PDA

View Full Version : Uncertainty With High Probabilities – July 29, 2011



Billy
07-29-2011, 06:14 AM
9558

The IWM robot entered a new long position at the open (79.96) yesterday. Make sure to adjust your stop at 78.28 to take the actual entry price into consideration. The robot’s indicative stop on the robot page is always in case of an entry at the exact limit price, but the actual stop must be adjusted for the actual execution price.

After some kind of hopeful dead cat bounce that stalled at the 2-day moving VWAP (blue line), IWM came back to the starting block and right above the same 200 dma (79.51) support confluence with WS2 (79.82). New weekly and monthly pivots will be in place on Monday and can change the floor support/resistance outlook significantly, so I won’t expand much on the current multi-pivot setup. Under normal market conditions, the 200-day moving average shouldn’t give up easily after such oversold conditions where IWM lost 5% in just four days. But market conditions are not normal and remain uncertain in the context of the debt crisis political irresponsibility.

As I took my breakfast this morning watching CNBC Europe, I was struck that the dozen of pundits interviewed and debating were UNANIMOUSLY in agreement about the US equity market: avoid and stay on the sidelines! This is the first day I saw such unanimity and it makes me wonder who’s left to sell? As Pascal mentioned in his daily IWM comment, the TEV extension on inversed ETFs is suggesting that large players who didn’t sell already are fully hedged on their remaining long positions. And that TEV extension is the last reason for the robot to refuse to sell and short this market.

In spite of high uncertainty, probabilities of gains and success for a new long entry rose to a very strong ST/LT buy signal setting today. If you are a proponent of high probability opportunistic trading, this can be your day! You won’t find often a statistical 3-day gain of 2.71% with 81.3% odds for a winning trade. Pre-market is weak at time of writing, so a low entry on a gap down makes the prospects for gain even better. As a side note, there will be a POMO (QE Light) today, so liquidity will be plentiful on top of all the cash parked on the sidelines.

9560

GDX was stopped out of its long position yesterday for a gain of 7.38% on this last trade. The robot could not find any edge for entering a new long or short position. The second chart below and comments are from the ETF Digest website.
Billy

9559
9557

roberto.giusto
07-29-2011, 08:35 AM
Billy,

I may be wrong, but if I remember well it often appears that when the robot gives a very strong signal it is in the middle of a rather NOT comfortable situation, at least on an emotional level.

Pascal is surely right when he says that we are at an infinitesimal distance from confirming the sell signal on the 20DMF, and very wisely suggests to keep a reduced position due to headline risk.

Still, the robot gives a VERY STRONG buy signal... my question is: is there a way to know if past signals were somehow affected by headline risk?

Or put another way: shouldn't the robot algorithm take into account such occurrences for the mere fact that "the price says it all"?

Excuse me if my question sounds as a dumb one, but I remember having the same doubt on one of the similar past occurrences.

Timothy Clontz
07-29-2011, 09:04 AM
I've also noticed that the best (and worst) times to invest are at the times of greatest fear.

Regardless, I think the Robots are operating correctly even with the current headline risk. If we get stopped out again, we get stopped out again.

As for lightening up -- those following both Robots are already lightened up because of GDX being on the sidelines.

Can't think of anything else that could be done here.

Billy
07-29-2011, 09:18 AM
I've also noticed that the best (and worst) times to invest are at the times of greatest fear.

Regardless, I think the Robots are operating correctly even with the current headline risk. If we get stopped out again, we get stopped out again.

As for lightening up -- those following both Robots are already lightened up because of GDX being on the sidelines.

Can't think of anything else that could be done here.

Roberto, I agree with Tim.

The robot is not a news or event predictor, but a follower of probabilities. Probabilities are never 100% and you can accumulate losses in the long run. But the robot is also programmed to limit losses to normal drawdowns.

The alternative is to trade with your guts feeling or not trade at all.
Billy

roberto.giusto
07-29-2011, 09:34 AM
The robot is not a news or event predictor, but a follower of probabilities.

