Is it possible for GTR1 to simulate trading on a specific day of the week using an average of the next trading day's opening and closing prices? For example, trade every Monday (using Friday's data) at the average of Monday's open and close. If Monday is a market holiday, trade the following day. Or trade every Wednesday at the average of Thursday's open and close using Tuesday's data.
No. The returns are always calculated using a daily close total return series. No opening prices are stored. Jim
For example, trade every Monday (using Friday's data) at the average of Monday's open and close.I have a SIP backtester with daily OHLC historical data. My primitive backtester isn't as capable as the GTR1 but it does allow for Monday testing. Currently the default options are "c-close,o-open, a-open+(H+L)/2+close, n-no Monday calcs". The limitation is that from 1998 on although I have many delisted company's data I don't have some of the low volume company's data. Running screens with minimal liquidity requirements I typically have only around 70% of Monday prices available. I estimate Monday trade performance compared to Friday's by taking the Loss/Gain ratio that I find on the equities I have and applying that ratio to the ones whose price I don't have.For example using the High Relative Volume screen I get:HRVstd Screen hold 5 tilldrop 7 period 28 days, slippage 0.50%Friday trade CAGR 23.1% GSD 25.1% Ulcer Index 12.0% MaxDD 37.4% Monday trade CAGR 20.3% using Avg(O,Avg(HL),C) Monday/Friday ratio 0.9981 , Monday prices avail 68.3%If you expected to find that trading on Monday using Fridays data was going to invalidate the backtest performance gains you will be disappointed. Of 33 screens I updated yesterday the average Monday trade disadvantage is only a -1.45% CAGR penalty. The largest penalty is if you try something like RS100 with weekly trading where there is more like a -7.5% CAGR difference.I was going to purchase a complete set of survivorship bias free historical data including delisted data but after examining the data I had already been collecting since 2004 I decided it wasn't necessicary.RAM
Of 33 screens I updated yesterday the average Monday trade disadvantage is only a -1.45% CAGR penalty.Compared to what, trading at Friday's close? Are you comparing to Monday's average of high and low? I've found it makes a great deal of difference whether trades fill near the high or near the low.It's a moot point anyway as regards GTR1, as GTR1 can't sync to the days of the week.
Compared to what, trading at Friday's close? Are you comparing to Monday's average of high and low?The comparison is to Friday's close as that is what is in the SIP data and the basis of Keelix's backtester. I can compare to Monday to Monday purchases based on Open, Close, or any calculation of Monday's O,H,L,C values.
the basis of Keelix's backtesterYou're clear that we're talking about Robbie Geary's GTR1 backtester, right?
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |