|
PureBytes Links
Trading Reference Links
|
At 02:08 PM 8/29/97 -0400, TWA7663@xxxxxxx wrote:
>I have a question for veteran S&P traders. I have recently seen many
>comments about slippage. I am backtesting several S&P systems and need a
>number to program slippage. I would like that number to be a % of the price
>to enable a conservative "average" of what the veterans would expect over the
>next few years. The % would also enable the code to adjust for the price
>changes of the future instead of using a specific number. I realize this
>would be a guess because of the dynamics of the market. However, it is
>necessary for as "close as possible" backtesting results. Please indicate if
>your % is for both entry and exit or just the entry or exit.
>
>In addition to the slippage I would be interested in what the average round
>trip commission is for those of you using discount brokers.
>
>Thanks in advance for the help.
>
>Russ
>
>
You don't say whether you are trading short term or long term. Long term,
it might be reasonable to make slippage a function of price. But, in
reality, it is a function of the pace of the market at the exact time you
place a buy or sell order. In the S&P, it is not uncommon to be caught is a
sellers only market (while trying to sell) for 60 seconds, in which time
the market may move up 2-4 points (1 point=$500).
If you want to backtest long term systems that only trade on the close, you
can get pretty close to the close by doing market on close orders (assume 2
ticks slippage on average. Opening prices are pretty much a guess.
For short term systems, I look at the tick-by-tick prices, and assume it
takes 1 minute for my order to get to the floor after my system generates a
signal. Then (if I'm selling), I assume I will get the price of the first
downtick thereafter.
----------------------------------------------------------------
Lawrence E. Lewis
Vice President, Chief Technology Officer, Chronology Corporation
EMAIL:lel@xxxxxxxxxxxxxx TEL:425-869-4227 x122
|