how much slippage to expect?

Discussion in 'Automated Trading' started by Kris, Jan 13, 2009.

  1. Kris


    Am I right in assuming that most automated systems are punching in and out of positions (removing liquidity)? For those of you who run automated intraday systems that trade a high frequency of trades, what kind of slippage should be factored in (assuming trading high volume equities > 5 mil/day with .01 cent spread)
  2. Baywolf


    My formula for slippage is (optimistic + likely + pessimistic case) / 3. This is based on historical bid/ask spread which I have collected on my own. The next thing I want to do is make an average spread based on the time. E.g., for x instrument, spread = y @ z time.

    I would not think too optimistic with slippage as low as .01. Try adding slippage somewhere from .03 to .07 if you are not accounting for the bid/ask spread, and see how your system does then.
  3. Kris


    Thank you, great info - I find your answers very helpful.
  4. Kris


    Slippage is a real reality check. I revisited an intraday system I've been working on, and at first glance this sucker looks like I'd be doing over 100% a year based on capital required...

    but because it's making so many trades, once you factor in slippage I'm looking at a break even situation :(
  5. This is my slippage table:
    Price Range Commission Limit Slippage Normal
    General .00913 0.051 0.074
    20 - 50 .00913 0.024 0.031
    50 - 80 .00913 0.052 0.072
    80 – up .00913 0.092 0.152
  6. Sorry about the table. The post reply submitted in the middle of an edit. I tacked slippage from my automation of daily stock trades over 8 months. I used stops (normal slippage) and limit (limit slippage) orders. Commision were one big bucket for me.

    To read the table, say".. price range 20 to 50 that a slippage of .031 and commision is used in the optimization.."

    I did this 2 1/2 years ago. It needs reviseing for todays volatility. But it still works for me.
  7. Let’s start over again (a phone call distracted me during writing, sorry) and I will make more sense of my post. I took out the limit column. My automation in Tradestation tries to look at the bid/ask and if the spread is too high and I set up a limit order that deviated from my original target price. But that is for a very small % of trades.

    This is general table I built 2 ½ years ago in the bull market. The automation traded about 162 trades in the daily time interval.

    Price Range Commission Stops

    General .00913 0.074
    20 - 50 .00913 0.031
    50 - 80 .00913 0.072
    80 – up .00913 0.152

    My general commission is .00913 and my general slippage is 0.074 per share not per trade. I use per share slippage because I roll back profits and losses in to the share calculation. So then shares traded increase or decrease causing per share calculations. Stops means I trade with buy stops and sell stops in my automation and that is how much per share I expect the deviation on average to be per trade under normal trading conditions from the calculate prices. For example if conditions are right and I trade a $60 stock the test optimization would use .00913 commission and 0.072 slippage. I found these numbers today are often inadequate when volatility changes.
  8. Kris


    Just to clarify - are we talking .07 cents round trip or .07 cents each way? I'm assuming round trip, ie .035 slippage to buy and .035 slippage to sell or vice versa... is this correct?
  9. Baywolf


    Kris, if slippage is a major concern for you, then just tell your algo to not enter a trade if the spread is beyond your threshold. Otherwise you will need to play the limit game, which is a whole different story.
  10. Kris – I swing trade an automated system in the daily and intraday (overnight) with stocks like AAPL, RIMM, etc…. Slippage is a big problem because of the pre-market spikes. In about one trade in nine the stops my system sets up get blown through causing large slippage amounts. This small group of trades throws off the total slippage calculation. And yes it is .074 per share and not per side.
    #10     Jan 16, 2009