Back to Community
Transaction cost model

For large accounts, transaction cost seems less an issue than small accounts. As we know, transaction cost does have an impact on overall return. An algo which conforms to the Q's contest requirement has a lower return because of its lower risk profile. Thus, transaction cost can further reduce the returns to a major degree. Such an algo which is profitable with a large account may lost it edge or even becomes a loser after transaction cost if trading with a small account.

Currently, Q has per share and per trade commission schemes to model transaction cost. Here is an interest find that points out that the real world transaction cost is around 0.06% ~ 0.12%, much lower than the academic estimates of 0.6%. Is it possible to add a percent commission scheme in Q ?

Here is link: https://alphaarchitect.com/2017/11/28/factor-investing-and-trading-costs/

3 responses

You may create your own slippage and commission models using set_slippage() and set_commission() in initialize.

Is it possible to have a constraint for transaction cost such that while maximizing alpha, it penalizes high transaction cost?

@ Vladimir
The two models you mentioned are per trade or per share. Is there a way to specify the cost as a percentage of transaction?