ON THE DOWNSIDE
If RSI is effective in
trading swings, it must be ineffective in trends, right? That’s true; it’s not
great at trend-trading . That said, it’s not a bad indicator, as it does confirm
when you’re in a trend, valuable knowledge by itself. Some tradables swing well
even while trending, but a resolute advance or decline will find RSI pegged above or
below its signal boundaries of 30 and 70. (See sidebar, “Calculating RSI.”)
Figure 3 shows such
a situation in Intel [INTC]. Here, RSIВ maxed out in the
very first runup, then drifted around the
signal boundary
with declining highs even as INTC roared up another $45 per share. RSI’s role here was
that, in going to 82 in the first place, it portended extraordinary strength,
the probable onset of trending. Had you been using RSI as an oscillator here, only
your experience-based stop would have protected you from the inevitable loss of
going short. Of course, there were few places to go short, since there are
higher highs and higher lows throughout the advance.
SUMMARY
RSI is a great
indicator, one of only three I’ve used religiously for 15 years (the other two
being moving averages and onbalance volume), but its limitations must be
understood. Given that, you can reliably use RSI in tradables that swing well,
futures contracts being the best bets. Hesitate before invoking RSI on a trending
equity, but pay attention to its hitting extreme values: that’s usually a clue
to trend-sustaining strength.
John Sweeney is
Technical Editor for STOCKS & COMMODITIES.
В
CALCULATING RSI
Though the
formulation can be made complex, the basic idea of the RSI is straightforward.
Considering the last 14
days, let’s compare
the strength of the upward moves to the downward ones. That should tell us if
the price, on average, is stronger in the bull’s direction or the bear’s. If
there are more and bigger upmoves than downmoves, the ratio of the two should
increase. If the reverse, it should decrease.
In practice, looking
at the past 14 days, you average the day-to-day upward changes (from close to
close) and then
the day-to-day
downward changes. There don’t need to be the same number of each. Then you
divide the average of the upward changes by the average of the downward changes
to get what’s called the relative strength, or RS.
Since the relative
strength ratio could be all over the lot, Wilder arranged to keep it in the
corral by using the following formula
RS I = 100 – (100 / 1 +RS)
where RSI is the relative strength
index.
The next day,
smoothing the result, you multiply the average up close by 13, add today’s up
change (if any) and
divide by 14. For
the downs, you multiply yesterday’s average down close by 13, add today’s
downward change (if any), and divide by 14. Then compute relative strength and
RSI as before. Since
1978, many other algebraic formulations have made this simpler; more complex;
more flexible; or more convenient, but this formulation is the essential idea.
This way, the idea can be maintained by hand (as Wilder originally did) or
easily programmed in a worksheet or interpreted computer language.
Wilder is reported
to have selected the averaging period of 14 through personal experience. I’ve
found that the
period is indeed
experience-related: whatever period you pick, you’ll soon be experienced with
it and interpret its
fluctuations
appropriately. Theoretically, at least, the period should be the half cycle of
the tradable, so truly
outlandish numbers
should be avoided, but 14 isn’t a bad number for most futures and 21 isn’t bad
for most equities.
Stochastic & RSI
| 
