10% success rate is better than 60% success rate?
- Thread starter botpro
- Start date
If knowledge of probability would lead to super performance, all successful traders would be phd's in math. In reality this is not the case, not even close. So being an expert in probabilities doesn't mean much in trading. Probabilities can also be managed in other ways than the way that was demonstrated here.
Your solution was not available in the poll, so not a valid answer. It was 10% or 60% system that had to be traded. Not 90% or 40%.
With equal R:R, wouldn't it be just as difficult to find a 10% success rate as it is a 90% rate? Good luck!
Developing a strategy with a 10% success rate is as difficult as developing a strategy with a 90% success rate; yes, this is true I would say.
Oh brother,....
Doesn't anybody know how to verbalize what expectancy is on a trading system? EITHER system could be better than the other. You could even have a system that FLIPS A COIN (50% win rate) and makes money. The key to making money and being "better" is in cutting losses short and letting profits run. So the risk amount taken and the profit and loss of each successful and unsuccessful trade respectively will help determine the best system expectancy.
To explain it in detail,... Let's assume a $100k trading account and each trade risks 1% or $1000 which we will call 1R. Each system will have profits and losses described as a multiple of R.
System #1. As per the OP has 10% wins. A system like this is most likely designed to have a very high profit for each trade that is successful and in this system we will assume has an average loss of 1R and an average win of 24R. That means that for every 10 trades we will lose 9 * 1R and win 1 * 24R. This system generates 15R over 10 trades.
System #2. As per the OP has 60% wins. A system like this is better than a coin flip probability and most likely designed to take significant bites out of a trend by trading with the trend. So this system we will assume has an average loss of 0.4R and average win of 1.6R. That means that for every 10 trades we will lose 4* 0.4R and win 6* 1.6R. This system generates 8R over 10 trades.
So which system is better? You can not answer the question without a basic understanding of expectancy and how systems are designed. In terms of probability system #2 is better. In terms of letting profits run system #1 is better. In terms of limiting losses system #2 is better because we tend to lose less than we risk. So in that system a stop loss is probably moved up as the trade matures and lessening risk as time goes on. Instead of losing the entire 1R risked when the trade was initiated we only lose 0.4R. In terms of maximizing profits over 10 trades system #1 is better at 15R. HOWEVER,.... We also did NOT define how many times a month you get 10 trades generated by the system. What if system #1 only generates 1 trade a month?? And system #2 generates 10 trades a month? NOW, in the course of a year system #1 would generate ABOUT 15R for January to October and lose 2R for November and December for about 13R/year...13% return. In the course of a year system #2 would generate 8R per month or 96R/year....96% return!!!
Obviously, these numbers are made up and these numbers could be just about anything. This question can not be answered without a basic understanding of expectancy and profit and loss and how systems are designed. Also we do not have a way of knowing maximal drawdown on the system but a system that has a higher probability of losses (system #1) will have longer streaks of losses before a winner and could end up having significant drawdowns since not every 10 trades will result in exactly 15R every time. It's entirely possible that if a Montecarlo style analysis were done that system #1 would have wide swings in portfolio funds.
So, I hope that answers the question to the OP satisfaction. You need to provide much more info to determine which one is better. Lastly, not all systems work perfectly as described with max of 1R or 0.4R losses. That is just the AVERAGE loss per trade. What if a trade generate 5R losses every 30 trades with a gap down past a stop loss? Again you see how you cannot answer the question with the info provided and you also need to define the risk each person is willing to take and the types of swings and max drawdown each trader is wiling to take. These answers are different for all of us. I may only tolerate a max drawdown of 10% and others may tolerate a max drawdown of 30% in which case swinging for the fences in a high R multiple win with low probability trades may be more acceptable for that trader.
Ok just noticed risk and reward is supposedly the same for each system as per OP Ina follow up post. That still does not provide enough info unless you assume the win results in 1R and loss results in 1R. Firstly, I would never trade with a 50/50 chance of winning what I risk (1R). Minimum reward risk ratio should be 2:1 but most would say 3:1. So I would not trade either of these systems in that case. However, using the stats above you have system #1 at 9R losses and 1R wins. This is -8R or a NEGATIVE expectancy system. I would obviously never trade this system. System #2 has 4R losses and 6R wins so 2R expectancy over 10 trades. This is probably on the low end of a system I would want to trade.
Doesn't anybody know how to verbalize what expectancy is on a trading system? EITHER system could be better than the other. You could even have a system that FLIPS A COIN (50% win rate) and makes money. The key to making money and being "better" is in cutting losses short and letting profits run. So the risk amount taken and the profit and loss of each successful and unsuccessful trade respectively will help determine the best system expectancy.
To explain it in detail,... Let's assume a $100k trading account and each trade risks 1% or $1000 which we will call 1R. Each system will have profits and losses described as a multiple of R.
System #1. As per the OP has 10% wins. A system like this is most likely designed to have a very high profit for each trade that is successful and in this system we will assume has an average loss of 1R and an average win of 24R. That means that for every 10 trades we will lose 9 * 1R and win 1 * 24R. This system generates 15R over 10 trades.
System #2. As per the OP has 60% wins. A system like this is better than a coin flip probability and most likely designed to take significant bites out of a trend by trading with the trend. So this system we will assume has an average loss of 0.4R and average win of 1.6R. That means that for every 10 trades we will lose 4* 0.4R and win 6* 1.6R. This system generates 8R over 10 trades.
