|
PureBytes Links
Trading Reference Links
|
I did a bunch of testing on OB using stop entries a couple of months ago.
Testing on SPX, the results were very encouraging. Smoother equity curve,
fewer trades, etc, etc. I knew that trading ES or SP would have different
results, so I put up a quick chart and examined the entries. Trades
occurred on the same bar for SPX and ES about 80% of the time. I also
changed the code to NOT enter a trade on a stop during the first bar of the
day. This prevents trading the SPX at an unrealistic price, as is being
discussed in another thread right now. I also cut the bars down to 30m
instead of 60m, to allow the stop entry to trigger a little sooner.
So how did the stop entry compare between SPX and ES? Terrible! Ok, not
terrible, but ES totally failed to show the dramatic improvement that SPX
showed using the stop entry. Here are a couple of quick system results over
600 or so trades:
SPX normal oddball: 1556 points profit, 2.52 points avg trade
ES normal oddball: $84678 profit, $146 avg trade
ES actually performed a little better, but ES and SPX were comparable.
SPX oddball stop entry: 1980 points profit, 4.63 avg trade, 200 fewer trades
than normal OB
ES oddball stop entry: $66800 profit, $151 avg trade, 130 fewer trades
ES stop entry performed poorer than regular OB, and had larger DD (not
shown). What gives? I thought most of the entries were the same?? So then
I started looking harder at these "same" entries. I knew that a system on
ES with stop entries would have some trades entered differently, due to the
spikier/noiser nature of ES. But the ES system was making less money even
on trades where the entries were on the exact same bars!
After staring at them for awhile, I realized what was happening. If you
enter on a buy stop above the top of the previous bar, you are entering when
the market is moving your direction. That's the whole point...to make the
market prove that long is the place to be. The problem is PREM! When the
market is moving up, premium is expanding, so you pay more for the privledge
of buying during a move up. Then when that trade is over and you are
reversing short with a stop at the bottom of the previous bar, what happens?
PREM has shrunk and you sell at a lower price.
Long story short, every trade you take using a stop entry on ES guarantees
that you pay a penalty equal to the current PREM. Right now, PREM is small
so its under a point. Early in the contract, or when rates were higher, the
PREM tax is higher. The extra 1 point per trade penalty takes alot of wind
out of the stop entry method on 30m and higher charts, particularly when
compared to the same system on SPX.
-Jeff
----- Original Message -----
From: "Bob Fulks" <bfulks@xxxxxxxxxxxx>
To: "BobR" <bobrabcd@xxxxxxxxxxxxx>
Cc: <omega-list@xxxxxxxxxx>
Sent: Friday, March 22, 2002 8:41 AM
Subject: Re: Oddball modification
> At 3:12 AM -0800 3/22/02, BobR wrote:
>
> >Try this backtestors. Oddball can be early with its signals so once the
> >signal is true then buy on a stop to delay the signal until price is
moving
> >in the direction of the signal. Remember, natural hour bars, sp or spx
in
> >data1, advancing issues in data2. Perhaps someone with lots of good
clean
> >data will post a comparison using the trade on a stop versus just buying
or
> >selling when the ROC conditions are satisfied.
> >
> >Inputs: BZ(3),SZ(1);
> >
> >If RateOfChange(close of data2,7)>BZ then buy at H[1] stop;
> >If RateOfChange(close of data2,7)<SZ then sell at L[1] stop;
> >
> >bobr
>
>
> A quick test of:
>
> Inputs: BZ(3),SZ(1), Len(1), Offset(0);
>
> If RateOfChange(close of data2,7)>BZ then buy at H[Len] + Offset stop;
> If RateOfChange(close of data2,7)<SZ then sell at L[Len] - Offset stop;
>
> 4 years ending 12/31/01
> $SPX cash index as data1 (BigPointValue = 1) trading 1 share
> $ADV NYSE advancing issues as data2
> Zero costs
>
> MB Inputs Inputs
> Original 3,1,1,0 3,1,1,4
>
> Net profit 1841 2470 2980 Points
> Trades 1080 754 536
> % Prof 48% 50% 54%
> Ave Trade 1.74 3.28 5.65 Points
> PF 1.38 1.66 2.06
> DD 260 180 130 Points
> ROA 706% 1370% 2280%
> Sharpe 1.58 2.34 3.20
>
> So your suggestions tests a lot better. Offsetting the stop 4 points
> tests even better still.
>
> Of course, trading the $SPX index on a stop is not really
> representative of the real world.
>
> A better test would require testing on the futures contracts or some
> instrument such as SPY and results would be poorer because of the
> increased noise. And entering on stops typically adds slippage.
>
> But it does indicate that holding entries until the price is at least
> heading in the right direction makes sense. (This should be pretty
> obvious. It always seemed pretty dumb to enter a long position when
> the price is clearly moving down...)
>
> I have tried similar approaches using the slope of a moving average
> with similar results.
>
> Bob Fulks
>
|