The Darvas Method

[S Chun-Li]

MM, u posted on another thread today about the oddball method and Zeev boxes, specific vol etc. I think this is praps a better thread to continue this train of thought due to less clutter.

The oddball system looks interesting and I haven't tested whether it works but the author claims it does!

It doesn't use volume or price as inputs but simply looks at the last hour of trading and compares the advancing issues on the NYSE in that hour to the same hour of the previous trading day.

If the advancing issues is up by 3% or more it is a buy. If it is down by 1% or more, it is a sell. It is always either long or short depending on the last signal.

If something so simple works then there would be a few ways I might tweak it to try to increase performance.

1) introduce the concept of stop losses
2) introduce a neutral state where it was 100% cash
3) allow the 3% and -1% triggers to move adaptively

I have a couple of ideas how I might implememnt tweaks 1 & 2, and some vague thoughts on tweak 3