Rigid mechanical system traders: Up to this morning I had six straight wins, one more than the longest streak in my backtesting. The system triggered a trade this morning. Would you take it, fade it, or stand aside?
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Poll: Are you statistically superstitious?
- Thread starter hypostomus
- Start date
i'm no expert here, but...
doesn't that depend on if there is dependency or not? if one trade has nothing to do with another, why would the distribution of wins and losses decide which trades you take??
this has been discussed before on ET using coin flips as examples. if flipping a coin is 50/50, each flip is 50/50 regardless of how many times you had one side in a row. it may be rare to get 30 heads in a row, but that is the series. each coin flip was still 50/50 regardless of how many times in a row there was heads.
so in your example, unless you know your next trade has a relationship to previous trades and you expect a loser, you should take every trade.
doesn't that depend on if there is dependency or not? if one trade has nothing to do with another, why would the distribution of wins and losses decide which trades you take??
this has been discussed before on ET using coin flips as examples. if flipping a coin is 50/50, each flip is 50/50 regardless of how many times you had one side in a row. it may be rare to get 30 heads in a row, but that is the series. each coin flip was still 50/50 regardless of how many times in a row there was heads.
so in your example, unless you know your next trade has a relationship to previous trades and you expect a loser, you should take every trade.
Quote from hypostomus:
Rigid mechanical system traders: Up to this morning I had six straight wins, one more than the longest streak in my backtesting. The system triggered a trade this morning. Would you take it, fade it, or stand aside?
I would think that you should have more confidence in your method when it is in sync with the market. I think trades are rarely independent. They are dependent on the market conditions they were designed to exploit. If such market conditions are in place, exploit them.
Thunderdog
P.S. Your string of 6 winners in your backtesting has very little predictive value. Don't give "statistics" too much credit aside from explaining the past.
P.P.S. Bottom line: if the trades are not independent, then you should take the trade because your method is in sync with the prevailing market conditions. If the trades are independent, then past trades have absolutely no bearing on future trades. (I think this exists more in theory than in fact.) Either way, you should go with the trade.
Quote from WDGann:
Each trade is independent.
. //period
Also, most of the guys don't get what system trading is.
It's not about each individual trade. It's about the cumulative edge.
Use your head... THINK!
Sorry, WDGann, I disagree that each trade is necessarily independent, although I'm sure you read that somewhere. It might be independent and, then again, it might not. Consider this basic example. Suppose you are using a method that catches trends but loses money during a trading range. To the extent that you are trading during a choppy period, your trade results are not exactly independent of one another because they are feeding off the same choppy environment. Yes, I have thought about it.
Regards,
Thunderdog
..and please pardon my innumeracy. I am (most of the time) putting absolute faith in my system. However, the equity curve is an output of the testing, and should have just as much statistical predictive value as the entries, stops, reverses, and takes of the system. Therefore I assert with absolutely no backup that I should stand aside when the drawdown exceeds history and not push my luck in a winning streak. My position is that backtesting ferrets out profitable conditional probabilities unknowable a priori. If not that, what the hell is it doing?
Quote from hypostomus:
..and please pardon my innumeracy. I am (most of the time) putting absolute faith in my system. However, the equity curve is an output of the testing, and should have just as much statistical predictive value as the entries, stops, reverses, and takes of the system. Therefore I assert with absolutely no backup that I should stand aside when the drawdown exceeds history and not push my luck in a winning streak. My position is that backtesting ferrets out profitable conditional probabilities unknowable a priori. If not that, what the hell is it doing?
Groovy. Good luck with that.
...if you must insult, at least do it cleverly in a foreign language. Trou de cul, for example, would sound so much more erudite than bunghole. Enculeur des mouches would be even more graphic and amusing.
Anyway, I deeply regret that I agree with you, but it IS about the edge. I for one would appreciate a brief exposition from you on your approach to system trading. I am not trying to bait you, I am sincerely interested in your views.
I created the poll because I experienced a deep existential and ontological dilemma this morning, which I resolved by going back to bed. The trade was in fact a loser.
Anyway, I deeply regret that I agree with you, but it IS about the edge. I for one would appreciate a brief exposition from you on your approach to system trading. I am not trying to bait you, I am sincerely interested in your views.
I created the poll because I experienced a deep existential and ontological dilemma this morning, which I resolved by going back to bed. The trade was in fact a loser.