Building Strategies: All Long/All Short vs Symmetrical

I’ve been evaluating the question of which is the “best” way to build strategies – to build them all one direction (i.e. only long or only short) or whether to build them symmetrically (i.e. both long and short). I would love to hear what others think and which one they have had the most success with in regards to building, testing, and real market performance. So far, I have only focused on symmetrical building under the philosophy that it allows me to approach the market without a bias and if the conditions are met going either direction, the algo takes/exits the trade. But I’m not sure that this is the “best” approach. Has anyone thoroughly tested both and what did they discover?

i always start with symmetrical approach for long and short. After i build basic pack of strategies, i am trying to find separate only long/short

also i generate 1 basic pack with asymetric approach

Thanks for sharing thoughts hankeys. Your approach is something worth considering.

Hello,

symmetrical strategies is a good start surely but it also depends on the market you intent to operate in. Some markets tend to be obviously “long biased” (look at Dow Jones historical chart) so when you try to develop a strategy for such market having Long rule = Short rule you will lose a lot of good “long only” strategies just because of poor performance on short side.

So my hint would be, look at a chart, get a feel about the historical rhythm of the market and use your own discretion here. Personally I see no edge in symmetry

Thanks for sharing Tomas. Points worth considering.

i will take it from other side – imagine that you build LONG only strategy for this market (dowjones) – what do you think will happen after 2018, when the behavior of market was totally changed?

During such periods a good robust long only strategy should make as few trades as possible