Stocks & Commodities V.

11:1 (11-18): Stochastic Momentum by William Blau

Stochastic Momentum
by William Blau

A new twist on the venerable stochastic formula was presented in a January 1991 article,
"Double-smoothed stochastics," written by William Blau. Here, Blau expounds on stochastic double
smoothing in a somewhat different form that emphasizes momentum characteristics.

ouble smoothing of both numerator and denominator of the original formula for %K of the stochastic

indicator aids in obtaining low-lag smooth-contoured indicator curves. In lieu of a single parameter to
specify the stochastic, the Ds-stochastic formulation provides an additional two parameters for a
double-EMA (exponential moving average) effect. The Ds-stochastic formula is given by:
DS(q , r , s) = 100

E s ( E r (Close LL: q ))
E s ( E r ( HH: q LL: q ))

where, for the numerator,

HH:q = highest high in a lookback of q-days
LL:q = lowest low in a lookback of q-days
Close-LL:q = numerator of Lane's q-day fast stochastic
Er(Close-LL:q) = r-day EMA of (Close - LL:q)
Es(Er(Close-LL:q)) =s-day EMA of Er (Close-LL:q)= double-smoothing
and, similarly, for the denominator.
Figure 1 depicts a train of high, low and close price bars. The dark bar on the right describes the range,

Article Text

Copyright (c) Technical Analysis Inc.

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

HH:q- LL:q, assumed by the price bars in a lookback of q-bars. The raw stochastic formula,
Ds(q,1,1),represents where the current close is relative to the low point of the stochastic range. A close
near the upper portion of the range will be near the highest high in the most recent q-days, for example.
Figure 2 shows the same train of high, low and close price bars but with different annotations. The
stochastic range remains unchanged. The close is now referenced to the midpoint, 0.5 (HH:q + LL:q), of
the range.
The distance of the current close from the midpoint defines stochastic momentum as
SM(q) = Close - 0.5 (HH:q + LL:q)
As a momentum, SM(q) takes on positive and negative values; the magnitude of the momentum is
determined by how large a displacement there is of the close relative to the midpoint of the range. When
the close is greater than the midpoint of the range, the stochastic momentum is positive. The largest plus
value occurs when the close equals HH:q. SM(q) takes on negative values when the close is less than the
midpoint having its greatest negative value when it is equal to the lowest low in the q-bar lookback. A
comparison is made in Figure 3 between close momentum and stochastic momentum. Close momentum
is simply the close, 20 days ago, subtracted from today's close. With stochastic momentum, a function of
the close 20 days ago is subtracted from today's close. Observing Figure 3, the similarity between the two
is evident.
STOCHASTIC MOMENTUM INDEX
Embracing stochastic momentum, the stochastic momentum index (SMI) is now formulated as:
SMI(q , r , s) = 100

E s ( E r (SM(q ))
0.5( E s ( E r ( HH: q LL: q )

where:
SM(q) = Close - 0.5 (HH:q + LL:q) = stochastic momentum
and
ASM(q) = Es(Er(SM(q)))
= average stochastic momentum (ASM)
= double-smoothing of stochastic momentum
The SMI is the bipolar form of the Ds-stochastic. Both have the same shape.
The basic configuration of the stochastic momentum index is shown in Figure 4 for aq = 13-day lookback
with EMA smoothings of 25 and two days, respectively. The indicator maps prices into a corresponding
range of from -100 to + 100 on its scale. Prices are considered to be at "high" levels when the indicator is
above its threshold overbought line (here set at +40). Prices are said to be at "low" levels when the
indicator is below its threshold oversold line (here set at -40). The signal line shown is the E MA of
SMI(q,r,s). It is normally in the range of three to 12 bars. When the SMI is above its signal line, a price
uptrend is indicated; a downtrend is defined when the SMI is below its signal line.
Double-smoothing in the stochastic momentum index, SMI(q,r,s), is obtained by using any two of the

Article Text

Copyright (c) Technical Analysis Inc.

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

q,r,s parameters while holding the remaining parameter fixed. Smoothing is readily evident in the r,s
parameters because the represent direct exponential moving averages. Smoothing is also intrinsic to the
q-parameter describing the stochastic momentum, SM(q). See Figure 5 for the smoothing effect of the
cumulative sum of the stochastic momentum. The result is comparable to taking an EMA directly on the
close.

