Judgmental Biases Continued
Trading System Testing
The degree-of-freedom bias is essentially the over use of system development parameters to fit a set of historical market data. I think this is somewhat erroneously named, but I understand why Tharp calls it what he does. He describes a degree-of-freedom as "a parameter that yields a different system for every value allowed." So for a moving average, each different number used gives a different value, giving infinite degrees of freedom. When developing systems, it is quite easy to keep changing moving average, MACD, or RSI indicators to get the perfect fit for the historical data.
I think I fell into this when I designed my candle system a few years ago. I remember being dissatisfied with the results and tinkering endlessly to improve performance. Tharp suggests that truly understanding the concept you are using will reduce the need to optimize the system. He also suggests that you think about various scenarios your system could encounter and how the system might be expected to behave. His final suggestion is to minimize the number of indicators used.
Postdictive error bias involves using information for testing that you would could only know in retrospect, like a price high, which you cannot know unless you are looking backwards. This one is sort of hard for me to conceptualize. Tharp says that exceptionally good testing results are probably a result of postdictive errors. That makes sense as the better your testing results are, the more likely they are to be a result of fitting your system to your data.
Lack of protection bias is really just having a system that is too risky by not preserving capital, not providing for an expedient exit if the trade goes against you, and having low expectancy (which is a great topic Tharp addresses later).
I know I did this when I first experimented with FX, and later I replicated it in my Roth IRA with stocks/ETFs. Through the magic of leverage I would risk as much as 20% of capital in the FX, and I would risk my entire Roth in a single position. While I considered these to be playing around in the market (I keep real retirement and investment monies elsewhere in much more sensible allocations), what I did was silly for real trading purposes. I want to be much more conservative and smart when I try this again in the FX game.
How you Trade the System
The gambler's fallacy bias is probably one of my favorites because it seems so common-sensical. This bias simply leads you to believe that because the market (or game or whatever) has been trending in a certain direction, it is "due" to change. So because the last 5 coin flips have been heads, the next has a higher chance to be tails, or because the last 4 days have been down on the Dow, the next day should have a higher chance of being an up day. To coincide with this belief is the tendency to wager a greater amount on that next iteration of the game, so you'll bet even more now since the market/game is more due for a change. Needless to say, this bias can take you out of the game very quickly.
I think I fell victim to this on the first FX go 'round. It just makes too much "sense"! Mostly this did me in because of the poor position sizing aspect of it. I'm finally learning that one should size their positions using a consistent methodology and not based on the idea of what the market should do, completely ignoring the element of randomness.
The conservative with profits but risky with losses bias is pretty much exactly what it says. I know I'm guilty of this one. The tendency is to take a win as soon as things go against you, but hold on to a loss in hope of it coming back. From later readings in the book, I've learned that taking a high percentage of losses is okay but only if you cut those losses short. I think this is highly related to the next bias.
"My Current Trade Must Be a Winner" bias is exactly what it says as well. This one is definitely a downfall for me, because it is extremely difficult to admit that you made a bad decision, or your analysis was flawed. Predictions speak to the ego and to be wrong is to take a hit to that ego. But as Tharp says "being right has nothing to do with making money."
The degree-of-freedom bias is essentially the over use of system development parameters to fit a set of historical market data. I think this is somewhat erroneously named, but I understand why Tharp calls it what he does. He describes a degree-of-freedom as "a parameter that yields a different system for every value allowed." So for a moving average, each different number used gives a different value, giving infinite degrees of freedom. When developing systems, it is quite easy to keep changing moving average, MACD, or RSI indicators to get the perfect fit for the historical data.
I think I fell into this when I designed my candle system a few years ago. I remember being dissatisfied with the results and tinkering endlessly to improve performance. Tharp suggests that truly understanding the concept you are using will reduce the need to optimize the system. He also suggests that you think about various scenarios your system could encounter and how the system might be expected to behave. His final suggestion is to minimize the number of indicators used.
Postdictive error bias involves using information for testing that you would could only know in retrospect, like a price high, which you cannot know unless you are looking backwards. This one is sort of hard for me to conceptualize. Tharp says that exceptionally good testing results are probably a result of postdictive errors. That makes sense as the better your testing results are, the more likely they are to be a result of fitting your system to your data.
Lack of protection bias is really just having a system that is too risky by not preserving capital, not providing for an expedient exit if the trade goes against you, and having low expectancy (which is a great topic Tharp addresses later).
I know I did this when I first experimented with FX, and later I replicated it in my Roth IRA with stocks/ETFs. Through the magic of leverage I would risk as much as 20% of capital in the FX, and I would risk my entire Roth in a single position. While I considered these to be playing around in the market (I keep real retirement and investment monies elsewhere in much more sensible allocations), what I did was silly for real trading purposes. I want to be much more conservative and smart when I try this again in the FX game.
How you Trade the System
The gambler's fallacy bias is probably one of my favorites because it seems so common-sensical. This bias simply leads you to believe that because the market (or game or whatever) has been trending in a certain direction, it is "due" to change. So because the last 5 coin flips have been heads, the next has a higher chance to be tails, or because the last 4 days have been down on the Dow, the next day should have a higher chance of being an up day. To coincide with this belief is the tendency to wager a greater amount on that next iteration of the game, so you'll bet even more now since the market/game is more due for a change. Needless to say, this bias can take you out of the game very quickly.
I think I fell victim to this on the first FX go 'round. It just makes too much "sense"! Mostly this did me in because of the poor position sizing aspect of it. I'm finally learning that one should size their positions using a consistent methodology and not based on the idea of what the market should do, completely ignoring the element of randomness.
The conservative with profits but risky with losses bias is pretty much exactly what it says. I know I'm guilty of this one. The tendency is to take a win as soon as things go against you, but hold on to a loss in hope of it coming back. From later readings in the book, I've learned that taking a high percentage of losses is okay but only if you cut those losses short. I think this is highly related to the next bias.
"My Current Trade Must Be a Winner" bias is exactly what it says as well. This one is definitely a downfall for me, because it is extremely difficult to admit that you made a bad decision, or your analysis was flawed. Predictions speak to the ego and to be wrong is to take a hit to that ego. But as Tharp says "being right has nothing to do with making money."
posted by Young Pip at 1:13 PM
0 Comments:
Post a Comment
<< Home