sub-grade applies there.
We will re-check the grade once it is clear how Goldwave Ultra runs on buyers' accounts over several months, the price changes or a larger decline occurs.
Who it does (not) suit
If you want to check it: Because the backtest says little per the vendor, first follow the signal for a few weeks and put your own trades next to it. Then a separate small account at a raw-spread broker with a VPS, the recommended settings instead of increased risk, and only money you can afford to lose. The scarcity note is no reason to buy without testing. Why a gold EA needs a VPS is covered in VPS for EAs; which brokers offer tight spreads on gold is compared in best broker for gold EAs.
Not suitable for anyone who wants an EA with disclosed logic or a meaningful backtest, for MT4 users without confirmation from the vendor, as a second gold EA next to another gold system, and for anyone who extrapolates the signal curve. For prop-firm accounts there is no evidence of how deep declines go with a lot size that grows with the account; basics: Gold EA at prop firms.
Verdict
Goldwave Ultra promises a selective gold EA without grid and martingale, and the linked signal does mostly close its trades in profit, with moderate closed declines so far. Open losses, however, were at times clear. Against that stand weighty reservations: the signal is older than the product and made most of its growth before the market launch, it runs on a small account with high leverage, the AI logic is a vendor claim without evidence, and buyers report rare and losing trades. The price ladder and the scarcity note change nothing about this. Our grade is therefore provisional. We grade Goldwave Ultra with the same formula as every EA. We earn nothing from the EA itself; we link the vendor and MQL5 only as sources, never as partner links. Links marked “Ad” on this page (VPS, ATS-listed alternatives) are partner links. Past results are no guarantee of future ones, and trading leveraged products can lead to the loss of the capital invested.
