Sunday, March 12, 2006

RULE #1

Never, ever, under any circumstance, should one add to a losing position ... not EVER! (see all the rules at The Big Picture)

A conundrum!
Each and every trading system I have tested has perfomed better, when I allowed adding to  the losing position compared to adding only when the trade was profitable. Actually quite a few systems turned negative the moment I let it add to a winning trade.

Of course it is correct that adding to a losing trade is the first step to disaster and I have lost a few arms and a leg already clinging to hope, while the right thing to do was just cover and reverse.

Let's look at an example and maybe some here can tell me, how they would trade it: Say the contract you trade made a nice move higher and is now in a 15 tick consolidation for the last 5 hours forming a nice and friendly bullish flag. So you "expect" the trend to continue higher and are looking for a long signal.

Prices tested the lower range, bounced, suddenly jump, you get a signal from your system and take the long as it happens just 2 ticks below resistance.
The Stop goes 3 ticks below support which means a 16 tick Stop on this trade. A bit wide, but imho there is no other logical place to put the stop.

Well this trade might work, prices might overcome resistance, having consolidated for 5h and decide to go up further. And assume that eventually prices will do exactly that...but not this time. Resistance holds and it's down again to support, for just another bounce. You, being not the dumb newbee, have placed your stop a few ticks below support and are not stopped out, even if support was broken by 2 ticks and you see yourself 15 ticks down. On 3 contracts, your regular trading size, that can be expensive, and just waiting for one hour before the breakout really happens will cost you a lot of mental energy.

So why don't you trade it differently:
Why no Scale In,
Why not enter the initial trade with 1 contract long, add another one at the middle of the range and add the last one 1 or 2 ticks above support. The first trade has a 16 tick stop, the second one in the middle of the range a 10 tick stop and the last one a 5 tick stop.
That means the Stop on the whole trade remains always 3 ticks below support, you make sure you get a bit of the cake, if the breakout happens right away, and if not, this technique actually saves you 17 ticks if stopped compared to the first example where you enter with the full position right away.

Just that this Scale-In is made into a losing position and it therefore violates Rule #1 or
does it really, if  the Stop remains the same.

As I said I already got my marks from violating Rule #1, so even if this Scale-In looks good, I'm always second guessing myself before I apply it. On the other hand, applying it makes for a very relaxed trading, so somehow it seems to be a technique my sub-contious mind likes. And one of the reasons is, that you no longer try to catch a top or bottom, try to pinpoint the exact entry, but work your position until being fully committed.

So if some traders reading this rather long post could tell me how they do it, I would really appreciate it.

2 comments:

Barry Ritholtz said...

There is a difference between scaling in, and "Doubling down"

If you are looking to establish a position for lets say a 6 month holding period, and you are willing to allow a draw down of 6%, then deciding to scale in before hand is fine -- let's say you scale into that trade at $64, 63.40, and 62.90 over 2 weeks -- thats ok.

Compare that with 1st buy at $64, adding at $52, and then again at $39 over 3 months.

In the first, you take advatage of short term volatility to obtain an advantageous entry.

In the second, you are ignoring what the market is telling you and merely throwing good money after bad.

Dave Johnson said...

I would have to agree with Barry. If the down impulse in which are accumulating is still moving down then taking positions before your predicted upward event would be wise. My backtesting on stocks has shown this to be a definite advantage.