I assert that it is easy to make a losing system. Given that, one of the following is true: 1) Creating a winning system is also easy, or 2) Reversing a losing system will not necessarily generate a winning system. Spread and commission seem so insignificant but they are a HUGE barrier to overcome in the long run.
Spread and commission are almost irrelevant on larger time frames. In other words, if you're scalping, it's a big deal. But if you're swing trading and the reason you are losing money is because of comissions, then you're doing it wrong. In other words, even Scottrade has only $7 commissions per trade. So if you're swing trading and making less than $14 profit per trade (0 after commissions) then your strategy sucks. So if the reason your losing system is still a losing system even after you switch buys and sells is because of commission, then your system wasn't good to begin with. But that stuff aside, when you swapped the buys and sells of your losing system, did you still lose money with it?
Yes, reversal trading systems are difficult to program. I can attest to that after 14 years of programming them as a retired professional programmer. With reversal systems the exit is your entry so you really learn what makes prices tick (pun intended). Otherwise a bad exit entry cost you on both trades.
I have a hard drive full of ideas that don't work. If I could reverse the non-working ideas and make money with them, then I wouldn't have a hard drive full of ideas that don't work; I'd have woven gold undies and flying cars. Overcoming spread and commission is an enormous chore regardless of time frame.
Not really in long timeframes. How badly are you trading that $14 is the cutoff between profitability and unprofitability? Are you trading super small lot sizes or something? Or are your target profits microscopic? Even if you only had $1,000 to invest in SPY (round it up to $130 per share), you could buy 7 shares ($910) plus commission ($7) for $917. This means you'd have to sell at $132 just to break even. You'd sell for a net of $924 minus the comission ($7) and you'd be at $917, which is what it cost you to buy. If you're swing or long term trading for a 1.5% gain ($2 on $130) then you need to reassess your trading plan. I personally say that $1,000 is a bit small of an amount to be trading with, so let's say you have $5,000 instead. You can now buy 38 shares for $4,940 + commission which is $4947. Now price has to go up about 38 cents per share in order for you to break even, at which point you would sell for $4954.44, minus commission, which puts you at $4947.44. If you're swing or long term trading for 0.2% gain (38 cents on $130) then you need to reassess your trading plan. If you're one of the ET millionaires who has figured out the secrets to make MACD (or any of the other indicators) profitable, then you probably have much more. Let's say you can buy a $20,000 position of SPY. Now you have 153 shares. You now need a $0.09 gain in SPY in order to break even after commissions. That's 0.06%. If your target profits on a long term swing trade are a 0.06% increase in stock price, then your trading plan needs to be revised. Just saying. Of course, ET millionaires who can use MACD profitably obviously know exactly when to exit a trade (since MACD tells them) so this entire post doesn't even apply to them.
Spread and commissions are always a factor because you have to overcome them just to break even. If you're trading at Scottrade at $14 a round trip, that's the first thing that you're doing wrong. One broker offers 1/2 a cent per 100 shares and it can be even lower if you go unbundled and provide liquidity. One dollar per round trip (or less) is a HUGE difference from $14. I doubt that you'll believe it but it is possible to make a handsome yearly return while making less than $14 per 100 shares. And no MACD required
Dood, I get the math. A few cents here and a dollar there doesn't add up to much. Do you have a better explanation for why losing systems still lose after you reverse the rules? When you're buying breakouts, you are going to get the best prices at the worst times and the worst prices at the best times. When you're fading breakouts, you are going to get all the bad fills and only some of the good ones. If you're exiting on a stop at 1000, you are going to sell if even 1 contract goes off at 1000. If you're buying at 1000 in your reversed system, that 1 lot sell order isn't going to fill you, and you might not get a fill at all. It's no coincidence that companies that capture spread and that collect commission are much, much more likely to make money than the average trader. Edit: this was a reply to 1a2b3cppp, not spindr0 in case that wasn't obvious.
You're so very right. Automating a reversal system (for me anyways) has been incredibly difficult and unsuccessful...still trying though. I have an intraday trend following automaton that does fairly well, but reversals are tough to automate. If you have any success executing this, please PM me and let me know. Obviously I don't want your secret (unless of course you want to share), I'd just like to have more assurance that it's possible ---- but damn't I know it is.
1) It's a mathematical fact that any system which is unprofitable with midpoint executions and no commission would be equally profitable under the same assumptions if reversed. So if your commission and spread are small, the basic argument for reversing losing systems is solid. 2) Be aware that all the errors of inference that make people think a system is profitable when it's not also cause people to think systems are unprofitable when they're not. Many poorly conceived systems are right about break-even. 3) Many systems, when reversed, have a changed risk profile. For example, a trend following system with a fixed stop become a mean reversal system willing to fade arbitrarily large losses to hit a fixed profit. The system may be profitable, but still be untradable. 4) Finding a working system is easy. Buy S&P when it closes 3 weeks above it's 40 week MA. Sell when it closes 3 weeks below. Now you've got one. It's whipped the market like a Tijuana whore for 30 years. The issue is not getting a working system, but getting one that trades in a way that suits what you want to do. 5) Much of trading is not a matter of improving your systems. It's frequently a matter of taking basic systems and figuring out where they will still work well. It's all about table selection.