Inverting the signals in a losing strategy must necessarily make it a winning one.

Any strategy that fails to exploit a real market inefficiency is effectively generating random signals. Inverting a random signal generator will only create yet another random signal generator, which will, therefore, be equally ineffective.

Why is it?

1. Transaction costs (spread, commission, slippage, swap, etc). These accumulate over time, making both win smaller and losses bigger. Like Jen says, the shorter your trading horizon, and the more frequently you trade, the greater the relative impact that these have. Many traders greatly underestimate their effect. Slippage never operates in your favor, but can have a severely destructive effect during market crashes or ‘black swans’.

2. Statistical fluctuation. If your strategy doesn’t exploit a genuine inefficiency — i.e. a recurring directional or behavioral bias in future price movement — then it is effectively a 50/50 crapshoot, and over a long enough period of time (say, a million trades) it will approximately break even (if we ignore transaction costs). But over any shorter time period (say, 500 trades) it could show an overall loss. That doesn’t mean it’s a losing system that can be reversed successfully, it means it’s an ineffectual system that happens to be in loss at the moment, but will eventually break even……. except that transaction costs will gradually bleed it to death.

   A system that appears to be losing heavily could merely be a statistical outlier (if you look at enough ineffectual systems, there will always be a few that have had an exceptionally ‘unlucky’ run to date) that has also been ravaged by transaction costs.

3. Overleveraging. This follows on from point 2. A strategy whose equity fluctuates above and below breakeven will at some future point fluctuate downward to the point where a margin call occurs. In other words, there’s an imbalance simply because there’s no upper limit to how high your account balance can get, but there’s a downside limit that’s an immediate showstopper. The bigger your position sizes, the greater the equity fluctuations, and the sooner this will happen. This is also somewhat akin to Gambler’s ruin: that a player with a smaller bankroll (retail trader) is at a disadvantage against a player with a larger bankroll (the market), simply because any fluctuations have a greater relative effect — in terms of pushing the player toward ruin — on the small player.
4. Broker mischief. With some brokers, unless you’re winning consistently, the br0ker may put you on his B-book, and actively trade against you. Granted, many newbie traders unjustifiably blame their br0ker as an excuse for losses (e.g. widening spreads during high impact news, when in reality this is merely a liquidity issue), but nasties like virtual dealer plugins nonetheless do exist, and I suspect that there are br0kers out there who use them.
5. Market reaction. This likely has minimal effect for retail traders, since their positions are relatively tiny, allowing them to ‘fly under the radar’, but the market is nonetheless dynamic and there is a reaction to every order placed. In other words, heavyweight players will react in a way that unwittingly crushes you, no matter what strategy you use, as they seek to implement their own agendas. Retail traders are effectively cannon fodder that (directly or otherwise) provide liquidity for the bigger fish. When I read comments like “it doesn’t matter what I do, the market always seems to move against me”, I can’t help but wonder if, in some cases, this may be a small contributing factor.
6. Psychological issues. In certain situations, there may be a small edge in cutting losses quickly and holding winning trades for longer. However, human nature being the way that it is, people tend to do the opposite. Hence reversing the entry rules, but nonetheless continuing to allow psychological foibles to impact exits, will continue to deliver a similar overall result. Of course, this isn’t reversing the entire strategy, and it doesn’t apply to automated systems.
7. Can’t be reversed. With many systems it is impossible to exactly reverse the exits or the MM, e.g. trailing stoploss, uneven scaling in/out, martingale sizing.

   Conclusions
   To win at any probability-based game, (1) the game must have some kind of exploitable bias, AND (2) you must apply a strategy that exploits the bias. A game that’s totally random can’t be beaten, and even a game that isn’t totally random can’t be beaten by applying a random (or ineffectual) strategy.