Using Bid&Ask Prices vs Only Mid-Price in Backtest

Slippage should be minimal for market orders in a liquid instrument that is not violently volatile. Stop orders definitely cause the need to account for slippage.

 

Aren't stop orders (other than stop-limit) essentially executed as market orders?

 

DEFINITION of 'Stop Order'
An order to buy or sell a security when its price surpasses a particular point, thus ensuring a greater probability of achieving a predetermined entry or exit price, limiting the investor's loss or locking in his or her profit. Once the price surpasses the predefined entry/exit point, the stop order becomes a market order.

Investors commonly use a stop order before leaving for holidays or entering a situation where they are unable to monitor their portfolio for an extended period.

Stops are not a 100% guarantee of getting the desired entry/exit points. For instance, if a stock gaps down, the trader's stop order will be triggered (or filled) at a price significantly lower than expected.

 

So yes, stop order is a market order. (again, other than stop-limit orders)

 

Any backtesting strategyb should simulate reality as closely as possible in order to avoid nasty surprises when you go live with your strategy. Using ask prices for long entries and bid prices for short entries would help improve simulation accuracy. Of course, as mentioned above the bigger the spread and the lower liquidity - the more important it becomes to use bid&ask.

The answer will also depend on what frequency your live strategy will be generating signals based in production - every second, every tick, X-min or X-hour bar close prices only? If you backtest on hourly bars using close prices, but in production your signals are generated based on 1-second price checks - your live trading signals maybe be very different from backtests.

 

Any backtesting strategyb should simulate reality as closely as possible in order to avoid nasty surprises when you go live with your strategy. Using ask prices for long entries and bid prices for short entries would help improve simulation accuracy. Of course, as mentioned above the bigger the spread and the lower liquidity - the more important it becomes to use bid&ask.

The answer will also depend on what frequency your live strategy will be generating signals based in production - every second, every tick, X-min or X-hour bar close prices only? If you backtest on hourly bars using close prices, but in production your signals are generated based on 1-second price checks - your live trading signals maybe be very different from backtests.

 

I hope you're aware stop market orders can lead to really awful fills. When P&G executes at $.01, I would guess that the seller in such cases used a stop without a limit, then when a mini-flash-crash hit (or a not-so-mini flash crash hit), his order got hit. You don't want to be "that guy" selling a $70.00 stock for $.01. Stop market orders make that very possible.