If a strategy loses on almost every trade, the obvious move is to take the other side of it. Flip every buy to a sell, every sell to a buy, and collect what the losing trader gives away. The arithmetic usually kills the idea, and not for any philosophical reason. Commissions and slippage are charged on both sides of the flip.
Reverse copy trading inverts a strategy's gross edge but not its trading costs, so the flipped version only nets a profit when the original was losing more per trade, before costs, than one round turn costs to execute. A strategy that loses because it bleeds commissions and slippage loses in both directions. At equal size and equal cost, the original's net expectancy plus the inverted version's net expectancy always equals minus two times the round-turn cost, so the pair can never sum to zero.
Does reverse copy trading actually work?
It works in one narrow case: the source strategy must lose more per trade, before costs, than a round turn costs to execute. That case is rarer than traders assume. Most consistently losing retail strategies are not badly wrong about direction. They sit close to a coin flip and die from friction, and friction does not care which side of the market you are on.
Reverse copying is shipping software, not a thought experiment. A developer write-up on MetaQuotes' own community site, covering reverse trade copying in a MetaTrader 5 signal copier, states the workflow plainly: master buy becomes slave sell, master sell becomes slave buy, with lot size taken from balance ratio, a fixed lot, or a multiplier. Native copy services mostly do not offer it. cTrader Copy's documentation describes investors copying a strategy as executed, on an equity-to-equity model, and documents no reverse-direction option. Treat reverse mode as a feature of copier tools, expert advisors and bridges rather than of the platform core, and check your own platform's current documentation.
So the question is never whether you can do it. It is whether the arithmetic survives the cost line. Usually it does not.
Why the edge inverts but the costs do not
Expectancy per trade is (win rate x average win) minus (loss rate x average loss), minus costs. Inversion swaps the two gross terms, because every trade that won now loses and every trade that lost now wins. That flips the sign of the gross term exactly. It does nothing to the cost term.
Costs are charged per round turn, on entry and exit, in both directions. A short pays the same commission as a long. Crossing the spread to sell costs what crossing it to buy costs. Exchange and clearing fees are side-agnostic. Whatever your all-in cost per round turn is, the reverse copy pays it too, on the same number of trades, at the same frequency. Gross flips. Net does not.
A worked example: a 40% win rate strategy, flipped
Use E-mini S&P 500 futures, one contract. ES is $50 x the index with a minimum price fluctuation of 0.25 index points, so a tick is $12.50, per CME Group contract specifications.
For costs, assume an illustrative stack of $4.00 round-turn commission plus one tick ($12.50) of average slippage, for $16.50 per round turn. That figure is illustrative, not a quote. Substitute your broker's schedule and your own measured slippage before trusting any conclusion below.
The strategy: 40% win rate, average win 10 ticks ($125.00), average loss 8 ticks ($100.00). The clean swap below assumes bracketed entries, where the stop and target trade places on inversion; discretionary exits break that symmetry, which the risk section covers.
| Per trade, 1 ES contract | Original | Inverted |
|---|---|---|
| Win rate | 40% | 60% |
| Average win | 10 ticks ($125.00) | 8 ticks ($100.00) |
| Average loss | 8 ticks ($100.00) | 10 ticks ($125.00) |
| Reward to risk | 1.25 | 0.80 |
| Gross expectancy | -$10.00 | +$10.00 |
| Cost per round turn | -$16.50 | -$16.50 |
| Net expectancy | -$26.50 | -$6.50 |
Original gross: (0.40 x $125.00) minus (0.60 x $100.00) = $50.00 minus $60.00 = negative $10.00. Net after cost: negative $26.50. A clear loser, exactly the kind of equity curve that makes people reach for the reverse switch.
Inverted gross: (0.60 x $100.00) minus (0.40 x $125.00) = $60.00 minus $50.00 = positive $10.00. Net after the same $16.50: negative $6.50 per trade. The flip improved the result by $20.00 per trade, twice the gross edge, and it was still not enough. The strategy's directional content is worth $10.00 in either direction, and it costs $16.50 to collect.
The invariant: both accounts sum to minus twice the cost
Add the two net figures: negative $26.50 plus negative $6.50 equals negative $33.00, exactly minus two times $16.50. That is not a coincidence of this example. At equal contract count and equal cost stack, the gross terms are equal and opposite by construction, so they cancel, and two cost charges remain. Over 500 trades the pair produces negative $13,250 and negative $3,250, summing to negative $16,500, which is 2 x (500 x $16.50).
