scose-no-doubt
Veteren member
- Messages
- 4,630
- Likes
- 954
The Coin Flip
- Thread starter higherSelfishness
- Start date
- Watchers 10
the hare
Senior member
- Messages
- 2,944
- Likes
- 1,282
well yes
The Leopard
Experienced member
- Messages
- 1,877
- Likes
- 1,021
Yes?
the hare
Senior member
- Messages
- 2,944
- Likes
- 1,282
scose-no-doubt
Veteren member
- Messages
- 4,630
- Likes
- 954
hmmmmm....
*ponders*
*ponders*
The Leopard
Experienced member
- Messages
- 1,877
- Likes
- 1,021
hmmmmm....?
scose-no-doubt
Veteren member
- Messages
- 4,630
- Likes
- 954
Spec-K Trading
Junior member
- Messages
- 14
- Likes
- 1
Lol!!! I'm getting dizzy.
To get back to the original question (because randomly generated charts look pretty similar can we conclude market action is random)...no, not just because of that rather superficial assessment. We could certainly some statistical tests that would give us some indications (but my days of upper division stats are long behind me). I wasn't trying to lead to that conclusion (that market behaviors are truly randomly generated), just make the point that our layering of pattern recognition and meaning is rather suspect.
Another interesting element of this is "entry selection" and the idea that through micro-level analysis of price/volume action we can choose "better" entries relative to probabilities of shortly upcoming price action. If I have a random process and I'm guessing future events, it makes no difference how I guess (i.e., if a random # from 1 to 36 is being generated, it makes no difference whether I guess the same # every time, a different number, or decide based on how many seconds until my butt itches again). Similarly, to the degree the markets are randomly driven, all our technical analysis to give us "high quality entries" may similarly be just mumbo-jumbo.
Bulkowski's work on chart patterns and following price action probabilities tries to prove statistical significance (non-random behavior), but I have a problem with his data for the simple reason that he has so few data points (tens to at most a few hundred).
One way to think about the problem is first to ask, what amount of data is needed, and what analysis on that data is needed, to conclude that data from a truly random process (coin flips work) is in fact random? If we have a good answer to that, apply it to financial market data.
I'm sure this area has been studied to death. My pragmatic view is that fine level TA is largely mumbo jumbo, trying to create/impose statistical significane when none exists. On the other hand, I do believe that significant trends that are more than just the result of random behavior in price action, they reflect statistically significant correlated behaviors by large numbers of traders/investors, and my trading technique is primarily predicated on that. But I haven't dug into the research that proves it so. My results are good, I'd hate to dissuade myself!
To get back to the original question (because randomly generated charts look pretty similar can we conclude market action is random)...no, not just because of that rather superficial assessment. We could certainly some statistical tests that would give us some indications (but my days of upper division stats are long behind me). I wasn't trying to lead to that conclusion (that market behaviors are truly randomly generated), just make the point that our layering of pattern recognition and meaning is rather suspect.
Another interesting element of this is "entry selection" and the idea that through micro-level analysis of price/volume action we can choose "better" entries relative to probabilities of shortly upcoming price action. If I have a random process and I'm guessing future events, it makes no difference how I guess (i.e., if a random # from 1 to 36 is being generated, it makes no difference whether I guess the same # every time, a different number, or decide based on how many seconds until my butt itches again). Similarly, to the degree the markets are randomly driven, all our technical analysis to give us "high quality entries" may similarly be just mumbo-jumbo.
Bulkowski's work on chart patterns and following price action probabilities tries to prove statistical significance (non-random behavior), but I have a problem with his data for the simple reason that he has so few data points (tens to at most a few hundred).
One way to think about the problem is first to ask, what amount of data is needed, and what analysis on that data is needed, to conclude that data from a truly random process (coin flips work) is in fact random? If we have a good answer to that, apply it to financial market data.
