[Python] Minimal Modification Approach to Port Python Factor Calculation to DolphinDB with...

I have a Python-based factor expression that I want to port to DolphinDB with minimal modifications while leveraging DolphinDB’s parallel computation engine. Could you advise on the most straightforward way to achieve this?

The factor calculates a cumulative bid/ask volume ratio with exponential decay. Here’s the Python implementation:

ask = df["av1"]
bid = df["bv1"]
p = df["mp"].iloc[0] # Scalar from the first value of 'mp'
for i in range(2, 11):
decay = np.exp(-10 * (i - 1) / p)
ask += decay * df[f"av{i}"]
bid += decay * df[f"bv{i}"]
vol_diff = 0.5 * np.log(bid / ask)


Assume df has columns:

• mp (scalar-like, same value for all rows)
• av1, av2, …, av10 (ask volumes)
• bv1, bv2, …, bv10 (bid volumes)

Expected Output:

A DolphinDB script that replicates the logic of the Python code while adhering to DolphinDB best practices (vectorization, parallelization if beneficial).

I’m aware of DolphinDB’s SQL-like syntax and vector programming features but need help with:

• How to replicate the loop over columns av2-av10/bv2-bv10 concisely.
• Proper syntax and functions for exponential decay and column arithmetic.
• Whether parallelization requires explicit configuration (e.g., ploop, partitioning).

Continue reading...