How wrong predictions can produce right trades
A reader just emailed me with an excellent question. He pointed out
that, in my
recent article, I mentioned that the combination of falling interest rates
and rising oil prices led me to look for sell setups. Despite these
events, the market has risen since then. Asks the writer, “How do you
know when your market predictions are wrong?”
A major pitfall, I would suggest, is treating quantitative inquiries into the
market as “your market predictions”. I am happily married to my
wife of 21 years, and I respect the polygamy laws of the State of
Illinois. So I’m not going to get married to a market opinion or
prediction. A statistical study of the market reveals historical
tendencies: nothing more and nothing less. Properly constructed, it tells
you whether what happened in the past is a random occurrence or whether there is
a market behavior that deviates from chance. To the extent that you
uncover a pattern that occurs with a greater-than-chance frequency, you have a
potential trading edge. You don’t have certainty.
I consider myself a discretionary trader, in that I rely on pattern
recognition rather than mechanical signals for entries, exits, and position
management.  Adding a layer of statistical analysis to the trade
takes one source of market edge and correlates it with a second,Â
independent advantage. The French scientist, Louis Pasteur, captured the
strategy most elegantly:
Dans les champs de l’observation le hasard ne favorise que
les espirts prepares.
In matters of observation, chance favors only prepared minds.
The goal of statistical analysis is to prepare the mind for moves that have
tended to occur in the past. Such analyses function as a heads-up for
traders, much like a long-term weather forecast for a traveler. If the
forecast calls for cool weather, I will pack my sweater. That doesn’t mean
I’ll wear it if the clouds dissipate and temperatures reach the 90s.
Let’s take a recent example of integrating discretionary trading with
statistical analysis. After the trading session of August 9th, I
posted a blog showing that, since July, 2003, we had experienced only twelve
occasions in which there had been three consecutive days of comparable selling
in the NYSE TICK and my TICK measure for large capitalization stocks.Â
Expectations after such three-day selling periods were decidedly bullish, with
ten of the occasions showing gains a week later, and only two showing
losses. The average amount gained over these twelve occurrences was a
respectable 1.31%. As a result, I started the next trading session
looking for long side setups. It didn’t take long for those to
materialize. The emini S&P opened four points higher and, within eight
minutes of the open, a very strong plurality of 845 more NYSE stocks were
trading at their offer price, rather than their bid. At that point, I
decided to wait for one of my long entry patterns: the first pullback in the
NYSE TICK after an upside breakout. This occurred at 9:45 AM ET, as the
TICK retreated toward the neutral zero level with price holding nicely. I
entered at 1240 on the ES, riding the next surge in the TICK to over 1000.Â
This provided a 2-1/2 point profit in the ES in only 15 minutes.
The value of the statistical forecast was several-fold:
- It kept me out of a bad trade, by allowing me to reverse my thinking
from market weakness to potential market strength; - It allowed me to pursue a good trade, rather than sit on the sidelines
and lament that I missed the opening four-point pop; - It gave me confidence in the trade, knowing that historical odds were
on my side.
Notice that the trade was not mechanical. Although the
statistical odds spoke to a five-day bullish period, the strategy was not to buy
and hold for five days, since that did not fit my risk parameters.Â
Moreover, had the first pullback in the NYSE TICK resulted in a significant
retracement of the opening gap, I would not have pursued the trade, as this
would have violated my buy setup. It
was the combination of favorable statistical odds and a favorable trading
pattern that allowed this trading shrink to pull the trigger.
Now let’s take an example of a “wrong” prediction. I went
into Wednesday’s trade looking for sell setups, focusing on the five-day outlook
following a period of rising T-note prices and soaring oil. My entry was
to sell the first bounce from the market’s first downward swing, as long as that
price was below the previous day’s volume-weighted average price. (This
entry is a short-term trend following pattern). The market opened well
above the average price benchmark, however, and then proceeded to trade still
higher in the opening minutes, with over 600 more stocks trading at their offer
price than their bid. When the first selling bout materialized shortly
after 10:00 AM ET, it only took the ES back to the average price. The
market then proceeded higher on very strong buying, with over 1000 more stocks
trading offer vs. bid.Â
By that time I was saying to myself, “This shouldn’t be happening if
we’re going to get weakness”. Downtrending markets don’t trade above
their average price on a sustained basis. I realized that the market was
not living up to its historical tendencies, which–in itself-was valuable
information. Even more valuable were my historical studies of stocks
that trade bid vs.offer. Since July, 2003 (N = 542 days), when my
normalized version of the NYSE TICK, averages 200 or over (N = 205), that day
has been up 185 times and down only 20. When the adjusted TICK has
averaged -200 or less (N = 184), the day has finished green 28 times, red 156
times. In other words, when traders are hitting bids on individual
equities, the S&P 500 tends to move lower. When traders lift offers on
stocks, the S&P tends higher.
Early on Wednesday, even after the initial drop, the adjusted TICK was well
above +200. The odds of it being a weak day were drastically reduced as
long as the TICK values stayed so positive. As it happened, the TICK
actually strengthened during the day, making it highly unlikely that the
earlier forecast would win out. By updating the forecast with current
market readings, it was possible to:
- Avoid bad (short) trades;
- Search for good (long) trades;
- Appreciate that this market was sufficiently strong to overcome historical
precedent.
Perhaps this latter benefit is most important of all. A market that
moves against a known edge is telling you that something special is going on.Â
Some unique influence is impacting the market that was not there during the
majority of past historical occasions. Fading market forecasts and going
with such special influences can be very profitable. In that sense,
“wrong” forecasts can produce right trades.
Brett N. Steenbarger, Ph.D. is Associate Clinical
Professor of Psychiatry and Behavioral Sciences at SUNY Upstate Medical
University in Syracuse, NY and author of The
Psychology of Trading (Wiley, 2003).
As
Director of Trader Development for Kingstree Trading, LLC in Chicago, he has
mentored numerous professional traders and coordinated a training program for
traders. An active trader of the stock indexes, Brett utilizes
statistically-based pattern recognition for intraday trading. Brett does not
offer commercial services to traders, but maintains an archive of articles and a
trading blog at www.brettsteenbarger.com