So which system is better? You can not answer the question without a basic understanding of expectancy and how systems are designed. In terms of probability system #2 is better. In terms of letting profits run system #1 is better. In terms of limiting losses system #2 is better because we tend to lose less than we risk. So in that system a stop loss is probably moved up as the trade matures and lessening risk as time goes on. Instead of losing the entire 1R risked when the trade was initiated we only lose 0.4R. In terms of maximizing profits over 10 trades system #1 is better at 15R. HOWEVER,.... We also did NOT define how many times a month you get 10 trades generated by the system. What if system #1 only generates 1 trade a month?? And system #2 generates 10 trades a month? NOW, in the course of a year system #1 would generate ABOUT 15R for January to October and lose 2R for November and December for about 13R/year...13% return. In the course of a year system #2 would generate 8R per month or 96R/year....96% return!!!
Obviously, these numbers are made up and these numbers could be just about anything. This question can not be answered without a basic understanding of expectancy and profit and loss and how systems are designed. Also we do not have a way of knowing maximal drawdown on the system but a system that has a higher probability of losses (system #1) will have longer streaks of losses before a winner and could end up having significant drawdowns since not every 10 trades will result in exactly 15R every time. It's entirely possible that if a Montecarlo style analysis were done that system #1 would have wide swings in portfolio funds.
So, I hope that answers the question to the OP satisfaction. You need to provide much more info to determine which one is better. Lastly, not all systems work perfectly as described with max of 1R or 0.4R losses. That is just the AVERAGE loss per trade. What if a trade generate 5R losses every 30 trades with a gap down past a stop loss? Again you see how you cannot answer the question with the info provided and you also need to define the risk each person is willing to take and the types of swings and max drawdown each trader is wiling to take. These answers are different for all of us. I may only tolerate a max drawdown of 10% and others may tolerate a max drawdown of 30% in which case swinging for the fences in a high R multiple win with low probability trades may be more acceptable for that trader.
Ok just noticed risk and reward is supposedly the same for each system as per OP Ina follow up post. That still does not provide enough info unless you assume the win results in 1R and loss results in 1R. Firstly, I would never trade with a 50/50 chance of winning what I risk (1R). Minimum reward risk ratio should be 2:1 but most would say 3:1. So I would not trade either of these systems in that case. However, using the stats above you have system #1 at 9R losses and 1R wins. This is -8R or a NEGATIVE expectancy system. I would obviously never trade this system. System #2 has 4R losses and 6R wins so 2R expectancy over 10 trades. This is probably on the low end of a system I would want to trade.
Last edited:
John Henry was a big and famous fundmanager. He had on average 1 win for every 2 losses. So 33% success rate. John Henry made big profits in his 33% winners and much smaller losses in his 66% losers. That's why he could close many years with big positive returns.
According to the inverting technique, proposed here as the best solution, the success rate would be 66% and would be much better. But in reality John Henry would have a 66% success rate with small profits (as these trades would go against the trend) and 33% losing trades with massive losses as all these trades would also go against the trend.
And in life REALITY counts, not probabilities. The REAL result is what matters. John Henry stayed long periods in his trades, no matter if the next day would be up or down. In fact he did not base his trades on higher or lower next day probabilities, so even if these probabilities are correct, they are useless.
Eganon explains very well that much information was missing to be able to judge. That's what I told earlier too, but no as extensive as Eganon did.
According to the inverting technique, proposed here as the best solution, the success rate would be 66% and would be much better. But in reality John Henry would have a 66% success rate with small profits (as these trades would go against the trend) and 33% losing trades with massive losses as all these trades would also go against the trend.
And in life REALITY counts, not probabilities. The REAL result is what matters. John Henry stayed long periods in his trades, no matter if the next day would be up or down. In fact he did not base his trades on higher or lower next day probabilities, so even if these probabilities are correct, they are useless.
Eganon explains very well that much information was missing to be able to judge. That's what I told earlier too, but no as extensive as Eganon did.
The OP did say, in other words, assume that all other 'things' are equal. This is a common/necessary assumption for many math/IQ test problems when the point being made or tested is not about the 'other things.'
The question is about probability, not total returns.
In still other words, the question specifically ignores the 'value' of the trades.
Asserting that the values matter even when the question states to ignore them doesn't change your score.
The 'thing' here was to test/illustrate: thinking outside of the box, imo. No big deal. Most get math/IQ test problems wrong. The OP may have gotten it wrong when he first encountered it, assuming he didn't 'invent' it.
If you're going to kill the messenger, maybe wait for a more solid reason.
The question is about probability, not total returns.
In still other words, the question specifically ignores the 'value' of the trades.
Asserting that the values matter even when the question states to ignore them doesn't change your score.
The 'thing' here was to test/illustrate: thinking outside of the box, imo. No big deal. Most get math/IQ test problems wrong. The OP may have gotten it wrong when he first encountered it, assuming he didn't 'invent' it.
If you're going to kill the messenger, maybe wait for a more solid reason.
So totally meaningless for traders as they want returns, not probabilities.
That's why I asked:
What color of shoes should you wear to make more profits? And would it help to take two left shoes?
My question has the same value for a trader as the OP's question. No value at all.
For traders everything that can influence their returns should be included in analysis and calculations to make a valid study.
I agree that it will be VERY difficult to make a strategy, 1) with 10% success rate and 2) with 90% rate.
It is probably because success rate in most game will be between 40% and 60%. For example, Baccarat has 48% success rate, if I heard correctly.
How about lotto? Many different style of lotto, in different states, has different success rates.
Furthermore, strategy with success rate p=0.6 might be better to long run profit than a strategy with p=0.7.
Roughly 60 years ago, Kelly gave answer for p and AveWin/AveLoss.
Please study Kelly criteria at http://www.investopedia.com/articles/trading/04/091504.asp , which may give answer for the OP.
Many site, such as http://www.bettingtools.co.uk/kelly-calculator , calculate answer for a special p and a special WL.