search

GuntharDeNiro / GunbotAMA

This repo is used to accept user's features requests for Gunbot AMA development
4 stars 1 forks source link

Position sizing/compounding #14

Open NCPHD opened 4 years ago

NCPHD commented 4 years ago

Gunbot would be more powerful with a choice of several automated sizing/money management options that could eliminate the need to manually size up or down frequently.

GuntharDeNiro commented 4 years ago

please be more specific about what kind of risk management took you would like.

GuntharDeNiro commented 4 years ago

please a feedback is needed on this issue, thank you.

NCPHD commented 4 years ago

Thank you for not letting this drop. I had been researching existing algorithmic strategies to try to find something as a model, but nothing quite seemed to be yet practical to suggest for Gunbot. So, because of the lack of an existing example, Here is what I'm thinking: a set of options that allow the bot to resize orders based on available cap increase (2%, 3%, 4%...) essentially compounding for us and eliminating the need to re-edit our limits every time we gained or lost) enough. Partial take profits/stop could be optional over full close to increase/decrease risk. R/R tolerance ranges could be set so that needs risky trades re limited. his could (or will) result in insufficient order sizes, but a trader who is into position sizing can handle that.

Traders can already DCA and pyramid with TV, and I know MB_CHECK as a permanent feature to allow scaled orders had been discussed a while ago. With sell targets or partial takes plus a hedging or layered ordering ability, we'd have the strong foundations for a complete system.

Ps, I spent a while looking at Var. It's not something I'm suggesting, but it did make me think that there have to be ways I could get GB to use its own performance in factoring a trade decision. Anyway, I know its a tall order, but thank you for opening this door and considering all requests.

GuntharDeNiro commented 4 years ago

ok i'll start with different sizing then we can proceed testing and eventually correct/implement. Thanks for asking.