When you're manually entering your stop orders or using TS "send stop orders to the TS network" option( not working actually ), your stop orders are already in the book and are filled way before the market order that is sent when the level is reached. When I make the transition from manually entering the stop orders to auto through TS( without the option I was refferring ) , my slippage came from almost 0 to 0.5 tick per side. I think stop orders are the best to use. Limit orders are the cause of many incredible equity publication but are the worst. You really have to use one tick per side of slippage to test limit orders. Brokers like TS are totally unable to fill sell orders on high of bars and buy orders on lows( if someone knows one, PM please ). To test for example a countertrend channel system, I use "Buy at LowerBand - 1 point limit " with one tick slippage to test "Buy at LowerBand limit" . It simply eliminates the signals that would not be filled and reduces considerably the interest of limit orders.I don't say you have got slippage on limit orders, I just say they are those in which you must count the max slippage in backtests. I don't know the last orders you're talking about , but that would be really interesting to use since they just have half a tick slippage and can be used the same as limit. I don't think we can use it on TS. Are you using it? Do you know a platform that can automate those orders ? Thanks.
Indeed, a poll of TS users indicates widespread need for an MIT order command: https://www.tradestation.com/Discussions/Topic.aspx?Topic_ID=55435 However, with TS8.x, and using the IntrabarPersist directive, I believe it is fairly straightforward to code your own MIT order. With the TS2000i version, you must wait till the next bar before placing the market order if the limit price was met in the current bar and there was no fill.
There is no such need for extra coding. When you write your strategy orders as limit orders, they can be implemented real-time as MIT orders by applying a setting to the strategy. Under the "Format Strategies" window for your particular strategy applied to a TS chart, click on "Properties". Under the "Automation" tab there is a pane titled "Strategy fill logic". You need to click the radio button for "Strategy will fill non-historical orders based on price activity", and check all three checkboxes below. This ensures that if your limit orders do not receive a first-in-line fill, they are converted to market orders after a specified delay. I use 0 seconds. If you are at the top of the queue, your order will be filled before TS has a chance to convert your order to market. If you are lagging in the queue, then you will just buy at ask or sell at bid. RoughTrader
walter, I always expect at BEST to get filled at the bid on sells and at the ask on buys...... That doesn't always happen.. but not ALWAYS for the bad... as I have discussed earlier in this thread, due to the nature of my when I am entering into a position I actually get what I consider "positive" slippage.. meaning I estimate at best to get filled at the bid on sales.. but sometimes I actually get filled a tick or so better... some times I get filled a tick or so worse..... just looking at the past 26 days it has traded live.... If I run a backtest and I use the bid/ask as my entries instead of my order simulation that I generally use for backtests, I find that the total profit matches nearly identical to what I experienced in real money... taking one day at a time that isn't true, but If I look at a few days added together it seems to all average out... So with that said at least for the past 26 days bid/asks is the extent of my slippage it seems.... that is after about 100 trades... going forward I do not know if that will be the case.. for what its worth my backtests using my order matching algorithm actually comes in with a slightly lower profit... which makes me trust my backtests even more going forward knowing that it may be undershooting the profitability a little...
Really helpful . I had never thought of it. Thank you very much, RoughTrader... Last question if you dare answer: What is the proportion of your orders that are filled as limit orders/ market orders? I suspect the number of filled limit orders would be very low for fast timeframes.
Good to know....I'm not using TS8 at the moment. I wonder exactly when this feature was implemented ? Seems that since they don't even mention MIT, then many users may be unaware of it's existence.
I only automate intraday systems, my EOD systems I just load the orders as MIT for the next session manually. On the hourly time frame, about 60 to 70% of my exit orders get filled as limit orders because they've been sitting in the queue at least an hour. Entries I would say maybe 30 to 40 % because the fill is issued soon after the order is made. On a 5 minute time frame, probably greater than 80% are converted to market. The times the limit orders get punched through cleanly are only when there is volume behind the move. RoughTrader