I'm sure this area has been studied to death. My pragmatic view is that fine level TA is largely mumbo jumbo, trying to create/impose statistical significane when none exists. On the other hand, I do believe that significant trends that are more than just the result of random behavior in price action, they reflect statistically significant correlated behaviors by large numbers of traders/investors, and my trading technique is primarily predicated on that. But I haven't dug into the research that proves it so. My results are good, I'd hate to dissuade myself!
Shakone
Senior member
- Messages
- 2,460
- Likes
- 665
Nice and sensible post Spec-K, nothing to really disagree with. I don't know what fine level TA is. I believe in Support and Resistance, Trend, that how you manage the trade is more important than where you got in, and that the markets exhibit various human/animal behaviours and so context is vital.
For me it's all of these factors and particularly the context that makes a lot of statitical analysis flawed (not all analysis). I come from a very math and stats background, so I don't say this lightly. But just as there is mumbo-jumbo TA, there is - if I can steal the hare's phrase - plenty of 'intelligent nonsense' statistical analysis as well.
Spec-K Trading
Junior member
- Messages
- 14
- Likes
- 1
A definite "yes" to viewing the markets through a statistical lens! I view stock price in the future on a probabilistic basis, informed by prior action, and trade accordingly. When I refer to TA mumbo jumbo relative to entry point selection, I'm talking about using indicators and the like, primarily focused on shorter term data, to pick an entry point that will in theory give substantially higher probability in the very short term to prices being higher than being lower, and from there trying to use tight stops enabled by our entry point wisdom. I don't believe that works very well, due to far more short term randomness than we'd like to believe exists in the market.
scose-no-doubt
Veteren member
- Messages
- 4,630
- Likes
- 954
I think this is the first seriuzz (is it even seriuzz?) and definately the tldr-iest post I’ve ever made on here. I know a lot of what I’m about to say is an egg sucking lesson, but it’s helping the flow of my though process so forgive me lulz are in order.
Scose’s musing and questions/points for discussion are as follows:
All financial markets are driven by macroeconomic activity.
I'm sure the vast majority will agree that they are (but this is trade2win after all and sometimes I think that the 200ma is give more significance than “funnymentals”) and so we are left with the question of what signifies the underlying economic trend or a change therein and what constitutes noise. The noise is what we’re referring to as a random variable/random price behaviour but what it really is imho, is the intra-day gaming which pushes the natural supply/demand relationship out of a conceptual (and I'm almost certain non-existent) equilibrium and into relative extremes. If this is the case then these extremes can be upper or lower bound and so we're left with what is essentially an unknown quantity which may or may not be treated as a random variable. But... people make a living from assessing and trading from this type of price movement so it's likely not random. Couple that with my favourite subject which is EMH and arbing (oomph!) and other pricing mechanisms and you're left with a market view that would have Salvador Dali keeled over the sh1tter with nausea. So does Scose agree that he short term market is random? Yes and no but also yes and no. Digging a little deeper, insofar as a multitude of parties being required to buy/sell at any given time, yes, but the functions of basic auction theory and methods of liquidity provisiorrrr prrrrrovide a structure of sorts, so no. Then you have front running algo stuff to consider. This b0llocks is imho, where the short term trading opps are and where the inefficiency is. Looking at anything else short-term and you’re kidding yourself.