When the stochastic momentum is taken over a very large

interval, the stochastic momentum takes on characteristics of
price shape.
STOCHASTIC MOMENTUM AS A PROXY FOR PRICE
When the stochastic momentum is taken over a very large interval, SM(q) for q, a very large number, the
stochastic momentum takes on the characteristics of price shape. Figure 6 shows a q=300-day interval of
stochastic momentum, SM(300). If the number scales on the right of the charts had been omitted, a user
would be hard pressed to determine which was the close curve and which was the stochastic momentum.
Figure 7 shows the large interval stochastic momentum (the first "smoothing", as in figure 6) with a
second smoothing of five days. The result is a stripping away of the high frequency noise fluctuations
producing a smoother curve. Because a short-duration EMA is employed, introduction of lag (delay due to
moving average effects) is minimal.
Figure 7 shows double smoothing using the parameter pair: q, r.The only lag introduced was due to the
r-day EMA. The q-day span for stochastic momentum did not (ideally) produce; any lag, although q = 300
days is a very large number of days. This is a very important characteristic of double-smoothed
momentum indicators: Generally, one of the smoothing functions will be of long duration; the other will
be of shorter or equal duration. Only one will (ideally) introduce lag. This is certainly not the case when
moving averages are taken on price: a 300-day span for a moving average produces a great deal of lag.
This feature of double-smoothing of momentum indicators is fundamental to their time liness and
smoothness as indicators of stocks and commodities. Currently, two of the most popular indicators, the
slow stochastic and MACD (moving average convergence/divergence), are double smoothed indicators.
Two recently introduced indicators, T SI (true strength index) and Ds-stochastic (double-smoothed
stochastic), also employ double smoothing.
Normally, the very long time spans of Figure 7 are rarely used. Figure 8 shows an arrangement with a
stochastic lookback of q = 20 days with r = five days. (It is implied that there is only double moothing
and the additional s-parameter is fixed at s=1.) The stochastic momentum after double smoothing,
dubbed the ASM(q,r,s), is seen as being timely and relatively free of noise variations. The S MI(q,r,s),
stochastic momentum index, is similar in appearance. Divergences appear when smaller time spans are
employed. A down divergence is identified via points A and B or A and C, indicating a possible end to
the price rally. Indeed, it ended with the plateau in late September to early October.
The slow stochastic and the stochastic momentum index (S MI) are compared in Figure 9. They
approximate each other well in terms of shape. The small disparity between the two is due to the use of
exponential moving average in the SMI versus a simple moving average in the slow stochastic. A
departure from the slow stochastic is made in Figure 10 by increasing the EMA smoothing interval from
five days to 20 days, producing SMI(q,r,s) = S MI(2O,20,1). The stochastic momentum index now appears

Article Text

Copyright (c) Technical Analysis Inc.

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

to trend as prices trend. (The slow stochastic is retained for comparison.) The major turning points are
highlighted in a timely manner with lag less than, or equal to, that produced by the same moving average
applied directly to price. Divergences now appear at B, C, D, E and F. A divergence from the beginning
of B to C signals the end of a rally in prices, or the beginning of a decline. At A, the slope of the SMI
trends up with prices. At A, the slow stochastic remains noisily flat in an over-bought region. As shown,
the stochastic momentum index is slowly varying, showing price trends, major turning points and
labeling of turning points via divergences with price.
The slow stochastic may now be used as an entry (or exit) vehicle for trading with the stochastic
momentum index defining the trend of prices. An example can be seen in Figure 11 for an SMI(q,r,s) =
SMI(20,60,1) defining the trend and a 20-day slow stochastic. Note the smoothness of the S MI due to the
increased smoothing to an EMA of 60 days with essentially very little lag introduction of the major
turning points.

