Absolute Strengh Histogram
This is an indicator which seems to originate in the forex world, probably at http://fxcodebase.com, but it seems difficult to track what the actual origin is.
Potentially first implemented in LUA here:
Which also says that an updated and revised LUA version is available at:
The same site hosts a version in MQL4
:
And another version for MQL5
can be found at:
All versions seem complex, due to the language and platform implementation,
with a whooping 227 lines in the MQL5
version (including some comments of
course), and have slight differences, so the need arises to settle for a sound
definition. The definition in pseudo-code after looking at the different
versions:
# p0 is the current price and p1 is the previous price if mode is RSI: bulls = 0.5 * abs(p0 - p1) + p0 - p1 bears = 0.5 * abs(p0 - p1) - p0 + p1 elif mode is STOCH: bulls = p0 - lowest(period) bears = highest(period - p1 avgbulls = moving_average(bulls, period) avgbears = moving_average(bears, period) smoothedbulls = moving_average(bulls, smoothing_period) / pointsize smoothedbears = moving_average(bears, smoothing_period) / pointsize ash = smoothedbulls - smoothedbears
The RSI
/ STOCH
naming was chosen by the original implementors and
pointsize
is also known as pipsize
which reflects its Forex origins. It
is unclear if the smoothing average has to be the same as the previous
moving average and some versions seem to work with an already averaged price
rather than with the standard price. And the moving averages are selected
using an integer.
Some decisions
-
The indicator will work with the raw priced passed to it in a data feed. If the user wants to run the indicator on an average price it can pass it, instead of passing the raw price.
-
The 0.5 multiplier in the
RSI
mode, will be a parameter -
The moving average won't be selected with any kind of integer. In backtrader one can pass the actual desired moving average as a parameter.
-
The smoothing moving average, unless specified as a parameter, will be the same as the moving average already in use in the indicator
-
The
pointsize
will not be used unless the user specifies a value for it as a parameter.
And here the implementation
class ASH(bt.Indicator): alias = ('AbsoluteStrengthOscilator',) lines = ('ash', 'bulls', 'bears',) # output lines # customize the plotting of the *ash* line plotlines = dict(ash=dict(_method='bar', alpha=0.33, width=0.66)) RSI, STOCH = range(0, 2) # enum values for the parameter mode params = dict( period=9, smoothing=2, mode=RSI, rsifactor=0.5, movav=bt.ind.WMA, # WeightedMovingAverage smoothav=None, # use movav if not specified pointsize=None, # use only if specified ) def __init__(self): # Start calcs according to selected mode if self.p.mode == self.RSI: p0p1 = self.data - self.data(-1) # used twice below half_abs_p0p1 = self.p.rsifactor * abs(p0p1) # used twice below bulls = half_abs_p0p1 + p0p1 bears = half_abs_p0p1 - p0p1 else: bulls = self.data - bt.ind.Lowest(self.data, period=self.p.period) bears = bt.ind.Highest(self.data, period=self.p.period) - self.data avbulls = self.p.movav(bulls, period=self.p.period) avbears = self.p.movav(bears, period=self.p.period) # choose smoothing average and smooth the already averaged values smoothav = self.p.smoothav or self.p.movav # choose smoothav smoothbulls = smoothav(avbulls, period=self.p.smoothing) smoothbears = smoothav(avbears, period=self.p.smoothing) if self.p.pointsize: # apply only if it makes sense smoothbulls /= self.p.pointsize smoothbears /= self.p.pointsize # Assign the final values to the output lines self.l.bulls = smoothbulls self.l.bears = smoothbears self.l.ash = smoothbulls - smoothbears
And here a view of how the indicator works