Now that we had added a sufficient degree of over-thinking the most basic facets of price movement, let us now move on to what constitutes randomness. This word is bandied about rather often on t2w and I’m not sure whether I’m being ignorant, everyone is lulzy or the word just isn’t really understood by some. The word random is tossed out and the sacrilege brigade fly off the handle and start Ross Hook the cheeks, auction hammer the faces or dojify (crucify + doji) those who dare utter the dreaded word. Statistics 101 or whatever the UK equivalent is (probably year 10/11 from some things I’ve seen) tells you that random doesn’t necessarily mean even distributed. What if at any given time our variable has a 60% chance of going one way and a 40% chance of going the other? Is that still not random? What if our random variable is normally distributed or leptokurtic (the concept that Robster w@nks over) or even displays some skew? Surely this is not outside the realms of possibility is it? If the markets are indeed driven by macro-economic trends which take time to play out wouldn’t we expect more people to be on one side of the trade than the other as they manage exposure or future expectations? Wouldn’t this effect supply and demand equilibrium? Wouldn’t this mean you’re more likely to make a profit buying rather than selling or vice versa as the macro case may be? Doesn’t this mean skewed returns? Doesn’t a skew in returns translate to a skewed probability distribution? Would this skewed distribution exhibit leptokurtosis (not going to go into why here)? How would all of this exhibit itself on a price chart? It would look like a trend wouldn’t it? Isn’t a trend just like a bull market or whatever the TA term is? Aren’t pullbacks etc extremely similar to how a chart of the inevitable outcome of a series of non-evenly distributed random events would look? While prices may or may not be random in and of themselves, given the above, shouldn’t a statistical process be a good enough proxy for analytical purposes-for providing a framework or approach to understanding?
So I ask, if you accept the above, is it not the case that some TA i.e. trendlines, pullbacks, S&R etc encompass a form of statistical analysis? Are these ‘lines’ not are an extremely crude, colouring-book approach to a set of more complex principles?
The question then, imho, becomes not one of whether TA is a nonsense, or mumbo jumbo, or witchcraft, or tea leaf reading. It becomes a question of the degree of rigorousness that should be applied before analysis can be relied upon. Touching on Spec-Ks notion of “fine level” TA trading, which I’m assuming is looking at bounces or patterns in isolation etc over the very short-term/watching specific levels in great detail, to me, this means that all you’re doing is using a very crude form of statistical analysis on a very small sample size. From this, can we not infer that over the longer term i.e. a more significant sample size, TA should become more accurate an indicator?
Now I ask what exactly is TA an indicator of? Well if we go back to TA being a crude form of statistical analysis which if carried properly would give us some sort of quantitative definition or framework to assess a probability distribution which in turn, in the context of trading, is a function of macroeconomic variables, then doesn’t TA give us a crude weighting of macroeconomic outlook- a crude weighting of which the performance of the multitude of components of our instruments pricing mechanism? So is TA not also a very crude for of macro-economic analysis? Well in this case the answer lies in the macro pricing mechanism itself.
Now me personally, as an accountant, I’m all about pricing and how much something costs. Side-stepping the whole discussion about why this is probably a bad mentality to have in regards to trading, in particular short term trading (especially so for someone of my unexceptional intellect), we have to think about what the pricing mechanisms are and their purpose and their structure.
In contrast to the whole statistical majiggy which is based on analysis of past prices for assumptions of the future- the main concern of detractors from this type of approach as we’re in a dynamic environment - pricing is based on assumptions about the future (which ironically are somewhat comprised by past events). As far as I’m aware, the best statistical framework from which to approach pricing is non-linear regression and my quant knowledge falls short of this stuff. For the rest of the uninitiated, what I do to compensate for my shortcomings is to assume a linear approach e.g.
P= a(?) +bx +b2x +b2x +b4x.... with however many variables you can shake a stick at.
And then I assume that that at any given time, one variable, or perhaps a small set of variables will have more weighting over the price than at others. This is evident at times like QE announcements etc when prices rip upwards despite Gadaffi saying he’s cutting off the world’s oil supply or whatever.
Now without getting into what the variables, which differ from instrument to instrument -but let’s say that they’re heavily interest rate dependant - are, if the market is priced by these expectations and it’s trending or has a major reversal, hasn’t TA captured the majority opinion in the case of the former and a fundamental shift in the case of the latter. Which variable it is would possibly be decipherable from the technical news sources but it’s quite irrelevant if you’re watching the resultant isn’t it? So does TA allow us to infer where the pricing mechanism is taking us? Is this linked into market psychology or to technical due diligence and complex analysis? I would probably argue yes in the short term (shake outs etc which brings us back to intra-day gaming etc) for the former and longer term for the latter which is where we would consider institutional and real money positioning.
And so we come full circle with each and every one of these concepts being inexorably linked to the other.
