Expert Advisor Review

Perceptrader AI Review MT4/MT5

Not Recommended MT4/MT5 AI-Filtered Grid, 6 Pairs From $1,497/yr Re-verified 27 Sep 2026

The Bottom Line

The Robotic Trader's verdict: Perceptrader AI is Not Recommended. The developer is legitimate and the account she features is genuinely strong, but two of her other real Perceptrader accounts fell 65.44% and 68.64%, and nothing published tells a buyer which settings produce which result.

We changed this verdict from Recommended to Not Recommended on 27 September 2026. Our original review rested on one account: the one the developer features, which shows a 21.96% maximum drawdown across its history on Myfxbook. When we reviewed her Waka Waka EA, we checked every Waka account she publishes. Applying the same check here, we found four real Perceptrader AI accounts on her Myfxbook profile, not one.

Two of them, including the featured account, show maximum drawdowns of 21.96% and 23.14%. The other two reached 65.44% and 68.64%. Two of the four started trading on the same day, 28 May 2023, and Myfxbook's equity charts put the deepest point of both on the same day, 29 September 2023 — two months after Perceptrader AI went on sale on the MQL5 Market, and the day Myfxbook records a deposit into the 65.44% account. One of that pair ended with a 65.44% maximum drawdown, the other with 23.14%. Same market, very different outcomes: the result depends on how the EA is set up.

That is the problem for a buyer. The accounts are labelled by trading mode, "Unique" or "Standard", but mode is not a risk level, and none of the four states which risk setting it runs. We cannot point you to the configuration that produced the 21.96% record, so we cannot stand behind a purchase. We treat losses of two-thirds of an account on the developer's own real accounts as disqualifying, as we did in our Waka Waka EA review.

What counts in its favour: the developer publishes all four accounts, including the two that hurt her case; the featured record is long, real and profitable, with 6,683 trades and a 1.50 profit factor on MQL5 as at 9–10 September 2026; and the grid is disclosed and comes with real risk controls. This is a verdict about the product's risk, not the developer's honesty. If you run it anyway, read how grid systems accumulate exposure first, and size capital and lot settings on the assumption that a drawdown of 65–69% can happen. What would change our verdict is set out in the disclosures below.

Methodology
ⓘ

Evaluation Methodology

This review was conducted under our Track 1 framework, which applies where an Expert Advisor has a genuine, independently tracked live record. The MQL5 Market listing links to the developer's MQL5 signal, which tracks the same account as her featured Myfxbook account: balance, trade count and profit factor match, so we treat the two as one record shown twice, not as corroboration.

On 27 September 2026 we re-assessed the review against the developer's full Myfxbook profile, opening all four real Perceptrader AI accounts listed there, and read the MQL5 listing and the developer's product page for pricing, risk settings and marketing claims. We also checked ForexPeaceArmy and Trustpilot and found no coverage of this EA. For how our verdicts are reached, read How We Test Expert Advisors.

Note: figures are dated snapshots read from public, independently hosted MQL5 and Myfxbook pages — not self-reported figures and not a live feed. The featured account's detailed figures are as verified on 9–10 September 2026; the four-account comparison was read on 27 September 2026. We did not buy the EA for this review.

The Developer's Four Real Accounts

Every Real Perceptrader AI Account on the Developer's Myfxbook Profile

All four are real-money MT4 accounts at ICE-FX, read on 27 September 2026. The names are the developer's own; "Unique" and "Standard" are trading modes, not risk levels, and no account states which risk setting it runs. Each name links to the account.

