Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
How we define performance?

I am a new bee here. after reading some post I have few questions. Question is when a quant call his algo perform better than SPY then we have no record about the transaction fee involved here.. If his or her algo is making $1 per transaction then does he/she subtracting the broker fee? Since most of the people are paying ~5 to 7$ to buy and sell a single stock. So overall per transaction they are losing 10 to 15$. Nobody has $1 million to buy SPY so they must be buying a max of 100 shares of SPY. So in their calculation do they count this number before coming to a conclusion that they over performed.

My understanding here is that a hedge fund manager has enough money to manipulate a stock. Normal trader deal with 100 share per transaction.

2 responses

Quantopian's backtesting engine has built in models to account for commission and slippage to make the simulations more realistic. The default commission used is $.03/share, but there's also 'per trade', 'per dollar', and minimum transaction costs models. It's always a good idea to look at the commission and slippage model used when assessing algorithm performance.

It's always a good idea to look at the commission and slippage model
used when assessing algorithm performance.

Yes, otherwise most the model claims that they outperform others. It would be nice to see how much realistic money was mode in real trade. Otherwise most algo will finish in some research paper.