Holy Grail
- Thread starter Cheese
- Start date
...my post was not aimed negatively at you, and I intended no offense. I don't personally know any of the posters here, they're just electronic phantoms to me. There are days when some posts seem to be written by word salad generators. I was merely drawing an historical parallel which I thought was amusing. In fact, many of our most loquatious posters don't need alternate personas, they're already talking to themselves. Best regards. - Mike
"It appears you are not accounting for the performance of either the market or the stock's sector. People that follow the 50/30/20 rule would say you are operating with only 20% of the available data."
You are quite correct in saying that my trading doesn't directly incorporate either the broad market or the stock's sector. The closest I come to broad market analysis is that I do look to make sure the S&P and/or NASDAQ tracking stocks are in an uptrend (in the same way I do for the stocks I want to buy). Dunno if that's an especially sound methodology for keeping me on the 'right side of the market'; but because I'm using a somewhat longer timeframe than some traders, my personal bias (and this may be totally wrong) is that if the market begins to tip over, I'll have a little bit of time to exit a given trade. So far as looking at sectors, I guess I probably should be looking to see the best performing sectors, then pick stocks from those, but I find it difficult to do so precisely because I tend to put stocks on my watchlist just at the time they first begin to move to the upside. That means I often find 'the horses leaving the starting gate are still running neck and neck'. I'd appreciate your comments or suggestions re. my opinions--anything which might refine the general model is valuable!
"I suspect your possible "blow out" is a feature of not following the market or sector as closely as you are the stock. You can easily fix this by applying your same analysis to the other contributing elements. It will also simplify the number of stocks you have to look at."
Just to clarify, I've never actually 'blown out' using this system (though I have on other occasions!) because I diversify and run correlation analysis against each stock to ensure it is weekly correlated with other stocks in the portfolio. My biggest threat, I think, would be for all stocks to take a hit as the broad market dropped suddenly. I confess I'm not sure how to prevent this, other than by diversification and limiting risk in each position. Ideas?
"You also look at your previous trades to determine if the current stock is going to perform, when a better indication might be the current stocks past behavior."
This is one thing I didn't communicate well, and I apologize. My backtesting is not done on a given candidate stock; rather, the general model was run on a randomly selected sample of stocks (fifty, to be precise) in the NYSE market. The only common criterion was that all trade over 1M shares/day, on average. I then selected the period Jan 2000 to mid 2003 in order to encompass the dire drop from Feb 2000 on, and the various upswings since then, assuming that this period, if anything, would savage a 'long system'. Basically, in each case the system kept me out except when the broad market was recovering, and limited loss on those trades which went sour (for some reason, mainly between April and July 2002, when many stocks fizzled).
Thank you very much for your comments, and I'd like to hear your remarks concerning what I just wrote.
You are quite correct in saying that my trading doesn't directly incorporate either the broad market or the stock's sector. The closest I come to broad market analysis is that I do look to make sure the S&P and/or NASDAQ tracking stocks are in an uptrend (in the same way I do for the stocks I want to buy). Dunno if that's an especially sound methodology for keeping me on the 'right side of the market'; but because I'm using a somewhat longer timeframe than some traders, my personal bias (and this may be totally wrong) is that if the market begins to tip over, I'll have a little bit of time to exit a given trade. So far as looking at sectors, I guess I probably should be looking to see the best performing sectors, then pick stocks from those, but I find it difficult to do so precisely because I tend to put stocks on my watchlist just at the time they first begin to move to the upside. That means I often find 'the horses leaving the starting gate are still running neck and neck'. I'd appreciate your comments or suggestions re. my opinions--anything which might refine the general model is valuable!
"I suspect your possible "blow out" is a feature of not following the market or sector as closely as you are the stock. You can easily fix this by applying your same analysis to the other contributing elements. It will also simplify the number of stocks you have to look at."
Just to clarify, I've never actually 'blown out' using this system (though I have on other occasions!) because I diversify and run correlation analysis against each stock to ensure it is weekly correlated with other stocks in the portfolio. My biggest threat, I think, would be for all stocks to take a hit as the broad market dropped suddenly. I confess I'm not sure how to prevent this, other than by diversification and limiting risk in each position. Ideas?
"You also look at your previous trades to determine if the current stock is going to perform, when a better indication might be the current stocks past behavior."
This is one thing I didn't communicate well, and I apologize. My backtesting is not done on a given candidate stock; rather, the general model was run on a randomly selected sample of stocks (fifty, to be precise) in the NYSE market. The only common criterion was that all trade over 1M shares/day, on average. I then selected the period Jan 2000 to mid 2003 in order to encompass the dire drop from Feb 2000 on, and the various upswings since then, assuming that this period, if anything, would savage a 'long system'. Basically, in each case the system kept me out except when the broad market was recovering, and limited loss on those trades which went sour (for some reason, mainly between April and July 2002, when many stocks fizzled).