What is the point of all of this you ask? Who knows? I was very bored and once I got started I couldn’t stop. As a side point, I’m sure that the only thing that hedge funds do is introduce volatility into the markets and I’d love someone who has an opinion on this to start a discussion on it some time and give me something to ponder on.
Now this is all likely absolute b0llocks and I have no idea of how to apply any of the above to the coin flip scenario but why bother with it anyway? If you’re into trading randomly, for me, you’re better off with a time interval based entry with the prevailing trend. This way you’ve got the distribution of returns on your side (shortmed-med term) and the event driven weighting of the macro-economic variables in your side (med-long term) too... until you haven’t.
All of my (possibly and probably rubbish) opinion is based on what I would say is short-med to long term trading so I’m very interested in what people have to say about random trading in the short term where I think you have to learn and think or die. I’m also interested in any discussion on the above obviously. Not now though, I’m going out and will probably be smashed in an hour.
Why did I write any of this?
Scose
Scose’s musing and questions/points for discussion are as follows:
All financial markets are driven by macroeconomic activity.
I'm sure the vast majority will agree that they are (but this is trade2win after all and sometimes I think that the 200ma is give more significance than “funnymentals”) and so we are left with the question of what signifies the underlying economic trend or a change therein and what constitutes noise. The noise is what we’re referring to as a random variable/random price behaviour but what it really is imho, is the intra-day gaming which pushes the natural supply/demand relationship out of a conceptual (and I'm almost certain non-existent) equilibrium and into relative extremes. If this is the case then these extremes can be upper or lower bound and so we're left with what is essentially an unknown quantity which may or may not be treated as a random variable. But... people make a living from assessing and trading from this type of price movement so it's likely not random. Couple that with my favourite subject which is EMH and arbing (oomph!) and other pricing mechanisms and you're left with a market view that would have Salvador Dali keeled over the sh1tter with nausea. So does Scose agree that he short term market is random? Yes and no but also yes and no. Digging a little deeper, insofar as a multitude of parties being required to buy/sell at any given time, yes, but the functions of basic auction theory and methods of liquidity provisiorrrr prrrrrovide a structure of sorts, so no. Then you have front running algo stuff to consider. This b0llocks is imho, where the short term trading opps are and where the inefficiency is. Looking at anything else short-term and you’re kidding yourself.
Now that we had added a sufficient degree of over-thinking the most basic facets of price movement, let us now move on to what constitutes randomness. This word is bandied about rather often on t2w and I’m not sure whether I’m being ignorant, everyone is lulzy or the word just isn’t really understood by some. The word random is tossed out and the sacrilege brigade fly off the handle and start Ross Hook the cheeks, auction hammer the faces or dojify (crucify + doji) those who dare utter the dreaded word. Statistics 101 or whatever the UK equivalent is (probably year 10/11 from some things I’ve seen) tells you that random doesn’t necessarily mean even distributed. What if at any given time our variable has a 60% chance of going one way and a 40% chance of going the other? Is that still not random? What if our random variable is normally distributed or leptokurtic (the concept that Robster w@nks over) or even displays some skew? Surely this is not outside the realms of possibility is it? If the markets are indeed driven by macro-economic trends which take time to play out wouldn’t we expect more people to be on one side of the trade than the other as they manage exposure or future expectations? Wouldn’t this effect supply and demand equilibrium? Wouldn’t this mean you’re more likely to make a profit buying rather than selling or vice versa as the macro case may be? Doesn’t this mean skewed returns? Doesn’t a skew in returns translate to a skewed probability distribution? Would this skewed distribution exhibit leptokurtosis (not going to go into why here)? How would all of this exhibit itself on a price chart? It would look like a trend wouldn’t it? Isn’t a trend just like a bull market or whatever the TA term is? Aren’t pullbacks etc extremely similar to how a chart of the inevitable outcome of a series of non-evenly distributed random events would look? While prices may or may not be random in and of themselves, given the above, shouldn’t a statistical process be a good enough proxy for analytical purposes-for providing a framework or approach to understanding?