Unique Mode, No Limits21.96% maximum drawdown — the account the developer features; 6,720 trades, profit factor 1.50, 1:500; $5,000 deposited, $8,000 withdrawn; trading since August 2019
Standard Mode, 5 Pairs with AI Filter23.14% maximum drawdown; 2,049 trades, profit factor 1.68, 1:300; $400 deposited; trading since 28 May 2023
Unique Mode, 5 Pairs with AI Filter65.44% maximum drawdown, deepest on 29 September 2023, with a deposit recorded the same day; 1,298 trades, profit factor 1.80, 1:300; $700 deposited in total; trading since 28 May 2023
Standard Mode, No Limits68.64% maximum drawdown, deepest in August 2021 — almost two years before Perceptrader AI was published on the MQL5 Market on 24 July 2023 — with a further deposit on 3 September 2021; 12,513 trades, profit factor 1.50, 1:500; $15,000 deposited, $21,101 withdrawn
What it showsBoth high-drawdown accounts have since recovered and are in profit. But one Perceptrader AI configuration lost almost two-thirds of its equity two months after the product went on sale, and a buyer has no published way to tell which configuration produces the 21.96% record.
Two figures we will not stateThe 23.14% account shows a highest balance of $58,049.63 against $400 deposited and a $1,593.00 balance, which cannot all be right, so we derive nothing from it. The dates of the deepest points come from Myfxbook's equity charts; the drawdown figures are Myfxbook's own.

The Featured Account

Verified Snapshot

Perceptrader AI — the developer's featured account

Verified: 9–10 September 2026 — a dated snapshot, not real-time. Previously recorded 13 August 2026. Use the links below for current figures.
6,683
Total Trades
65.58%
Win Rate
1.50
Profit Factor
17.06%
Max Drawdown — by balance, MQL5, Apr 2023 – Sep 2026
17.94%
Max Drawdown — by equity, MQL5, Apr 2023 – Sep 2026
21.96%
Max Drawdown — full history, Myfxbook
12.82
Recovery Factor
0.05
Sharpe Ratio
Monthly growth0.63% (MQL5)
Annual forecast7.62% (MQL5 projection from recent performance, not a measurement)
Expected payoff per trade2.12 USD
Currency pairs tradedGBPCHF, NZDCAD, AUDCAD, NZDUSD, USDCAD, AUDNZD
Trades per week18
Average holding time19 hours
Best / worst trade+899.61 USD / −264.93 USD
Max consecutive wins / losses20 / 7
Broker and leverageICEMarkets-Server, 1:300 on MQL5 (changed once within the 1:300–1:500 range); Myfxbook lists 1:500
Algorithmic trading99% of trades
Signal monitored since8 April 2023 — account recorded by MQL5 as 371 weeks old
Account funding5,000 USD initial deposit; 8,000 USD withdrawn to date
Slippage dataNot published on this signal
View Live MQL5 Signal → View Live Myfxbook Account →
MQL5 and Myfxbook track the same trading account here — balance, trade count and profit factor match — so we present them as one record viewed two ways. The two drawdown periods differ because MQL5 has only monitored the signal since April 2023, while Myfxbook covers the longer history. We no longer publish a growth figure for this account: with $5,000 deposited and $8,000 withdrawn, percentage growth measures the funding pattern rather than the strategy. We report the Sharpe ratio for completeness but do not use it to decide a verdict. Past performance, including verified live performance, is not a guarantee of future results.

Price and Licence Terms

Developer's siteSold only inside a six-EA portfolio: $1,497 a year to rent, or $4,997 once to own; 10 MT4 or MT5 licences per EA
Free trial / refund14-day free trial; 30-day money-back guarantee on the portfolio (both as stated on the developer's site)
MQL5 Market$2,800 on its own — MT4 version, with an MT5 version; 10 activations; free demo for the Strategy Tester; published 24 July 2023
PlatformMT4 and MT5; M5; one chart trades all six pairs; VPS required by the developer
Risk settingsThe developer's site lists "3 risk sets"; the MQL5 listing describes "4 predefined presets". Fixed, balance- or equity-based, and deposit-load lot sizing are also available, with optional maximum-drawdown and daily-drawdown closes
Leverage guidanceThe developer advises no higher than low-risk settings on an account under $6,000 at 1:30 leverage, and up to significant-risk settings on $1,000 at 1:100
Prices recorded27 September 2026, from the developer's site and the MQL5 Market. On 10 September 2026 the developer's site listed it alone at $1,960, discounted from $2,800