ong trends are often troublesome to those indicators that are inherently oscillators. Broadly speaking,

when the time duration of the trend exceeds the indicator's natural period of oscillation, the indicator may
not properly show the trend direction. Figure 12 depicts such a situation. The lag is determined by the
smallest value of q, r in the stochastic momentum index, SMI(q,r,s). The turning points to be retained, the
"major" turning points, are filtered by the largest value of the double smoothing (q,r) parameters. In
Figure 12, the anomaly of having a downtrend in prices erroneously indicated is corrected by increasing
the r parameter from 20 to 150.
TWO-DAY STOCHASTICS
When q=2, there are two price bars in the lookback period as shown in Figure 13. Both the Ds-stochastic
and the stochastic momentum index are described. The latter is more appropriate for our present
purposes, as it permits the use of stochastic momentum, which here becomes:
SM(2) = Close - 0.5 (HH:2 + LL:2)
To determine if the two-day stochastic is a reasonable proxy for price, a large (300-day) moving average
is employed in the stochastic momentum index, SMI(q,r,s) = S(2,300,1). Comparison with the price chart
on Figure 14 reveals it to be an excellent stand-in for price. An example of two-day stochastics is shown
in Figure 15 for Treasury bonds where S MI(q,r,s) = S MI(2,20,20). The fixed parameter is q=2; the two
variable parameters for double smoothing are the r-day EMA, which is subsequently smoothed by an
s-day EMA. The effect of the double smoothing is to remove high-frequency fluctuation, obtaining a
relatively smooth trending curve (see sidebar, "Stochastics momentum index"). The curve is observed to
be timely, low lag, with access to major turning points in prices. The overbought and oversold thresholds
here are set at plus and minus 20, respectively. Generally, a buy is indicated for the S MI crossing above its
signal line; a sell is indicated for the SMI crossing from above to below its signal line.
An interesting situation is depicted in Figure 16: the effect of a key reversal pattern. A key reversal
occurring in mid-September produces a two-day stochastic momentum index differently than does a true
strength index (TSI), both using the same smoothing parameters (see sidebar, "True strength index"). The
TSI(close,40,20) tracks the close-to-close price curve well remaining unaffected by the key reversal.
However, the long spike of the key reversal has a dramatic response with the two-day S MI(2,40,20). This
is to be expected, because two-day stochastics are very sensitive to the location of the close relative to the

Article Text

Copyright (c) Technical Analysis Inc.

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

extreme prices today and yesterday. Note the sudden drop in the response of the S MI(2,40,20).
ONE-DAY STOCHASTICS
When q = 1, the nomenclature suitable for Ds-stochastics and the stochastic momentum index is shown
in Figure 17. With a lookback of q = 1 bar, the stochastic momentum is defined within the highest high
and lowest low of the day:
SM(1) = Close - 0.5 (high + low) = one-day stochastic momentum
The stochastic momentum index becomes:
SMI(1, r , s) = 100

E s ( E r (SM(1))
0.5( E s ( E r ( high low )))

The one-day stochastic is sensitive to the location of the close relative to the high and low of the day.
This characteristic is useful as a sentiment, or trend identification, indicator, such as in Figure 18, where a
true strength index, TSI(close,100,20), is compared with a stochastic momentum index, SMI(1,100,20),
for a 20-minute December S&P 500 futures contract. The TSI is seen to track the close-to close variations
of the price curve: up, down, sideways. The one bar stochastic, on the other hand, gives more of a sense
of the overall direction of the market. Often, the one-bar stochastic may be used as a direct trading
vehicle signaling major turning points on a smooth and timely basis.
In Figure 19, the 20-minute price curve is shown with up sentiment as indicated by SMI(1,40,20). A
TSI(close,40,20) curve is included for comparison. The TSI curve tracks the close-to-close prices and is
greatly affected by the gap opening. The one-bar stochastic is unaffected by the gap opening continuing
its rise from one day to the next.
William Blau, (407) 368-9095, is an independent futures trader.

REFERENCES
Appel, Gerald [1985]. The Moving Average Convergence-Divergence Trading Method, Advanced
Version, Scientific Investment Systems.
Blau, William [1991]. "Double-smoothed stochastics," Technical Analysis of STOCKS &
COMMODITIES, Volume 9: January.
___ [1991] . "True strength index ," Technical Analysis of STOCKS & COMMODITIES, Volume 9: November.
___ [1992]. "Trading with the true strength index," Technical Analysis of STOCKS & COMMODITIES, Volume
10: May.
___ [1991] . "True strength index & double-smoothed stochastics," presented at the CompuTrac TAG14
Technical Analysis Seminar, New Orleans, November 20-22.
Chande, Tushar [1991]. "The midpoint oscillator," Technical Analysis of STOCKS & COMMODITIES Volume
9: November.
Lane, George C. [1984] . "Lane's stochastics," Technical Analysis of STOCKS & COMMODITIES, Volume 2:
Chapter 3.

