Typical slippage faced by automated systems?

I

InvertedChance

Guest
Hi, I'm currently developing an automated system which works nicely in backtesting.

I'd like to know if anyone with experience has a ballpark of what kind of slippage due to missed fills that I'd be facing once the system runs live.

I'd be constantly shooting (and if necessary canceling) limit orders rather than using market orders.

I'm looking to run the system on NASDAQ and NYSE hybrid stocks, over roughly a 1-2 minute frame.

Much appreciated.
 
You need to give us more information on what kind of system you have developed. It depends on the stocks you're trading - how thick are they - how volatile... and more importantly, are you chasing ripping stocks? Are you betting on mean reversion? Are you trying to add liquidity or take it away?
 
Chasing ripping stocks, generally. My current system is designed for stocks that trade at least 1m shares/day, but have a good degree of volatility. It's not meant to trade stocks like MSFT, for example.

I developed the system backtesting against primarily AAPL. Otherwise I've run tests against GOOG, CME, and RIO; all three of these seem gentle in comparison to AAPL though.
 
Quote from InvertedChance:

Chasing ripping stocks, generally. My current system is designed for stocks that trade at least 1m shares/day, but have a good degree of volatility. It's not meant to trade stocks like MSFT, for example.

I developed the system backtesting against primarily AAPL. Otherwise I've run tests against GOOG, CME, and RIO; all three of these seem gentle in comparison to AAPL though.
More...

The biggest limitations you'll face are the execution speed of your platform, your internet connection speed, and the algorithms used to find your setups and send your orders in. CME has very little liquidity so you could potentially face lots of slippage on a cent basis, but not necessarily more on a percentage of the stock's movement basis.

Test with 100 shares. If you chase a ripping stock, you should be able to get it if you cross the market up a few cents. The hybrid market is pretty fast and efficient for getting stock when you need it. If you limit higher than the best bid to buy, you'll get price improvement as long as stock is available. You may need to use market orders to get out, and on the volatile, crazy stocks, that can be bad.

On NE today, I went market to sell and tried to hit a bid at xx.52; I was filled at xx.41, because someone one printed that bid. I don't generally experience slippage that's too bad, and I'm a scalper, for whatever my subjective experience is worth to you.

You really need to do some tests with 1000 shares and sacrifice the few hundred, or thousand dollars, to see how slippage will effect your performance.
 
Quote from NanoTick:

What is a ripping stock?
More...

A stock that's going up, fast, hard, with aggressive buyers competing to take out all available offers while the sellers back off, knowing they can dump their stock at higher prices.
 
Quote from InvertedChance:

Hi, I'm currently developing an automated system which works nicely in backtesting.

I'd like to know if anyone with experience has a ballpark of what kind of slippage due to missed fills that I'd be facing once the system runs live.

I'd be constantly shooting (and if necessary canceling) limit orders rather than using market orders.

I'm looking to run the system on NASDAQ and NYSE hybrid stocks, over roughly a 1-2 minute frame.

Much appreciated.
More...


I don't trade stocks but I think the number of missed fills is only related to your place in the book, and the frequency of sending new limits.

For example on futures like ES, your limit buy order would never get filled on the low of the bar if your order has been sent only 2 minutes prior. You get much more of those fills if you trade 30 + min timeframe.

I've once tested a system on ES that would generate 10 + trades in the day on limit orders. I never got a fill before price has gone below my limit( and the backtest was totally false )

To avoid this problem, retest your system with entry point 2 c below your normal entry point and with 2 c slippage. With this, your test will not take "buy at the low" and "short at the high" hypothetical trades into account( rude awakening sometimes ).

BYE.
 
You must log in or register to reply here.
Back
Top