Key Evaluation Criteria

  • ✅Evidence authenticity: passes. All four accounts are real, public and long-running, and the developer publishes the two that weaken her case. A deleted Myfxbook listing for her Waka Waka EA proved, on archived copies, to be the same account re-listed.
  • ✅Strategy disclosure: passes. The grid is declared on the MQL5 listing and the developer's site, with its risk settings documented.
  • ❌Drawdown on the developer's own accounts: fails. 65.44% and 68.64% on two real Perceptrader AI accounts. Our methodology treats a near-wipeout as disqualifying.
  • ❌Configuration transparency: fails. The same product produced 21.96% and 65.44% drawdowns, and no account states its risk setting, so a buyer cannot tell which result they are buying.
  • ⚠️Marketed figures: the developer's site features only the 21.96% account, and shows "42/48" months in profit, while the MQL5 listing claims "80 consecutive months in profit". The two claims do not reconcile.
  • ⚠️Recent return rate: monthly growth of 0.63% and an annual forecast of 7.62% on the featured account (MQL5, 9–10 September 2026), against a four-figure purchase price.
  • ⚠️Independent corroboration: none found on ForexPeaceArmy or Trustpilot, and the MQL5 and Myfxbook records are the same account.
  • ⚠️Marketing tactics: recurring scarcity and urgency pricing across promotional history, and one unsourced testimonial.

Suitability & Risks

What's Genuinely Positive

To be fair to the product

The developer is legitimate, and she publishes her record in full, including the accounts that weaken her case.

  • ✅A featured real-money record running since August 2019, with 6,683 trades on MQL5 and a 1.50 profit factor.
  • ✅All four real accounts are public, and both high-drawdown accounts have since recovered.
  • ✅The grid is declared, with real risk controls: presets, lot caps and optional drawdown closes.
  • ✅A 14-day free trial and a 30-day money-back guarantee through the developer's site.
Key Risks

Why we will not recommend it

Every item here is sourced to the developer's own accounts or listings.

  • ⚠️Real-account drawdowns of 21.96%, 23.14%, 65.44% and 68.64% across four Perceptrader AI accounts.
  • ⚠️No published way to know which settings produce the lower figures.
  • ⚠️Recovering from a 65.44% drawdown takes a 189.4% gain — try the arithmetic in our drawdown recovery calculator.
  • ⚠️A recent return rate of 0.63% a month against a four-figure price.
  • ⚠️If you run it, size capital and lot settings on the assumption that a drawdown of 65–69% can happen.