Figures

Copyright (c) Technical Analysis Inc.

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 1: In a lookback period of q-bars, The range of the bars is HH:q-LL:q. The close is referenced
to the low point of the range.

Figures

Copyright (c) Technical Analysis Inc.

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 2: In a lookback period of q-bars, HH:q, the range of the bars is HH:q - LL:q. The close is
referenced to the midpoint of the range. Stochastic momentum is defined as: SM(q) = Close - 0.5(HH:q
+ LL:q). SM(q) is positive when close is greater than the midpoint of the range, direction up. SM(q) is
negative when close is less than the range midpoint, going down. Amount of movement of market is
determined by how large SM(q) is relative to its range.

Figures

Copyright (c) Technical Analysis Inc.

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 3: COMPARISON OF CLOSE MOMENTUM AND STOCHASTIC MOMENTUM. Close

momentum = Close - Close [20 days ago], while stochastic momentum = Close - A function of the close
[20 days ago].

Figures

Copyright (c) Technical Analysis Inc.

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 4: STOCHASTIC MOMENTUM INDEX, BASIC CONFIGURATION. Aq = 13-days stochastic is

shown. The indicator maps prices into a range from -100 to +100 on its scale. Prices are deemed to be
at high levels when the indicator is above its overbought line. Prices are said to be at low levels when
the indicator is below its oversold line. The signal line is the EMA of the SMI (q,r,s). It is normally in the
range of three to 12 bars. When the SMl is above its signal line, the trend is said to be up; a downtrend
is said to be in force when the SMl is below its signal line. Chart notation: SMI (q,r,s) = SMI (13,25,2)
means a q = 13-bar stochastic lookback with r = 25-bar first smoothing, s = two-bar second smoothing.

Figures

Copyright (c) Technical Analysis Inc.

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 5: SMOOTHING EFFECT OF STOCHASTIC MOMENTUM. For numbers of price bars in

excess of two bars, the stochastic momentum has a smoothing effect. This becomes evident when the
cumulative sum of the stochastic momentum is made. A 20-day lookback is shown. The cumulative
sum appears as a smoothed version of the price curve inclusive of the attendant lag of smoothing. A
20-day EMA of the close is also shown for comparison.

Figures

Copyright (c) Technical Analysis Inc.

10

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 6: STOCHASTIC MOMENTUM AS A PROXY FOR PRICE - VERY LARGE INTERVAL . A

large q = 300 day interval of stochastic momentum, SM(300), produces a curve that is a good
approximation of price shape. The unsmoothed stochastic momentum index, SMI(300,1,1), is an
amplitude compressed version of stochastic momentum. Here, the ASM and SMI have almost identical
shapes, although their amplitude scales are different.

Figures

Copyright (c) Technical Analysis Inc.

11

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 7: STOCHASTIC MOMENTUM SMOOTHING FOR NOISE REDUCTION. The 300-day

stochastic momentum is smoothed by a five-day EMA, resulting in a noise-reduced curve.

Figures

Copyright (c) Technical Analysis Inc.

12

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 8: SHORTER INTERVAL, ASM & SMI. A reduction in the stochastic lookback interval to 20
days with five days of smoothing sharpens the turning points and produces divergences in the SMI
relative to price where they did not exist before. Again except for scaling, the ASM and SMI appear very
similar in shape. The differences are due to compression effects in the normalization of the SMI

Figures

Copyright (c) Technical Analysis Inc.

13

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 9: LANE'S SLOW STOCHASTIC AND THE STOCHASTIC MOMENTUM INDEX. SMl(q,5,1)
approximates the slow stochastic in shape and timing with the exception of scaling. The small
differences are mainly due to the five-day EMA in lieu of the three-day SMA built into the %D (q) of the
slow stochastic.

