Dynamic system backtest with actual live current positions

[GolovinAlexx]

I found this comment from Rob.

Yes those are before buffering and the algo

After buffering (static system): system.accounts.get_buffered_position( instrument_code, roundpositions=True)
After greedy algo (dynamic system): system.optimisedPositions.get_optimised_position_df()

(Note in production I don't do this - I run the buffering and or greedy algo on the unrounded positions, taking into account my actual live current positions, rather than what the backtest thought it had on yesterday)

Originally posted by @robcarver17 in #529 (reply in thread)

How can I insert actual live positions as previous positions for a dynamic system backtest?

[robcarver17]

Why would you want to do this?

[GolovinAlexx]

I use only the 'sim' part of pysystemtrade. And don't use 'production' part at all. For example, because I don't have IB.

How I use 'sim' for a live trading.

In a static system, everything is simple: every day I run a backtest, and at the output I get top and bottom buffers for each instrument. I know the current values of my positions (from my broker). In a simple spreadsheet, I calculate the orders that I must execute (manually).

But the backtest of a dynamic system outputs the final positions. I can not use the buffering stage, when we must calculate tracking error of the optimised portfolio versus the positions we already hold. The portfolio changes too often.

What I would like. To understand where and how I can enter my current positions into the code (for example, as a simple dict) and run some piece of code that will calculate buffering.

[PurpleHazeIan]

I wanted to do something vaguely connected. Dynamic optimisation positions are path dependent. I wanted to see what positions I would have if I started from nothing today, and compare with my actual positions, as a sanity check of sorts. As a pointer, my solution used calculate_optimised_positions_data from sysexecution.strategies.dynamic_optimised_positions.

But you first need to understand buffering in DO (read the fine blog). It's not +/- on each position. It might or might not meet your need.

[robcarver17]

One option is to use the production part of the system in a limited way; generating orders but not executing them, and then manually entering fill prices (all in interactive order stack). The docs explain how to do this here.

Otherwise you can just call optimisedPositions.get_optimal_positions_with_fixed_contract_values of the