The two strategies were just an extreme example of how a losing strategy can improve a portfolio. You're certainly right that there is no such extreme anticorrelation in real strategies. But I have indeed some strategies with a correlation in the -0.2 .. -0.3 range, so this is not just a hypothetical question. Trend based and cycle based strategies on the same time frame are often slightly anticorrelated.
Making money with a losing strategy
- Thread starter jcl
- Start date
I haven't seen anything where that correlation is negative & also stable.
To clarify, I mean correlation between *strategies*.
Obviously there are related structures / instruments that offset neatly with known dynamics.
Good thinking. Study Parrondoâs paradox. Mike Harris had a blog last January about this. I thought he was joking but after reading your post I now think it is quite possible and maybe it is an edge of some sort, paradoxically enough:
http://www.priceactionlab.com/Blog/...ing-investment-strategies-to-win-the-paradox/
http://www.priceactionlab.com/Blog/...ing-investment-strategies-to-win-the-paradox/
Thanks for the link. Interesting idea.
Please note this thread (from 2007) on ET about Parrondo:
http://www.elitetrader.com/vb/showt...225&perpage=6&highlight=parrondo&pagenumber=2
Can't say it better than the last post (by Kevin Schmit).
If you find a strategy that has the properties of game "B2" you are better of trading it outright. No need to play loosing strategies.
http://www.elitetrader.com/vb/showt...225&perpage=6&highlight=parrondo&pagenumber=2
Can't say it better than the last post (by Kevin Schmit).
If you find a strategy that has the properties of game "B2" you are better of trading it outright. No need to play loosing strategies.
How come? Both strategies if played alone are losers by definition.
Just to avoid misunderstandings: The Parrondo paradox has nothing to do with improving a portfolio with a losing strategy.
The Parrondo paradox is based on a hypothetical game rule that gives you a large win when you've lost before. Portfolio management is just based on the correlation of the portfolio components, and does not require special game rules. The latter is thus very useful for trading, the former isn't.
The Parrondo paradox is based on a hypothetical game rule that gives you a large win when you've lost before. Portfolio management is just based on the correlation of the portfolio components, and does not require special game rules. The latter is thus very useful for trading, the former isn't.
On my test system, usually for days (very rarely like 1 in 30) if my win/loss hits >60%, mostly likely im in for a good day of positive.
If u have a system A with 80% and B with 40%, i'll say discard the 40% and work fully on the 80%.
Why complicate and mix it up ? Unless you say are LTCM and afraid that the prime brokerages or country party are following or copying your trades, moving a few billions in and out every day.
If u have a system A with 80% and B with 40%, i'll say discard the 40% and work fully on the 80%.
Why complicate and mix it up ? Unless you say are LTCM and afraid that the prime brokerages or country party are following or copying your trades, moving a few billions in and out every day.