Picture the plumbing. One signal leaves the source strategy and goes two places: straight into the master account, and through a copier bridge that rewrites the side field before sending it to the reverse account. Each of those two orders passes a toll gate on its way to the exchange, and each toll is $16.50 whichever way the order points. The two branches are mirror images in gross terms and identical in cost terms, so wherever the market goes the pair is pinned at minus $33.00 per paired trade.
That sum belongs to the exchange, the broker and whoever took the other side. No lot size cancels it. Both sides of an inversion can lose, and usually both do.
Two accounts trading opposite sides of the same signal do not sum to zero. They sum to the cost of trading.
How wrong does the original have to be?
Wrong by more than one round turn of cost, measured gross. On ES with a $16.50 cost stack, that threshold is $16.50 / $12.50 = 1.32 ticks per trade. The example strategy loses only $10.00 / $12.50 = 0.8 ticks gross. It sits below the threshold, so it is cost-bled rather than directionally wrong, and flipping it fixes nothing.
In win-rate terms the threshold is uncomfortably tight. Hold the payoffs fixed and write the original's gross expectancy as a function of win rate w: 125w minus 100(1 minus w) = 225w minus 100. The inverted gross is 100 minus 225w. Set that equal to the $16.50 cost: 225w = 83.50, so w = 37.11%. The original has to win less than 37.11% of the time before the flip even breaks even.
Cross-check from the other side. The inverted version wins $100.00 and loses $125.00, so its gross breakeven win rate v solves 100v = 125(1 minus v), giving v = 55.56%. Each extra percentage point of win rate is worth 225 x 0.01 = $2.25, so covering $16.50 of cost needs 16.50 / 2.25 = 7.33 points more, or 62.89%. It only has 60%. And 100% minus 62.89% is 37.11%, matching exactly.
Now make the original genuinely bad: 35% win rate, same 10-tick win and 8-tick loss. Gross becomes (0.35 x $125.00) minus (0.65 x $100.00) = negative $21.25, net negative $37.75. Inverted gross is positive $21.25, net positive $4.75 per trade. Invariant check: negative $37.75 plus $4.75 = negative $33.00. In gross ticks the original loses 1.70 ticks per trade, clearing the 1.32-tick threshold.
Five percentage points of win rate is the entire difference between a flip that loses $6.50 a trade and one that makes $4.75. The decision boundary is razor thin, and it sits in a region where sampling noise dominates.
Why risk management does not invert
A copier rewrites the side field on an order. It cannot rewrite an exit policy. Take a common reason a discretionary strategy loses: it cuts winners early and lets losers run. Invert the direction and that behaviour survives intact. The flipped version now cuts its more frequent winners early and lets its larger losers run. The pathology lives in the exit rules, and the exit rules were never inverted.
Stops and targets also swap roles, which changes what gets hit. The reverse copy inherits the master's stop distance as its target and the master's target distance as its stop. In the example that is an 8-tick target against a 10-tick stop, so the inverted account risks $125.00 per contract where the original risked $100.00, a 25% increase in dollar risk at identical contract count. Whether that wider stop is hit less often depends on your instrument's typical bar range, a question for your fill history, not a general rule.
Two more failure modes deserve naming. Tail shape: a strategy that loses through a few catastrophic trades rather than a steady negative edge inverts into a lottery ticket, many small losses waiting on a rare large win. Under a hard daily loss limit, the string of small losses can end the account before the rare win arrives. Positive expectancy plus a hard drawdown gate is still a losing combination when the sequence runs the wrong way. Selection: losing traders stop trading. The signal source you are inverting is, by definition, the one most likely to change behaviour, cut size, or disappear, precisely because it has been losing.
The copy is not a mirror of the master
Two accounts on opposite sides of one signal are not reflections. They are two participants paying their own way into the book at two different moments. The master detects its signal at time T and crosses the spread on its own side, lifting the offer to buy. The copier then has to observe that fill, transmit it, and place an opposing order, so the reverse account crosses the other side of the book at T plus whatever the hop takes. It pays its own half-spread plus whatever the market moved during the gap. Measure that hop on your own stack: it depends on your bridge, your broker and your colocation, and any vendor number quoted without your infrastructure behind it is marketing.
Fill quality, not inversion logic, decides the marginal cases. Same strategy, same flip, two cost stacks: at $16.50 the inverted version nets negative $6.50 per trade, and at $4.00 (commission only, passive limit fills, zero slippage) it nets positive $6.00. Over 500 trades that is negative $3,250 versus positive $3,000, a $6,250 swing produced entirely by execution. Held to $4.00 commission, the allowable slippage budget is ($10.00 minus $4.00) / $12.50 = 0.48 ticks per round turn, meaning average slippage under half a tick, every trade, in both directions.