So I ask, if you accept the above, is it not the case that some TA i.e. trendlines, pullbacks, S&R etc encompass a form of statistical analysis? Are these ‘lines’ not are an extremely crude, colouring-book approach to a set of more complex principles?
The question then, imho, becomes not one of whether TA is a nonsense, or mumbo jumbo, or witchcraft, or tea leaf reading. It becomes a question of the degree of rigorousness that should be applied before analysis can be relied upon. Touching on Spec-Ks notion of “fine level” TA trading, which I’m assuming is looking at bounces or patterns in isolation etc over the very short-term/watching specific levels in great detail, to me, this means that all you’re doing is using a very crude form of statistical analysis on a very small sample size. From this, can we not infer that over the longer term i.e. a more significant sample size, TA should become more accurate an indicator?
Now I ask what exactly is TA an indicator of? Well if we go back to TA being a crude form of statistical analysis which if carried properly would give us some sort of quantitative definition or framework to assess a probability distribution which in turn, in the context of trading, is a function of macroeconomic variables, then doesn’t TA give us a crude weighting of macroeconomic outlook- a crude weighting of which the performance of the multitude of components of our instruments pricing mechanism? So is TA not also a very crude for of macro-economic analysis? Well in this case the answer lies in the macro pricing mechanism itself.
Now me personally, as an accountant, I’m all about pricing and how much something costs. Side-stepping the whole discussion about why this is probably a bad mentality to have in regards to trading, in particular short term trading (especially so for someone of my unexceptional intellect), we have to think about what the pricing mechanisms are and their purpose and their structure.
In contrast to the whole statistical majiggy which is based on analysis of past prices for assumptions of the future- the main concern of detractors from this type of approach as we’re in a dynamic environment - pricing is based on assumptions about the future (which ironically are somewhat comprised by past events). As far as I’m aware, the best statistical framework from which to approach pricing is non-linear regression and my quant knowledge falls short of this stuff. For the rest of the uninitiated, what I do to compensate for my shortcomings is to assume a linear approach e.g.
P= a(?) +bx +b2x +b2x +b4x.... with however many variables you can shake a stick at.
And then I assume that that at any given time, one variable, or perhaps a small set of variables will have more weighting over the price than at others. This is evident at times like QE announcements etc when prices rip upwards despite Gadaffi saying he’s cutting off the world’s oil supply or whatever.
Now without getting into what the variables, which differ from instrument to instrument -but let’s say that they’re heavily interest rate dependant - are, if the market is priced by these expectations and it’s trending or has a major reversal, hasn’t TA captured the majority opinion in the case of the former and a fundamental shift in the case of the latter. Which variable it is would possibly be decipherable from the technical news sources but it’s quite irrelevant if you’re watching the resultant isn’t it? So does TA allow us to infer where the pricing mechanism is taking us? Is this linked into market psychology or to technical due diligence and complex analysis? I would probably argue yes in the short term (shake outs etc which brings us back to intra-day gaming etc) for the former and longer term for the latter which is where we would consider institutional and real money positioning.
And so we come full circle with each and every one of these concepts being inexorably linked to the other.
What is the point of all of this you ask? Who knows? I was very bored and once I got started I couldn’t stop. As a side point, I’m sure that the only thing that hedge funds do is introduce volatility into the markets and I’d love someone who has an opinion on this to start a discussion on it some time and give me something to ponder on.
Now this is all likely absolute b0llocks and I have no idea of how to apply any of the above to the coin flip scenario but why bother with it anyway? If you’re into trading randomly, for me, you’re better off with a time interval based entry with the prevailing trend. This way you’ve got the distribution of returns on your side (shortmed-med term) and the event driven weighting of the macro-economic variables in your side (med-long term) too... until you haven’t.
All of my (possibly and probably rubbish) opinion is based on what I would say is short-med to long term trading so I’m very interested in what people have to say about random trading in the short term where I think you have to learn and think or die. I’m also interested in any discussion on the above obviously. Not now though, I’m going out and will probably be smashed in an hour.
