Running hundreds of trading strategies concurrently

Problem I see with multiple strategies, the winners get diluted by the losers.
For example, it's not difficult to be making excellent money on one hand, only to see the losing positions rob you of your edge.
Like diversifying into multiple stocks, it might add for safety, but like an insurance policy, there is a price to pay.
I would rather run maybe one good algo than numerous algos, or for those who are particulaly gifted, maybe a handful of algos.
Just my opinion, I'm no expert in this and aware there is heaps of different ways to make money.
 
With so many strategies going around, possible to apply Pareto's 80/20 rule? 20% of the strategies that account for 80% of the profits.:cool: Just a thought. I am using trend following with 50 SMA on daily data. The simplest strategy that can be.:sneaky:
 
Peter8519 said:
With so many strategies going around, possible to apply Pareto's 80/20 rule? 20% of the strategies that account for 80% of the profits.:cool: Just a thought. I am using trend following with 50 SMA on daily data. The simplest strategy that can be.:sneaky:
More...

What products are you running that on? Trend following doesn't work well for equities.
 
traider said:
What products are you running that on? Trend following doesn't work well for equities.
More...
I am using Amibroker. A trend will not last forever and one must know when to get in and out. Not a recommendation but an example.
Trend that has broken: CBOE, CAT, TTGT
Still trending: AMED, WCG, UVE
 
Peter8519 said:
I am using Amibroker. A trend will not last forever and one must know when to get in and out. Not a recommendation but an example.
Trend that has broken: CBOE, CAT, TTGT
Still trending: AMED, WCG, UVE
More...

What do you mean, if you are fully systematic, trend is well defined and so are entries and exits. If you run tests on a huge bunch of stocks, u will find that TF sucks compared to just BuyHold. Are you selecting only stuff that you think will perform well with trend rules?
Or u just filter and then u use your discretion which means you are not fully systematic.
 
traider said:
What do you mean, if you are fully systematic, trend is well defined and so are entries and exits. If you run tests on a huge bunch of stocks, u will find that TF sucks compared to just BuyHold. Are you selecting only stuff that you think will perform well with trend rules?
Or u just filter and then u use your discretion which means you are not fully systematic.
More...
Do a frequency plot on the number of days above 50 SMA and that will give an important clue on when to get it.
 
I do long only. I use multiple rules for exit. Optimize the number of days above 50SMA. TF NO sucks. :D
In the game of trading, it's all about winning odds. :cool:
 
themickey said:
Problem I see with multiple strategies, the winners get diluted by the losers.
For example, it's not difficult to be making excellent money on one hand, only to see the losing positions rob you of your edge.
Like diversifying into multiple stocks, it might add for safety, but like an insurance policy, there is a price to pay.
I would rather run maybe one good algo than numerous algos, or for those who are particulaly gifted, maybe a handful of algos.
Just my opinion, I'm no expert in this and aware there is heaps of different ways to make money.
More...
I try and think in terms of capital utilization. Let's say I have algos A, B and C with the following stats:

A: Avg trade result: 3%
Avg trades per month: 10
Avg holding period: 5 days

B: Avg trade result: 8%
Avg trades per month: 2
Avg holding period: 5 days

C: Avg trade result: 4%
Avg trades per month: 8
Avg holding period: 5 days

Algo B is the top performing algo but it does not trade as often. A and C have similar performance metrics. I don't want to only trade algo B because my capital utilization will be unnecessarily low. What I would do however is rank the algos. In this case, algo B would have a rank of 1 and A and C would have a rank of 2 which means a signal from algo B takes priority over a signal from algo A and C. If a signal from B is encountered but there is no more capital available because of open trades in A and C, trades from A and C will be partially or fully closed to make room for a trade in algo B.
 
One of the main purposes of running multiple systems/strategies/algos, whatever someone wants to call them, is simply to minimize the basket's DD.

If your 'best' strategy got whacked a couple times a month with a large loser but you had another non-correlated strategy that always offset the losers of the 'best' strategy why wouldn't you allocate whatever funds were required to it?
 
You must log in or register to reply here.
Back
Top