The zero-slippage version is not free either. Passive limit entries change which trades fill at all: you get filled on the ones that trade through you and miss the ones that run away. That adverse selection never appears on a commission statement, so "just use limit orders" is not a clean fix. It moves the cost from a line item into your fill distribution. If you have not measured how much slippage your own orders actually pay, you have no way to know which side of the threshold you sit on.
Two notes on the threshold. Higher frequency does not help: at 5 trades a day across 250 sessions (1,250 round turns) the original bleeds $33,125 and the inverted version $8,125, each account handing $20,625 to costs. The flip saves $25,000 of loss and still loses. Micro contracts do not soften the hurdle either. A Micro E-mini tick is $1.25 against the E-mini's $12.50, so the same dollar commission is ten times larger measured in ticks. Micro commission rates are usually lower per contract, but check your own schedule for whether they fall by anything close to a factor of ten.
How many trades before you can trust a negative edge?
Hundreds, not dozens. For the baseline example the per-trade outcome is +$125.00 with probability 0.40 and negative $100.00 with probability 0.60. E[X squared] = (0.40 x 15,625) + (0.60 x 10,000) = 12,250. Variance = 12,250 minus 100 = 12,150, so the standard deviation is $110.23 per trade, roughly eleven times the $10.00 edge being measured.
Trades needed for the measured edge to sit z standard errors from zero: N = z squared x variance / edge squared. At two standard errors, N = 4 x 12,150 / 100 = 486 trades. At three, N = 1,094. Those numbers belong to this payoff geometry, so recompute with yours. The formula transfers, the answer does not.
Call a strategy consistently losing after 30 to 50 trades and you are working from roughly one twelfth of the evidence needed to separate a genuine inverse edge from an ordinary drawdown. Sizing off that sample is the same mistake as sizing off a lucky streak, with the sign reversed. In R multiples, with R set at the original's $100.00 stop, one standard deviation of a 40-trade sample here is $110.23 x sqrt(40) = $697, about 7R of pure noise before any edge exists.
Not trading has an expectancy of exactly $0.00, no margin requirement and no drawdown exposure. The inverted example expects negative $6.50, so doing nothing beats it by $6.50 every time the copier fires.
Flipping the sign column in a backtest spreadsheet produces hypothetical performance in the regulatory sense. Under 17 CFR 4.41(b), simulated or hypothetical performance of a commodity pool operator, commodity trading advisor or any principal thereof may not be presented unless it is accompanied by a prescribed statement, which says such results have certain inherent limitations and do not represent actual trading. That rule governs presentations of those registrants' records rather than your own private research, but the caution transfers: an inverted equity curve has never met a real order book.
The sizing rule most reverse copiers get wrong
A mechanically correct reverse copier inverts direction while preserving risk per trade, not contract count. In the example, holding risk at $100.00 when the stop widens from 8 ticks to 10 means sizing at $100.00 / $125.00 = 0.8 contracts per master contract. A 5-lot master signal becomes 4 contracts on the reverse side, not 5. Left on 1:1 volume mirroring, the copier silently runs the inverted account 25% hotter than the account it is inverting.
Copier software will generally not do this for you. The MQL5 write-up is explicit that reverse mode does not affect the scaling logic and only changes trade direction while keeping lot calculation consistent, and the lot modes on offer (balance ratio, fixed lot, multiplier) all key off the master's volume rather than the inverted stop distance. If your tool behaves the same way, its default configuration is the mechanically wrong one. Contract counts are also integers, so the ratio only expresses cleanly at size: a 1-lot master cannot become 0.8 of anything on the E-mini, though 8 Micro E-minis express it exactly, at the harsher cost hurdle described above.
Correcting it does not rescue the trade. At the 0.8 ratio the inverted net becomes negative $6.50 x 0.8 = negative $5.20 per master signal. Gross edge and costs scale together, so sizing changes how fast a negative expectancy shows up and never whether it does.
What this means on a funded account
Two constraints bind before expectancy gets a chance to work. First, the inverted profile carries the wider stop, so a hard daily loss limit or trailing drawdown arrives on fewer losing trades than the original would have needed. Positive per-trade expectancy is not protection against a drawdown gate, because the gate does not average.
Second, reverse copying one of your own accounts into a second account at the same firm creates simultaneous opposing positions across accounts. Many funded-account programs restrict that under hedging or group-trading rules, and the stated consequence is often account closure rather than a warning. Terms differ by firm and change, so read your firm's current rulebook before wiring anything up, and never assume that a copier supporting a configuration means your program permits it.
So when is inversion the right tool?
