Serving the Quantitative Finance Community

 
User avatar
HorseRider
Topic Author
Posts: 0
Joined: December 8th, 2004, 4:12 am

transaction cost and slippage assumption in back testing

February 20th, 2012, 8:23 pm

I read somewhere in this forum that transaction cost assumption for equity should be around 10 - 20 bps. I feel it is too high. Let's say one does a pairs trading, long a $50 stock and short another $50 stock for 100 shares each.15 bps would means 15 * 0.0001 * (50 + 50) * 100 = $15?that is impossible, since even Interactive Broker offer 0.01 per share.So what is a reasonable transaction cost assumption for high frequency strategy? and what about medium frequnecy (30 minutes holding period)?Thanks,Steve
 
User avatar
Stew
Posts: 0
Joined: January 13th, 2011, 12:53 pm

transaction cost and slippage assumption in back testing

February 21st, 2012, 2:07 am

In the US you usually have the choice of 'per share' or 'fixed percent' for equity brokerage. It may make sense to use 'per share' trade shares with high value. Most other equity exchanges use fixed percent brokerage. Most have a minimum brokerage. At a retail level brokerages of 4-10bp per side are common.If you pay for a algo interface to your broker, FIX etc, you'll probably find that they discount this according to the amount of brokerage you accrue during each month.Spreads and slippage at the DMA level vary hugely depending on the liquidity of the stock in question.
 
User avatar
HorseRider
Topic Author
Posts: 0
Joined: December 8th, 2004, 4:12 am

transaction cost and slippage assumption in back testing

February 21st, 2012, 4:17 pm

thank you, sir
 
User avatar
Marine
Posts: 0
Joined: July 17th, 2003, 7:56 am

transaction cost and slippage assumption in back testing

February 23rd, 2012, 12:24 am

Transaction Costs:If you are trading then you should know exactly what your transaction costs are. If not then use IB as an example. They are available on their website. Don't forget the stock lending fees.Slippage ... Market ImpactFrom my experience 20 bps is a good estimate. When you trade you should be able to calculate what your slippage is per pair and then feed this back into your backtest. Then you will have a better idea. If you are trading illiquid stocks or in huge sizes you will definitely have a market impact. So if you are only paper trading then 20 bps is a good estimate otherwise use your trading results. If you are trading then this should be one of the first things you implement. You need to have real-life and simulation inline and you should be able to determine your market impact from the results.$.02
 
User avatar
HorseRider
Topic Author
Posts: 0
Joined: December 8th, 2004, 4:12 am

transaction cost and slippage assumption in back testing

February 27th, 2012, 7:46 pm

thank you
 
User avatar
strategyard
Posts: 0
Joined: December 8th, 2011, 8:12 pm

transaction cost and slippage assumption in back testing

March 12th, 2012, 12:33 am

It is important not to overlook the transaction costs, especially for high-frequency trading. Anything between 15 - 50 bps is a reasonable estimate.