Why did I write any of this?
Scose
Spec-K Trading
Junior member
- Messages
- 14
- Likes
- 1
Reading that is certainly sending me to the kegerator for a shot of courage!
Now this is all likely absolute b0llocks and I have no idea of how to apply any of the above to the coin flip scenario but why bother with it anyway? If you’re into trading randomly, for me, you’re better off with a time interval based entry with the prevailing trend. This way you’ve got the distribution of returns on your side (shortmed-med term) and the event driven weighting of the macro-economic variables in your side (med-long term) too... until you haven’t.
Scose
I agree with what you say, err, I think
I thought that I was talkative!
Your paragraph that starts "Now this is all likely absolute b0llocks", apart from the unfortunate start, is what I think that I agree with most. You've got to go with the prevailing trend because you have to, at least, try get it right but, once done, it is random. My (current) response to a bad trade is not to look at it---cut it and be ready to start again. That happens more frequently than not but it makes me a profitable (and mediocre) trader. Thank the Lord that it is a hobby! I'd hate to maintain the wife and kids on this!
scose-no-doubt
Veteren member
- Messages
- 4,630
- Likes
- 954
And now back to our scheduled program
Nob'eads (beat. Ro Jista) - Dick Limerick Academy - YouTube
Nob'eads (beat. Ro Jista) - Dick Limerick Academy - YouTube
Spec-K Trading
Junior member
- Messages
- 14
- Likes
- 1
My trading system doesn't do that, and you might want to explore alternatives. Tight stops on "precisely timed entries" combined with "let your profitable trades run" tends to lead to lots of smaller losers and winners that aren't so big because trades run too long and greed and fear causes poor sells that don't capture big moves. (I'm not saying you are doing that, only that it's pretty common.) Overall, such systems tend to struggle to have an average per trade profit. It's pertinent to this thread because such methods (I call it generally the "let profits run and cut losses short" method) underestimate the degree of randomness in price behavior. Precise entries fail to eliminate a negative price excursion much more often that we expect.
My system takes way more winners than losers, with wInners much smaller than the less common loser. It takes positions low (very low) in the immediate or setting up range, or off the primary trendline (and the system only selects stocks for trading with excellent uptrending behavior in the intermediate and longer term ). The volatility and recent ranging behavior determine how much profit to go for, but unless the stock is highly volatile (>5%) it's pretty small (a handful or less of percentage points). The system calls for spreading out money among a significant number of simultaneous trades to smooth out the equity curve overall, and as trades exit (which happens frequently and fast), immediate reentry, with compounding. This methodology keeps money working efficiently by moving it to the best setups at the best times for quick high probability profits, rather than waiting around for "profits to run", yet allows for "random" wandering of price in a negative excursion without taking a quick loss. Losses are evil, and worth waiting a bit for the trades that don't work quickly to recover from, even if it takes months. While slow to win (or lose, it happens!) trades are developing, most of the money is popping out winner after winner, with compounding.
It's a different way to trade. While my system tries quite hard to buy in situations where price is likely to move up at least a bit in short order, it doesn't hang it's hat on that attempt, due to the reality of short term randomness in price action.
Shakone
Senior member
- Messages
- 2,460
- Likes
- 665
Scose, interesting post, and I hope we'll see more of those. Think your idea of TA as a crude form of stat analysis is excellent. With regard to funnymentals, I'm one of those weirdos who believe that people move price. Fundamentals move people, but whether they move the price, depends upon what else is moving the people and thus the price at the time.
Spec-K, getting average profit lower than loss is fine if win rate is high, although I personally don't feel that comfortable with it. 'Aiming' for a lower profit than loss though, is a problem in my view. You should always be trying to aim for a large profit and leave that possibility open. That's one of the reasons I dislike fixed arbitrary targets.
Spec-K, getting average profit lower than loss is fine if win rate is high, although I personally don't feel that comfortable with it. 'Aiming' for a lower profit than loss though, is a problem in my view. You should always be trying to aim for a large profit and leave that possibility open. That's one of the reasons I dislike fixed arbitrary targets.
