One can create a custom VWAP slippage function to use in place of the standard slippage models. The slippage function get's as an input the order object and then outputs the number of shares filled and the price it is filled at. This function is called each minute for every open order.

So, how to create a custom slippage model (https://www.quantopian.com/docs/api-reference/algorithm-api-reference#zipline.finance.slippage.SlippageModel)? The attached algo shows one possible implementation. It uses historical minute volumes (21 days in this example) to calculate the percent of each securities total day volume which was historically traded during each minute. This is sort of a 'traded minute volume profile'. Orders are then 'filled' at the end of the day at a price equal to this average minute volume x actual minute close prices. This is the Volume Weighted Average Price (VWAP) for the day.

There are some caveats with this implementation:

• currently calculates VWAP for the entire day even if order placed later than market open
  
• doesn't check that enough trading volume actually exists each minute
  
• requires a fixed slippage be entered by user (could be 0)
  
• uses minute close_price * average_volume (for each minute) to determine VWAP (some VWAP calculations use a combo of low, high and close)
  
• half day trading is problematic since average_volumes are based on full day numbers
  
• there's a small bit of lookahead bias in determining the weights since the volume data includes the current day. This could be fixed by removing the top 390 minutes (ie the current day) from the data.

Use or adapt to fit your needs. Good luck.