As a diagnostic, not an income stream. A large-sample negative edge measured gross of costs says something about the market that its author did not intend to discover. That is worth researching. Build a real strategy around the observation, with its own entries, exits and sizing, tested at the frequency you plan to trade. Do not bolt a reverse flag onto somebody else's order flow and inherit their exit policy, their trade selection and their decision to quit.
The CFTC's advisory on commodity trading systems sold on the internet makes the same point from the regulator's side: system promoters may also fail to take into consideration the impact on profits of commissions and fees charged by brokers.
Reverse copy trading is not a way to convert a bad trader into an income stream. It is a narrow tool for a strategy with a demonstrated, large, gross negative directional edge measured over hundreds of trades, resized to hold risk constant, executed at a cost level you have measured rather than assumed, on an account whose rules permit it. Nearly every real case fails at least one of those tests. The correct default response to a strategy that loses money is to stop trading it. Trading it backwards keeps you paying the same commissions, the same slippage, the same platform fees and the same margin, on a signal you have already proven you do not understand.
Frequently asked questions
Does reverse copy trading actually make money?
Only when the original strategy loses more per trade, before costs, than one round turn costs to execute. Inversion flips the gross edge and leaves commissions and slippage untouched, so a strategy that loses $10 per trade gross while costing $16.50 to trade still loses $6.50 per trade after being flipped. Most losing retail strategies fall below that threshold because they are cost-bled rather than directionally wrong.
Why does copying the opposite of a losing trader still lose money?
Because both accounts pay the full cost of trading. The master pays commission and slippage on its round turn and the reverse copy pays commission and slippage on its own round turn, so at equal size the two accounts sum to minus twice the round-turn cost rather than to zero. With a $16.50 cost stack, an original at negative $26.50 per trade and its inversion at negative $6.50 per trade add to exactly negative $33.00.
How bad does a strategy have to be before inverting it is worth doing?
Its gross loss per trade must exceed the total round-turn cost. On ES with an illustrative $4.00 commission plus one tick of slippage, that threshold is $16.50 / $12.50 = 1.32 ticks per trade gross. In the worked example a 40% win rate is not bad enough (0.8 ticks lost) while 35% is (1.70 ticks lost), so five percentage points of win rate separates a flip that loses from a flip that pays.
Do trade copiers support reverse or inverse copying?
Third-party copier tools, expert advisors and bridges commonly do, and a developer write-up published on MetaQuotes' MQL5 community site describes the workflow directly as master buy becomes slave sell. Native platform copy services mostly do not: cTrader Copy's documentation describes copying a strategy as executed on an equity-to-equity model and documents no reverse-direction option. Treat reverse mode as a feature of copier software rather than of the trading platform, and check your platform's current documentation.
Should a reverse copy use the same lot size as the master?
No, because the stop and target swap roles, so identical contract counts silently increase dollar risk. In the worked example the master risks 8 ticks ($100) and the reverse copy risks 10 ticks ($125), which is 25% more risk per contract. Holding risk constant means sizing at 0.8 contracts per master contract, so a 5-lot master signal becomes 4 contracts inverted.
Can better position sizing make a reverse copy profitable?
No. Sizing scales the gross edge and the costs by the same factor, so it changes the rate of loss and never its sign. Scaling the negative $6.50 example to 0.8 contracts produces negative $5.20 per master signal, still negative. Only lower costs or a genuinely larger negative edge in the source can change the outcome.
How many trades do I need before concluding a strategy has a real negative edge?
Hundreds. For the worked example the per-trade standard deviation is $110.23 against a $10 edge, so reaching two standard errors takes 486 trades and three standard errors takes 1,094. Compute your own with N = z squared x variance / edge squared. Calling a strategy consistently losing after 30 to 50 trades is a decision made on a small fraction of the evidence needed.
Can I reverse copy one of my funded accounts into another account at the same firm?
Check the firm's written rules first, because that setup creates simultaneous opposing positions across accounts. Many funded-account programs restrict this under hedging or group-trading rules and the stated consequence is often account closure. Terms differ by firm and change, and a copier supporting the configuration says nothing about whether your program permits it.
Is reverse copying easier to make work on micro contracts?
No, it is harder. A Micro E-mini tick is $1.25 against $12.50 on the E-mini, so the same dollar commission is ten times larger measured in ticks, which raises the cost threshold the inversion has to clear. Micro commission rates are usually lower per contract, so check whether yours falls by anything close to a factor of ten and recompute the threshold with your own schedule.
What should I do instead of reverse copying a strategy that loses money?
Stop trading it. Trading it backwards keeps you paying the same commissions, slippage, platform fees and margin on a signal you have already proven you do not understand. If a large-sample gross negative edge really exists, treat it as a research observation and build a strategy around it with its own entries, exits and sizing rather than inverting someone else's order flow.