Important Disclosures

Verification History
13 August 2026 — initial assessment, Recommended. 6,626 trades, 65.63% win rate, 1.51 profit factor, 17.06% maximum drawdown, 12.77 recovery factor.
9–10 September 2026 — re-verified against both platforms, Recommended. 6,683 trades, 65.58% win rate, 1.50 profit factor, 17.06% drawdown by balance and 17.94% by equity, 12.82 recovery factor, 0.63% monthly growth, 7.62% annual forecast. Myfxbook confirmed 21.96% maximum drawdown across the full account history. This re-verification also corrected two errors: the 17.06% drawdown had been presented alongside a seven-year track record, when it covers only the period monitored since April 2023; and the MQL5 and Myfxbook records had been described as independent corroboration, when they track the same account.
27 September 2026 — pricing updated and one finding corrected, Recommended. Portfolio-only pricing recorded; the deleted Waka Waka EA Myfxbook listing established from archived copies as the same account re-listed; two statements we could not source removed.
27 September 2026 — verdict changed to Not Recommended. Re-assessed against all four real Perceptrader AI accounts on the developer's Myfxbook profile, applying the same full-account check we used for her Waka Waka EA. Two show maximum drawdowns of 65.44% and 68.64%, and none states its risk setting. We also stopped publishing a growth figure for the featured account, because its deposits and withdrawals distort it; removed a statement that the developer trades her strategies privately before publishing them, which we could not source; removed the affiliate links from this page; and stood down our own account running this EA. Featured-account figures were not re-checked on this date.
We do not edit published figures to track current values — we add dated entries here instead.
What would change this verdict. So that this does not rest on our judgement alone, these are the conditions under which we would reassess Perceptrader AI:
  • The developer stating which risk setting each of her real Perceptrader AI accounts runs.
  • A real account on a named configuration a buyer can select reaching 24 months with a maximum equity drawdown below 30%.
  • The product page showing every real account's drawdown beside the featured one.
The first two are required; the third would strengthen the case. None requires the product to stop being a grid system.
On the two platforms. The featured account appears on both a public MQL5 signal and a public Myfxbook account. They track the same trading account, so we treat them as one record, not two independent sources. MQL5 lists ICEMarkets-Server at 1:300, with one recorded leverage change within the 1:300–1:500 range, while Myfxbook lists 1:500; we flag the difference rather than resolve it.
Marketing findings. The developer's promotional history shows recurring scarcity and urgency pricing, and we identified one testimonial attributed to an unnamed, unverifiable industry source. Her site shows "42/48" months in profit for the featured account while the MQL5 listing claims "80 consecutive months in profit"; the two do not reconcile. None of this reflects on whether the accounts are real — they are — but each is disclosed.
On the replaced Waka Waka listing. A Myfxbook listing for another of this developer's products, Waka Waka EA, was found deleted. Archived copies (last captured 16 August 2025) show it tracked the same account the developer now publishes under a new Myfxbook listing. The record was re-listed, not removed.
We ran this EA ourselves, and have stopped. We ran Perceptrader AI on our own real account from 19 August 2026 and stood it down on 27 September 2026, after this re-assessment. We will not run an Expert Advisor we tell readers not to buy. That account was far too short to count as evidence and was never part of the verdict.
What we verified, and what we did not.

We read the MQL5 signal and the featured Myfxbook account on 9–10 September 2026, and all four real Perceptrader AI accounts, the MQL5 listing and the developer's product page on 27 September 2026. We did not buy the EA for this review, inspect its default settings, or re-verify trade-by-trade execution or broker conditions. The dates of the two deepest drawdowns come from Myfxbook's equity charts; every drawdown figure is Myfxbook's own.

Affiliate Disclosure: The Robotic Trader is an affiliate partner of ValeryTrading, which sells this developer's Expert Advisors. This review was first published on 13 August 2026 and the partnership was approved on 17 August 2026, after the original verdict was reached. The programme covers the portfolio Perceptrader AI is sold in, so a Recommended verdict could earn us commission. We have reached Not Recommended, and no link on this page is an affiliate link.
Risk warning. Trading foreign exchange on margin carries a high level of risk and you can lose more than you expect. An Expert Advisor removes the delay between a signal and a trade; it does not remove risk. Past performance, including verified live performance, is not a reliable indicator of future results. The Robotic Trader is an independent publisher, not a regulated financial adviser, and nothing on this page is investment advice.