Spec-K Trading
Junior member
- Messages
- 14
- Likes
- 1
I know my system approach raises hackles; it's different and contrarian re: the general trading "rules". However, "you should always" is rarely correct in trading, and it's not accurate here (=> always aim for a large profit). There are many ways to get an edge (positive expectation), and I'm certain I have one while violating this trading maxim.
In the abstract, going for the large profit on a per trade basis means riding the price increase driven by the laterally upward moving price range (while price gyrate up and down in that range). That's a relatively slow boat ride to profits. It ignores the values of time (quick successive profits => quick compounding => faster growth in equity) and trade probabilities (every additional several percentage you are putting on your exit target dramatically reduces probability of success). If there's discretion in when to exit when going for large per trade profits, such an approach challenges our emotionally driven tendencies as well (greed => hold out for more, fear => exit at a loss before more is lost).
My system's approach grabs part (the lower, higher probability part) of one of those moves in the ranging behavior, then moves the money (with profits) immediately to grab another such move on another stock; it leverages the volatility inside the ranging behavior to get a much faster profit than riding the price up as the range moves laterally upward. It compounds that return into another similar high probability quick hit trade. It's violates those traditional rules of cutting losses short (which fails to manage random price excursions and general volatility well) and letting profits run, at least on a per trade basis (on a macro basis it salutes them, but uses very different per trade rules to achieve them overall). It's different and maybe relatively unique. Years of working systems development and testing looking for a method that was consistent across markets got me to it. I'll never trade stocks any other way. I have no problem with getting out of a position and then noticing later that that stock has gone to the moon; most don't, and hanging on for the larger win is generally less efficient in terms of profit generation than moving the money on to a new trade.
The goal of letting winners run and the goal of taking small profits and finding new excellent set ups and entering a fresh trade immediately are identical; maximize rate of returns. In my experience the latter approach is easier and more consistent.
In the abstract, going for the large profit on a per trade basis means riding the price increase driven by the laterally upward moving price range (while price gyrate up and down in that range). That's a relatively slow boat ride to profits. It ignores the values of time (quick successive profits => quick compounding => faster growth in equity) and trade probabilities (every additional several percentage you are putting on your exit target dramatically reduces probability of success). If there's discretion in when to exit when going for large per trade profits, such an approach challenges our emotionally driven tendencies as well (greed => hold out for more, fear => exit at a loss before more is lost).
My system's approach grabs part (the lower, higher probability part) of one of those moves in the ranging behavior, then moves the money (with profits) immediately to grab another such move on another stock; it leverages the volatility inside the ranging behavior to get a much faster profit than riding the price up as the range moves laterally upward. It compounds that return into another similar high probability quick hit trade. It's violates those traditional rules of cutting losses short (which fails to manage random price excursions and general volatility well) and letting profits run, at least on a per trade basis (on a macro basis it salutes them, but uses very different per trade rules to achieve them overall). It's different and maybe relatively unique. Years of working systems development and testing looking for a method that was consistent across markets got me to it. I'll never trade stocks any other way. I have no problem with getting out of a position and then noticing later that that stock has gone to the moon; most don't, and hanging on for the larger win is generally less efficient in terms of profit generation than moving the money on to a new trade.
The goal of letting winners run and the goal of taking small profits and finding new excellent set ups and entering a fresh trade immediately are identical; maximize rate of returns. In my experience the latter approach is easier and more consistent.
DionysusToast
Legendary member
- Messages
- 5,965
- Likes
- 1,501
Agreed - one of the problems with backtesting is that it doesn't deal with context.
Look at the Euro problems of late. I still feel the pain of 23 ticks slippage on the Bund when Angela Merkel (the gobsh!te) let rip with some off-hand comment that made the markets go ga-ga.
Your backtesting could work/fail just because of such occurences when in real time you could manage scenarios like this at your discretion. Tough to see how you can backtest this sort of thing.
