That because the basis of your question - "...successful traders usually advocate a reward-to-risk where the reward is several times the risk - isn't necessarily true. Where did you come up with that one? I trade a very nice system that's 1:1 with a 65% win ratio. Others on this same thread have reported much the same. There are countless approaches to systems development, and some systems like an even stop and target, while other entries benefit from a different ratio. There is no single way to the promised land. The bottom line is YOUR equity curve. Either what you are doing is adding to it, or detracting from it. Period.
There are many aspects to consider when comparing all in/out vs scaling strategies. Scaling can help counter some of the variance of trading and smooth out the equity curve. This can potentially cut overall risk allowing for more leverage and more consistent compounding of returns which can result in greater overall profit. Of course this is all money management theory. From experience, I developed systems that have improved considerably after applying a scaling strategy, however this is not always the case. Systems that are curve fitted typically can't be improved with a scaling strategy.
No. If you are following a system, scaling out will not increase your overall profit. This has been discussed at length here, I urge you to read the rest of this thread and the link to the other one about scaling out. If you don't agree, please post an example with specific numbers using any target/stop parameters you want. This will help you learn why it is inferior.
Most of the discussion on that thread does not factor in variance. Back testing is done in a controlled environment as the historical data never changes so any system developed is subject to curve fitting. Curve fitted system will usually work best using an all in/out method vs scaling. However this does not reflect what will work best forward testing due to anticipated market variance. Let me give an example. Trend following system xyz uses a simple strength indicator to trigger entries and exits. The strength indicator value range is 0 to 10. This system gets the best results entering the market at value 8 and exiting at value 3 for a back test of 5 years. However, entering at value 8.5 and exiting at 3.5 works best for year one and entering at value 7 and exiting at value 4 works best for year 2 and so on with each year doing better with a different value set. While it is easy to determine what values worked best in the past, it is unknown what values will work best going forward. However the back testing shows a range of values that generally worked well during back testing and it is anticipated those values will generally work well during forward testing. Lets look at two ways to trade the system. All in/out Go all in when the indicator hits 8 and all out when the indicator hits 3. (historically the best performance) Scale in and out. Scale in the market during ranges 7-8.5 and scale out at ranges 4-3. (historically not as good as first method but very close) Once again, due to the variance of the markets, it is unknown which method will do best forward testing. However with a little help from LLN (law of large numbers) among other things, using the scaling method has a better chance of replicating past performance than an all in/out method. I'm not saying that all in/out is inferior to scaling, only that all in/out is not always the best method and there are good reasons to use scaling. Edit. I didn't expand on why scaling can (but not always) produce bigger profit. Without going into detail, scaling can increase the winning percentage of the position taken. This may be at the cost overall profit for the individual trade, however consistent winnings compound faster and potentially provide a greater overall profit. Compounding returns can add up fast with frequency and consistency winnings. Scaling can effectively make better use of compounding due to the increase in consistency provided by the LLN advantage that scaling provides.
Setting aside the scaling vs all in/out debate. With good money management, a trading method that generally produces 65% 1:1 with a frequency of 1 or more trades a day will provide a nice return.
Statistically, 1/1 ratio with 65% probability is a good method, no doubt about it. Equally, 4/1 ratio with 40% probability is a good method too. To a scalper, 1/3 ratio with 90% probability is acceptable. by the way, scaling in and scaling out is a stupid thing in my opinion. If you should get out, get out altogether. If you eat a banana, you eat all of it, not leaving half of it in your mouth and the other half in your throat.
Are you amusing that 100% of bananas are 100% edible? If 65% of bananas are edible, how many bananas are needed to guarantee getting 1 that is edible?
I totally agree with lindq (who, btw, it a much more experienced and successful trader than I). I have already shown you the stats to one of my systems, which is now my 'favourite' due to its consistency. I have another system which is more of a swing style. The risk:reward is truely horrible, mainly because it doesn't really use a stop (only a multi standard deviation 'disaster' stop - hit once in the last 7 years back in Jan 2008). However, the AVERAGE loser is only 2-3x the av winner with a win rate of 90%+. This swing system is v v profitable over the long run with a lovely smooth eq curve for the most part, with the odd divot. However, I size these trades v conservatively. Why? (a) Because the psychological impact of the stop being hit back in Jan was pretty bitter and (b) because how does one know when a system has lost its edge? Ultimately by incurring more losers than expected. This may be palettable at 3:1 reward:risk, or even 1:1, but not when you use the kind of stops this system does. Eventually I may look to drop this system from my traing entirely, despite the long-run profit factor of 4+. Overall, I think that any robust strategy which put the odds in your favour, whether due to a favourable R:R, a high win rate, or a combination of both in real trading (not just back-testing) is worth trading, providing it fits well wtihin your portfolio of strategies and you have a suitable warning mechanism to alert you when the edge has diminished or disappeared.
To labour the point slightly more - in trading I try to emulate a casino, playing on the house side (like many other traders). i.e. Develop a game (trading strategy) with the odds in your favour and play it as many times as possible. Think about roulette. The R:R for the casino is TERRIBLE for each spin when a gambler plays the individual numbers - 35 or 36:1 depending on the casino. However, the casino doesn't stop offering roulette. The high win rate compensates for the poor R:R and the casino makes money over enough games.
Is there not a slight flaw in your equasion (amusingly)? What if there's only one banana? In keeping with your speculation (and the way I'd play that particular game as you laid it out), should you get a bad one, your prospects of having a good one will exponentially increase the more bad ones you select.