Thank you very much for your comments, and I'd like to hear your remarks concerning what I just wrote.
Okay, that's fair enough. I'll be the first to admit the system is imperfect, and I probably could tweak it to eke out more profit. My concern is that I don't want to inadvertently fit the system to a given data set, and want to keep it as 'generally useful' as I can.
Any ideas on the 'early exit'? You're quite right that it's 'early' if I follow the exit rule of 'leave when the stock is overbought'. My custom is to use this as a 'heads up', but to then manually intervene and manage the exit myself by merely tightening the stop and keeping an eye on the stock. Any ideas to automate this a bit?
What are the 'other factors'? I'd be grateful for whatever light your own experience can shed on this.
VT
> To Steve
> "It appears you are not accounting for the performance of either the market
> or the stock's sector. People that follow the 50/30/20 rule would say you
> are operating with only 20% of the available data."
>
> You are quite correct in saying that my trading doesn't directly incorporate
> either the broad market or the stock's sector. The closest I come to broad
> market analysis is that I do look to make sure the S&P and/or NASDAQ
> tracking stocks are in an uptrend (in the same way I do for the stocks I
> want to buy). Dunno if that's an especially sound methodology for keeping me
> on the 'right side of the market'; but because I'm using a somewhat longer
> timeframe than some traders, my personal bias (and this may be totally
> wrong) is that if the market begins to tip over, I'll have a little bit of
> time to exit a given trade. So far as looking at sectors, I guess I probably
> should be looking to see the best performing sectors, then pick stocks from
> those, but I find it difficult to do so precisely because I tend to put
> stocks on my watchlist just at the time they first begin to move to the
> upside. That means I often find 'the horses leaving the starting gate are
> still running neck and neck'. I'd appreciate your comments or suggestions
> re. my opinions--anything which might refine the general model is valuable!
The market, sector, and stock will ideally have strong/healthy trends. Strong meaning price movement backed up by volume. If the market or sector have weak trends, you will generally not want to tie up your capital in the stock (assuming it has a positive beta value). You might try just trading indices for a bit, you will quickly gain awareness of how the market is doing. Even something simple like diamonds will get you focused. Being aware of how a stocks sector is doing is very important, a strong stock may be the last to fall in its sector, but fall it often will. I consider this scenario the most likely cause of any "blow out" you might encounter, and it is easy to avoid.
> "I suspect your possible "blow out" is a feature of not following the market
> or sector as closely as you are the stock. You can easily fix this by
> applying your same analysis to the other contributing elements. It will also
> simplify the number of stocks you have to look at."
>
> Just to clarify, I've never actually 'blown out' using this system (though I
> have on other occasions!) because I diversify and run correlation analysis
> against each stock to ensure it is weekly correlated with other stocks in
> the portfolio. My biggest threat, I think, would be for all stocks to take a
> hit as the broad market dropped suddenly. I confess I'm not sure how to
> prevent this, other than by diversification and limiting risk in each
> position. Ideas?
There are various ways of measuring market strength, generally volume is the key. Minimizing correlation coefficients is great, more than most people are willing to do. Choosing an optimal subset of stocks is a TSP, often solved via a best guess, heh. Travelling Salesman Problems plague most attempts at optimization, I find it fascinating solutions are generally not provable to be optimal. Alan Turing would have had a good crack at them.
>
> "You also look at your previous trades to determine if the current stock is
> going to perform, when a better indication might be the current stocks past
> behavior."
>
> This is one thing I didn't communicate well, and I apologize. My backtesting
> is not done on a given candidate stock; rather, the general model was run on
> a randomly selected sample of stocks (fifty, to be precise) in the NYSE
> market. The only common criterion was that all trade over 1M shares/day, on
> average. I then selected the period Jan 2000 to mid 2003 in order to
> encompass the dire drop from Feb 2000 on, and the various upswings since
> then, assuming that this period, if anything, would savage a 'long system'.
> Basically, in each case the system kept me out except when the broad market
> was recovering, and limited loss on those trades which went sour (for some
> reason, mainly between April and July 2002, when many stocks fizzled).
The problem with backtesting is that your data sample is unlikely to contain examples of stocks that have been delisted. A friend of mine lost millions because his sample data had no stocks in it that went bust. I would not consider your test data to be representative of the real data unless you are *certain* that it is. Most commercially available datasources weed out delisted stocks, making them worthless sources for 'backtesting data'.
> Thank you very much for your comments, and I'd like to hear your remarks
> concerning what I just wrote.
I would advise you to get as comfortable with market and sector strength as you are with stocks. If you like building systems (and who doesn't?), at least you will know your test data is complete. There is DJIA data that goes way back.
The longer your investment period, the more the 50/30/20 rule comes in to play. The other 80% is well within your reach.
Steve.
> "It appears you are not accounting for the performance of either the market
> or the stock's sector. People that follow the 50/30/20 rule would say you
> are operating with only 20% of the available data."