Frequently Asked Questions

Is Perceptrader AI legit?
Yes. The developer, Valeriia Mishchenko, publishes real-money accounts on MQL5 and Myfxbook, including the ones that weaken her case, and states openly that Perceptrader AI is a grid system. Our Not Recommended verdict is about risk, not honesty: two of her four real Perceptrader AI accounts fell 65.44% and 68.64%, and nothing published tells a buyer which settings produce which result.
Why is Perceptrader AI Not Recommended?
Because the developer's own real accounts show that the same product can lose around two-thirds of an account. Of four real Perceptrader AI accounts on her Myfxbook profile, two show maximum drawdowns of 21.96% and 23.14%, and two show 65.44% and 68.64%. The 65.44% account hit its low two months after the product went on sale, with a deposit recorded the same day. None of the accounts states its risk setting, so a buyer cannot choose the configuration behind the better record. Our methodology treats a near-wipeout as disqualifying.
Why did you change the verdict from Recommended?
Our original review, published on 13 August 2026, rested on the one account the developer features. When we reviewed her Waka Waka EA in September 2026 we checked every account she publishes for it, and applying the same standard here meant checking all four real Perceptrader AI accounts. Two of them reached drawdowns we treat as disqualifying, so on 27 September 2026 we changed the verdict to Not Recommended. The earlier review is recorded in the verification history above.
What is the maximum drawdown of Perceptrader AI?
There is no single answer. The featured account shows 17.06% by balance and 17.94% by equity on MQL5 (April 2023 – September 2026) and 21.96% across its full Myfxbook history. The developer's other three real accounts show 23.14%, 65.44% and 68.64% (Myfxbook, read 27 September 2026). Your own figure would depend on settings the developer does not publish for any of the four.
Which settings give the lower drawdown?
We cannot tell you, and neither can the published record. The accounts are named by trading mode, "Unique" or "Standard", but mode controls whether each copy of the EA takes its own trades, not how much risk it takes. None of the four accounts states its risk preset, and the developer's site and the MQL5 listing even disagree on how many presets there are (three and four).
What return should I expect?
On the featured account, MQL5 reported monthly growth of 0.63% and an annual forecast of 7.62% as at 9–10 September 2026. The forecast is a projection from recent performance, not a measurement, but against a four-figure price it matters: on a modest account the purchase price could take a long time to recover. We do not publish a growth figure for this account, because deposits and withdrawals make percentage growth an unreliable measure.
How much does Perceptrader AI cost?
Checked on 27 September 2026, the developer sells it only inside a six-EA portfolio: $1,497 a year to rent or $4,997 once to own, with 10 MT4 or MT5 licences per EA, a 14-day free trial and a 30-day money-back guarantee. It is also listed on its own on the MQL5 Market at $2,800. On 10 September 2026 the developer's site listed it alone at $1,960, discounted from $2,800.
Is Perceptrader AI available for MT4?
Yes. It runs on both MT4 and MT5, and the developer's licence covers 10 MetaTrader 4 or MetaTrader 5 activations per EA, so you are not buying separate versions.
What currency pairs does Perceptrader AI trade?
Six crosses on the featured signal: GBPCHF, NZDCAD, AUDCAD, NZDUSD, USDCAD and AUDNZD, averaging 18 trades a week with a 19-hour average holding time. Several overlap with the developer's Waka Waka EA, so running both adds correlated exposure.
How long has Perceptrader AI been live?
The featured account has traded since August 2019 and MQL5 has monitored it since 8 April 2023, but Perceptrader AI was only published on the MQL5 Market on 24 July 2023, so the early years of that record predate the product as sold. Of the four real accounts, the two started on 28 May 2023 are the nearest to the product's launch, and one of them is the 65.44% account.
Do you trade Perceptrader AI yourselves?
Not any more. We ran it on our own real account from 19 August 2026 and stood it down on 27 September 2026, after this re-assessment, because we will not run an Expert Advisor we tell readers not to buy. That account was far too short to count as evidence either way and was never part of the verdict.
Could Perceptrader AI be recommended again?
Yes. The conditions are published in the disclosures above: the developer stating which risk setting each real account runs, and a real account on a named, buyable configuration reaching 24 months with a maximum equity drawdown below 30%. We would reassess on that evidence.
Review Our Complete Testing Methodology

Discover More EAs

Explore our directory of independently verified trading algorithms. Every system is assessed against the same published standards, and where the evidence changes, so does the verdict. Our NightVision review is another example: a long live record, no deception found, risk disclosed honestly, and still Not Recommended.

Back to Best Expert Advisors →
100% Data-Driven & Independent Reviews
The Robotic Trader Logo
Risk Warning & Affiliate Disclosure: Trading foreign exchange and CFDs on margin carries a high level of risk and may not be suitable for all investors. Past performance, including a verified live track record, is not a guarantee of future results. The Robotic Trader may receive compensation from partner brokers, VPS providers and prop firms through affiliate links, at no extra cost to you. This supports our research and does not constitute personal financial advice. Commercial relationships never affect our verdicts, and every one is disclosed on the page it affects. Always trade with money you can afford to lose.

© 2026 The Robotic Trader. All rights reserved.