Luna AI Pro EA Review (2026): An 89% Win Rate, Losses Six Times the Wins and an 87% Drawdown on the 'Low-Risk' Signal

49.3 / 100

weighted verdict score on eight published criteria

88.9% win, 0.16 payoff

win rate and average win over average loss on the vendor's 420-week signal, 2,516 trades

87.0%

maximal drawdown by balance on the signal the listing calls low risk, against 30.1% relative by equity

The verdict at a glance: a long record that shows how the loss arrives

Luna AI Pro earns points for how much it shows. Eight years of platform-built statistics is a rare thing, and the seller has built a reputation over two decades. It loses points on risk, where a long record is a mixed blessing: the longer it runs, the more it shows what a very high win rate costs. A system that earns 2 USD a win and gives back 13 USD a loss needs to be right nine times in ten just to break even, and its worst day is worth ninety winners. The scorecard and radar show how the total was built.

What Luna AI Pro actually is

Luna AI Pro MT5 is an Expert Advisor for MetaTrader 5, sold on the MQL5 Market by Profalgo Limited of Malta, whose seller profile shows a 4.0 rating from 1,017 votes, 27 products and 29 signals, with a website at forexeasolutions.com. The developer describes himself as developing EAs since 2005. It was published on 4 July 2023, showed version 1.60 and was last updated on 28 October 2026, with 10 activations per licence and a rating of 5 from 5 reviews. The listing calls it 'a very advanced night scalper', developed with 'the mean-reverse strategy' and 'only the best pairs and techniques'.

The design has a clear shape. Recommended pairs are EURAUD, GBPAUD, AUDUSD and AUDCAD, with secondary pairs GBPCHF, EURUSD, GBPUSD, USDCHF, GBPCAD, AUDJPY, USDCAD, EURCAD, EURCHF and EURGBP. A OneChartSetup runs all pairs from one EURUSD M5 chart. An Individual Performance Monitor cuts the risk on any pair that stops performing and raises it again when the pair recovers. A correlation filter limits exposure to the same currencies, a swap filter avoids high swap-cost days, a rollover filter skips pairs that could slip at rollover, and a news filter pauses trading around high-impact events. The listing says it does not use martingale, grids without a stop loss or very wide stop losses, that there are 'no fake or manipulated backtests', and that the live account has been running for more than a year.

How we judged it, and who has something to sell

Evidence on EAs is never neutral. Here is every party near this product and what each stands to gain, ours included.

• The developer sells the EA and its 26 siblings, rents them, sells signal subscriptions (30 USD a month on this signal, with no subscribers shown) and runs the live accounts. His record is public, which counts for something, and it is also his sales page.
• MQL5 and MetaQuotes host the Market, the reviews and the signals. The signal statistics and every notice on its page are the platform's own.
• Buyers who wrote reviews are, in the part of the page we could read, all positive and short. Testimony, not data, and none can be checked from outside.
• ForexCopilotEA sells a listing for this EA. We gain if you buy it, which is why this verdict opens with that fact and applies the same rules as every other.

What the live signal shows, number by number

A platform-computed signal is the best public evidence a vendor can show, and this one is much longer than most. It still has limits: the vendor runs the account, the broker is one broker, the setting is the vendor's choice, and the account received deposits and paid out withdrawals. One more detail needs a plain statement. The signal page shows a 'Started' date of 27 April 2022, while its weeks counter reads 420 and its lifetime 2,935 days, which is about eight years. The two cannot both describe the first trade, and the platform's lifetime figure is the longer. The Market product itself was published in July 2023, so the early history was made by earlier versions of the strategy. With those limits in mind, here is what the signal shows on 29 October 2026.

Five readings stand out. First, the payoff: winners of 2.12 USD and losers of 13.46 USD. With a win rate of 88.87%, the expected payoff is 0.8887 times 2.12 minus 0.1113 times 13.46, which is 1.88 minus 1.50, or 0.38 USD, matching the platform's 0.39. The whole edge is the gap between 88.9% and the break-even rate, which is 86.4%. A fall of just three points in the hit rate, to about 86%, takes the expected payoff to zero.

Second, the worst trade. A single loss of 194.52 USD equals about 92 average wins and is about 11% of the account's current balance. The signal's longest run of wins is 70 trades worth 552.28 USD, which is less than three average losses of that size. Third, the drawdowns on the same account: 30.1% by equity, 46.8% relative by balance and a maximal balance drawdown of 87.0% (368.15 USD at an early, small balance). The listing's 'low risk' label sits beside those figures, and a buyer should read the figures. Fourth, the profit: 976.63 USD in eight years on a 213.93 USD start with deposits of 933.83 USD. The platform's monthly growth is 0.47% and its annual forecast 8.09%, and it states that 80% of all growth was earned in 27 days. Fifth, the deposits and withdrawals: the account is not a single compounding line, so the +247.93% is the platform's adjusted figure and the dollar profit is the cleaner number.

How a mean-reversion night scalper makes and loses money

A mean-reversion system bets that a pair that has drifted from its recent average will drift back. At night, when liquidity is thin and the majors are quiet, the AUD and CHF crosses trade in tight ranges, and a small, frequent profit is available to a system that sells strength and buys weakness. That is why Luna AI wins nearly nine trades in ten, and why the average win is small: price does not need to move far to reach the target.

The loss arrives differently. A pair that drifts away from its average and keeps going, because of news, a central-bank decision or a regime change, turns a small favourable bet into a large adverse one. The stop loss caps the damage at 13 USD on average and 194 USD at worst, but nothing caps how often. The Individual Performance Monitor is the system's answer: it reduces risk on a pair that is losing, which limits a bad streak on one pair. It cannot prevent several pairs from losing together when the same currency moves, and the AUD-heavy list of recommended pairs makes that a real exposure to Australian dollar events.