I didn't mean I was worried about a potential loss, but I guess this is the answer to my question.

Thanks to you both,


Roberto

asomani
07-29-2011, 10:05 AM
In my opinion, a major point to mechanical trading is trading with the probabilities particularly when emotions around you are running high (and it is not atypical for this to be the case when headline risk is elevated) and when you are also likely to be at least somewhat emotionally driven (as is currently the case). Incidentally, this is also often the time that mechanical trading provides one the greatest benefit and the largest gains (in my experience, anyway).

Timothy Clontz
07-29-2011, 10:34 AM
I don't know all of the ins and outs of the model, but I think the Robot adjusts stops based on volatility. That is, if the price rises while volatility rises, the stop might rise less than the price. If the price falls while volatility falls more, the stop might actually rise.

But the Robot (if I understand correctly) NEVER lowers the stop.

Under normal situations that's entirely correct, I'm sure.

Of course, that begs the question of defining "normal." It's like what the judge said about porn "I can't define it, but I know it when I see it."

Billy -- I'm sure you've tested allowing stops to lower as volatility rises. Maybe if you shared the edge the current non-lowering approach has, that might address (on a statistical level) what would happen if you tried to adjust for "headline risk" (which would be reflected in such an approach).

That might answer everyone's questions with specific numbers.

On an anecdotal level, I'd like to compare this with poker. Do you increase the bet when you get worse cards? Or do you fold quickly for a small loss and wait for a better hand? Winning players take the small loss and are grateful for the small loss.

brrim
07-29-2011, 10:47 AM
Billy "the actual stop must be adjusted for the actual execution price".
Do you mean by this that for a new IWM buy today the stop at 78.52 is 2.07% lower than the entry price of 80.18. If you bought at open today at 78.72 maintaining the relationship the adjusted stop would be 77.09. Similarly for TNA the numbers are buy 74.03, stop 69.45-a 6.18% difference, so a buy at open of 70.04 would give you a stop of 65.71.
Is this correct?
Thank you.
Robert

Billy
07-29-2011, 10:52 AM
Billy "the actual stop must be adjusted for the actual execution price".
Do you mean by this that for a new IWM buy today the stop at 78.52 is 2.07% lower than the entry price of 80.18. If you bought at open today at 78.72 maintaining the relationship the adjusted stop would be 77.09. Similarly for TNA the numbers are buy 74.03, stop 69.45-a 6.18% difference, so a buy at open of 70.04 would give you a stop of 65.71.
Is this correct?
Thank you.
Robert

Robert,

Yes this correct.
Billy

Billy
07-29-2011, 11:24 AM
I
On an anecdotal level, I'd like to compare this with poker. Do you increase the bet when you get worse cards? Or do you fold quickly for a small loss and wait for a better hand? Winning players take the small loss and are grateful for the small loss.

Tim, you couldn’t provide a better analogy. The robot always fold quickly for a small loss and waits for a better hand.
Risk management is all about the position’s risk, not external headline and event risks.
The stop management will start with a tight stop at the beginning of a trade.
Then the initial stop will be fixed (not trailing) until a certain backtested profit threshold has accumulated in the trade on a daily close basis. From there, the stop will be trailed with a larger allowance for volatility than for the initial stop. That’s why a trailing stop is never lowered: it already takes into account a wide enough volatility tolerance after it becomes profitable and can allow for more wiggle room.
The robot excels in trend-following, but we haven’t had a sustained intermediate trend to follow for 7 months now. When trend-following will be back –and it will!- the volatility will contract on an ATR basis and the trailing stop will consequently become tighter and tighter to secure maximum accumulated profits.
In choppy ranging markets like now, the stop will often remain very tight because the profit threshold for trailing wider stops will seldom be reached.
The objective is not to get stopped out easily from trending positions and to take only minor losses in choppy markets.
Billy