Figures

Copyright (c) Technical Analysis Inc.

14

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 10: INCREASING THE EMA SMOOTHING THE INTERVAL. Increasing the EMA interval to
20 days obtains SMl(q,r,s) =SMI(20,20, 1), which no longer has the appearance of a fast oscilator. The
SMI now appears to trend as prices trend. Divergences appear at B, C, D, E and F. A divergence from
the beginning of B to C signals the end of a price uptrend. Also, at A the slope of the SMl trends with
price. The slow stochastic is shown for comparison; the region A remains noisily flat in an overbought
condition.

Figures

Copyright (c) Technical Analysis Inc.

15

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 11: FURTHER INCREASE OF EMA SMOOTHING INTERVAL. SMI(20,60, 1) produces a very
smooth trendline with essentially no lag introduced for the main turning points(compare with Figure 10).
Divergences exist now only at D. E, F and G. The divergences that were at B and C are now absent.
The major downtrend in March is keyed by the long divergence at G and is further supported by the
slow stochastic divergence at C. The slow stochastic may now be used as an entry (or exit) vehicle for
trading with the stochastic momentum index defining the trend of prices.

Figures

Copyright (c) Technical Analysis Inc.

16

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 12: LONG TRED ANOMALY. In the middle panel SMI(20,20, 1) is timely and smooth.
however, it gives erroneous indications of trends when the price trend is of very long duration. Starting
in April, it indicates an uptrend while, in fact, prices are declining. In the bottom panel SMI(20,150,1)
removes the error so that the indicator again trends correctly with price. Timeliness and smoothness
are present with very little additional lag.

Figures

Copyright (c) Technical Analysis Inc.

17

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 13: Designations for the DS-stochastic and stochastic momentum index formulas are
shown;q=2 defines a two-day stochastic (two daily price bars) .

Figures

Copyright (c) Technical Analysis Inc.

18

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 14: TWO-DAY STOCHASTICS AS A PROXY FOR PRICE . A large interval 300-day moving
average shows the two-day stochastic to be a good approximation of price (bar graph).

Figures

Copyright (c) Technical Analysis Inc.

19

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 15: TWO-DAYS TOCHASTIC ON T-BONDS . The smoothing interval is reduced to 20 days. A
second interval also of 20 days is used to smooth the result of the first smoothing. The effect of double
smoothing is to remove the high-frequency fluctuations and obtain a relatively smooth trending curve
that is timely (low lag) with access to major turning points in prices.

Figures

Copyright (c) Technical Analysis Inc.

20

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 16: KEY REVERSAL PATTERN . Two-day sochastics are sensitive to the location of the close
relative to the extreme of prices today and yesterday. A key reversal pattern has a large displacement
of the close from the extreme (high) today relative to yesterday's extreme (low). An immediat downtum
is generated in the SMI(2,40,20). Compare with a TSl(Close,40,20), which closely tracks the
close-to-close variations in price.

Figures

Copyright (c) Technical Analysis Inc.

21

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 17: Designation for the DS-stochastic and stochastic momentum index formulas are shown; q
=1 defines a one-day stochastic (one price bar).

Figures

Copyright (c) Technical Analysis Inc.

22

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 18: TREND IDENTIFICATION (SENTIMENT) USING ONE-DAY STOCHASTICS. One-day

stochastics detect any propensity for the close to favor either the high side or low side of the price bar.
This is a form of a sentiment indication or trend identification. See SMI(1, 100,20)of a 20 minute bar
chart of the December S&R By contrast, the TSl(close, 100,20) tracks the close-to close daily changes.

Figures

Copyright (c) Technical Analysis Inc.

23

Stocks & Commodities V. 11:1 (11-18): Stochastic Momentum by William Blau

FIGURE 19: GAPS & GAP OPENING. The gap opening takes time for the EMA(close,20) and
TSl(close,40,20) to absorb. The one-bar stochastic, SMI(1,40,20), is unaffected by the gap continuing
its ascent The overall sentiment is up. Examination of the bars reveals the closes are dominantly above
the midpoint of their 20-minute rangeover the smoothing interval.

Figures

Copyright (c) Technical Analysis Inc.

24