|
PureBytes Links
Trading Reference Links
|
Aongus O'Gorman wrote,
>As we are all system tester/traders I'm interested in how different
>individuals deal with slippage in backtesting. What type of slippage do
>people use (fixed, per market, per round trip?), and does it depend on
>the particular trading strategy?
>I'm testing a momentum based system looking at 30 markets in FX, Bonds,
>and Commodities. Round trips work out at c.3000 per annum per $mn. (1000
>for FX and Bonds, 2000 for Commodities). Does anyone have any
>suggestions on the level of slippage I should use.
Slippage is usually over and underestimated by traders and by book vendors.
I saw a publication by Lars Kestner where he tested "29 Popular Trading
Approaches". In these tests he used $100 p.r.turn for all markets. The
markets included everything from Corn to the S&P. Once you start trading
you realise that $100p.r.turn is way too high for Corn and way too low for
the S&P. Lars Kestner himself stated in his book that his slip assumptions
need to be adjusted for each market, however, that did not stop him from
producing a whole book of worthless statistics.
I've been trading for 12 months full-time and I have kept a record of my
slippage for ALL my trades. The best method to manage your slippage
assumption is to use a realistic hypothetical figure in your testing (e.g.
Corn = $40 and USTBonds=$90) and as you trade in real-time, analyse your
average real-time slippage results and actively change the slippage amount
in your hypothetical testing. The objective is to match your average
real-time slippage with your hypothetical slippage. One thing is for sure,
there are alot of traders out there that have discounted alot of good
systems purely because they "overestimated" their slip. It's OK to be
conservative but more importantly you must be a realist. When I analysed
my real-time slip over the last 12 months, I have found that the average
slip amount is much lower than I expected.
For example, what do people use for slip on the Swiss Franc? I originally
used $50 per round turn initially in my testing. After 12 months of
real-time trading, I have transacted 57 times in this market and my
real-time slip is only $17 per round turn. My best slip is $0 and my worst
is $100 (8 points). Please note I trade only 1 contract in the Swiss. With
this real-time information I would now use $22 p.r.turn ($17+$5) for my
Swiss system testing. I add an extra $5 p.r.turn to adjust for any
unfavourable movements in the future.
Another note worth mentioning is that the slip depends on your trading
methodology also. This is very important. In my trading I enter using a
stop order based around the open of tomorrow and I usually have an exit on
the open at market so that helps keep my slippage down. I generally have
high slip on the entry signal but hardly any slip on the exit. If you use
stop orders on both entry and exit then your slippage on that system may be
higher than on mine.
Once you have some real-time slip statistics, then you need to RETEST ALL
YOUR SYSTEMS AGAIN. You probably have some systems where you have
under/overestimated your slip. Be aware that there are people out there
that DO NOT analyse their slip results at all. They just say, "slip in the
S&P500 is $xxxx" without any mathematical foundation to their assumption.
I hope this helps.
I would like to ask any trader out there what their average slip is on the
S&P500 and Coffee ??? I do not trade these markets but I'd love to see who
has some real-time trade stats on their slip on these 2 markets.
Regards,
Robert Bianchi
Email:R.Bianchi@xxxxxxxxx
Email:RobertBianchi@xxxxxxxxx
,-._|\
/ Oz \
\_,--._/
v
|