Risk: a stop on every trade, and a payoff shape that punishes a bad run

Luna AI Pro is not a grid or a martingale, and it has stops. Why is it still High risk? Because the design the signal shows risks 13 to 194 USD to make 2 USD, and the verified drawdowns, 30% by equity and 47% by balance on the vendor's own low-risk setting, are those of a system that can lose a large share of an account in a stretch. The label follows the measured drawdown and not the vendor's word.

• Negative skew. The average loser is 6.3 times the average winner and the worst trade 92 times. A few losers in a week can erase a month.
• A thin edge. A fall of three points in the win rate removes the expected payoff. Costs, slippage and a wider spread at night all push the hit rate down.
• Currency concentration. The recommended list leans on AUD crosses. An Australian dollar shock hits several pairs at once, whatever the correlation filter says.
• Night execution. Thin liquidity, wider spreads and rollover slippage matter more for a system targeting 2 USD wins. The rollover and swap filters help and cannot remove it.
• Setting risk. The risk input is the buyer's. The signal's low setting produced the drawdowns above; a higher one multiplies them.

What buyers say

The listing shows a rating of 5 from five reviews, too few to carry weight. We could read three. One, from 6 July 2026, is the single line 'Win is the number one in mql5', which we treat as praise for the seller and not as information about the product. One, from 13 May, was bought 'to diversify my portfolio' as an EA 'I can use long-term without worries', trusts that the seller's EAs 'are working long-term and will recover each time' and praises the support. One, from 12 May, has used it 'for a bit now', finds it 'pretty solid' and 'way more reliable' than other EAs, and likes waking up to trades already in profit.

Two cautions apply. The word 'recover' in the second review is accurate and worth noticing: the signal does recover from drawdowns, over long periods. And 'waking up to trades in profit' is the high-win-rate experience in one sentence, which is exactly what makes the rare large loss feel unfair. We read the reviews as positive about support and reliability, and silent about the payoff shape.

Price: $399, rental options and a $2,000 'final price'

The Market price on 29 October 2026 was $399, with the listing saying only one copy remains at that price and the final price is $2,000. As with other products on the Market, treat the scarcity line as a sales message. The rental options are the useful part of the pricing: $150 for a month, $199 for three months or $299 for a year. A year's rental costs 25% less than buying, and three months lets a buyer run the live product on a demo or small account through enough nights to see how it handles the spread and rollover at their broker. Rent first if you can; a system with a thin edge should be tried on your own costs before you commit to a purchase.

Set the price against the signal: 976.63 USD of profit in about eight years, with a forecast of 8% a year, at the vendor's low-risk setting. A buyer can scale the risk up, with the drawdown scaling with it. Remember the other costs: a VPS, a low-spread account and the swap on overnight holds. If you buy from a source other than the Market, such as our own listing, the price, version and support are different. Our listing is a separate purchase with its own terms and shows version 1.5 against 1.60 on the Market, so check the version and what is included.

What a bad month looks like for a system like this

Picture a quiet fortnight in which the system wins 40 trades at about 2 USD, banking roughly 80 USD, and then an Australian rate surprise sends three AUD pairs through their stops at 13 USD each and one at the 194 USD worst-case size. The month nets out near zero or below even though the win rate for the month is still above 90%. A buyer who looks only at the win rate sees a system that is working; a buyer who looks at the equity curve sees a month of nothing. Judge the system by the curve and the worst trade, and by how long a recovery took, which on this signal can be many months.

The score, criterion by criterion: 49.3 out of 100

The total is 49.3 out of 100. The grade A on evidence is earned by a 420-week platform signal; the score stays at 6 because the vendor runs it, its early history predates the product, the account had deposits and withdrawals and the risk label is the vendor's. Risk, at 3.5, is the lowest of the criteria and the reason for the High label: the measured drawdown and the payoff ratio, not the absence of a grid. Profitability, at 4, reflects a profit factor of 1.26 and a monthly growth of 0.47%.

Who Luna AI Pro fits, and who it does not

• It fits a trader who wants a long-established, multi-pair mean-reversion system, accepts a low, thin edge, sets the risk very low, runs a VPS on a low-spread account and judges it over years.
• It does not fit anyone who needs strong returns, who will raise the risk after a run of wins, or who cannot sit through a 30% equity drawdown.
• It is risky for prop-firm accounts, where a single 194 USD-sized loss repeated across pairs can breach a daily limit.
• It is a poor choice for a buyer who is paying mainly because only one copy remains at $399. Rent for a month or three first.

How it compares with other EAs we have scored

Luna AI Pro comes from the same seller as Gold Reaper, and the two share a habit of showing long records. Next to One Man Army, which also wins often and loses big, Luna AI has a longer record and a deeper verified drawdown; its 49.3 and High label reflect that. Next to Pulse Engine, which wins half its trades with equal payoffs, it is the opposite shape. For how these system types differ, see our strategy comparison and our forex EA category.

Questions to put to the developer before you pay

1. Which version and risk setting does the signal run, and what does the 'low risk' label correspond to?
2. Why does the signal show a started date of April 2022 and a lifetime of 420 weeks?
3. What is the largest drawdown by equity at the low-risk setting since the Market launch in July 2023?
4. What is the break-even win rate at the current payoff, and how has the live win rate varied by year?
5. How does the correlation filter treat several AUD pairs opening at once?
6. Can you publish a real-tick tester report for the current version?
7. What risk do you advise for a 1,000 USD account?