>
> You are quite correct in saying that my trading doesn't directly incorporate
> either the broad market or the stock's sector. The closest I come to broad
> market analysis is that I do look to make sure the S&P and/or NASDAQ
> tracking stocks are in an uptrend (in the same way I do for the stocks I
> want to buy). Dunno if that's an especially sound methodology for keeping me
> on the 'right side of the market'; but because I'm using a somewhat longer
> timeframe than some traders, my personal bias (and this may be totally
> wrong) is that if the market begins to tip over, I'll have a little bit of
> time to exit a given trade. So far as looking at sectors, I guess I probably
> should be looking to see the best performing sectors, then pick stocks from
> those, but I find it difficult to do so precisely because I tend to put
> stocks on my watchlist just at the time they first begin to move to the
> upside. That means I often find 'the horses leaving the starting gate are
> still running neck and neck'. I'd appreciate your comments or suggestions
> re. my opinions--anything which might refine the general model is valuable!
The market, sector, and stock will ideally have strong/healthy trends. Strong meaning price movement backed up by volume. If the market or sector have weak trends, you will generally not want to tie up your capital in the stock (assuming it has a positive beta value). You might try just trading indices for a bit, you will quickly gain awareness of how the market is doing. Even something simple like diamonds will get you focused. Being aware of how a stocks sector is doing is very important, a strong stock may be the last to fall in its sector, but fall it often will. I consider this scenario the most likely cause of any "blow out" you might encounter, and it is easy to avoid.
> "I suspect your possible "blow out" is a feature of not following the market
> or sector as closely as you are the stock. You can easily fix this by
> applying your same analysis to the other contributing elements. It will also
> simplify the number of stocks you have to look at."
>
> Just to clarify, I've never actually 'blown out' using this system (though I
> have on other occasions!) because I diversify and run correlation analysis
> against each stock to ensure it is weekly correlated with other stocks in
> the portfolio. My biggest threat, I think, would be for all stocks to take a
> hit as the broad market dropped suddenly. I confess I'm not sure how to
> prevent this, other than by diversification and limiting risk in each
> position. Ideas?
There are various ways of measuring market strength, generally volume is the key. Minimizing correlation coefficients is great, more than most people are willing to do. Choosing an optimal subset of stocks is a TSP, often solved via a best guess, heh. Travelling Salesman Problems plague most attempts at optimization, I find it fascinating solutions are generally not provable to be optimal. Alan Turing would have had a good crack at them.
>
> "You also look at your previous trades to determine if the current stock is
> going to perform, when a better indication might be the current stocks past
> behavior."
>
> This is one thing I didn't communicate well, and I apologize. My backtesting
> is not done on a given candidate stock; rather, the general model was run on
> a randomly selected sample of stocks (fifty, to be precise) in the NYSE
> market. The only common criterion was that all trade over 1M shares/day, on
> average. I then selected the period Jan 2000 to mid 2003 in order to
> encompass the dire drop from Feb 2000 on, and the various upswings since
> then, assuming that this period, if anything, would savage a 'long system'.
> Basically, in each case the system kept me out except when the broad market
> was recovering, and limited loss on those trades which went sour (for some
> reason, mainly between April and July 2002, when many stocks fizzled).
The problem with backtesting is that your data sample is unlikely to contain examples of stocks that have been delisted. A friend of mine lost millions because his sample data had no stocks in it that went bust. I would not consider your test data to be representative of the real data unless you are *certain* that it is. Most commercially available datasources weed out delisted stocks, making them worthless sources for 'backtesting data'.
> Thank you very much for your comments, and I'd like to hear your remarks
> concerning what I just wrote.
I would advise you to get as comfortable with market and sector strength as you are with stocks. If you like building systems (and who doesn't?), at least you will know your test data is complete. There is DJIA data that goes way back.
The longer your investment period, the more the 50/30/20 rule comes in to play. The other 80% is well within your reach.
Steve.
You're quite correct, my dataset doesn't include delisted stocks. I'll have to look at this issue further; at the moment my best defense against seeing a position drop like a rock is that I'm diversified, so (I hope!) while any one stock might tank, I wouldn't suffer a crippling loss--just one that drove me to drink for an evening or two. My other defense lies mainly in picking stocks which are in an uptrend--I'd expect that a stock would begin to trade with less optimism as soon as any rumors of serious health issues began to revolve around the company. This is something I'm going to research further, and thanks for pointing it out.
You're right about looking at the broad market, and I'm going to see what effect sector analysis has upon the profitability of my stock selections using this system.
Thanks for these suggestions!
Alternatively I might just buy that 'green light/red light thingy.
Just kidding!!!
Vermonter
You're right about looking at the broad market, and I'm going to see what effect sector analysis has upon the profitability of my stock selections using this system.
Thanks for these suggestions!
Alternatively I might just buy that 'green light/red light thingy.
Just kidding!!!
Vermonter
Yoda: "Much anger in this one."
Steve.