Open TheMightiestCarrot opened 2 years ago
also all parameters that start with 'nbdev' are defined as integers but the actual function expects float indicators: BBANDS, STDDEV, VAR
Where do we define nbdev
as integers?
I only see them as double nbdev
in the _func.pxi
and _stream.pxi
files.
In abstract function info
for name, value in abstract.Function("BBANDS").info['parameters'].items():
print(name ,":",type(value))
timeperiod : <class 'int'>
nbdevup : <class 'int'>
nbdevdn : <class 'int'>
matype : <class 'int'>
for example T3 vfactor is float as its supposed to be
for name, value in abstract.Function("T3").info['parameters'].items():
print(name ,":",type(value))
timeperiod : <class 'int'>
vfactor : <class 'float'>
Oh interesting, I'm not sure.
For sure the C function takes floats:
TA_RetCode TA_BBANDS( int startIdx,
int endIdx,
const double inReal[],
int optInTimePeriod, /* From 2 to 100000 */
double optInNbDevUp, /* From TA_REAL_MIN to TA_REAL_MAX */
double optInNbDevDn, /* From TA_REAL_MIN to TA_REAL_MAX */
TA_MAType optInMAType,
int *outBegIdx,
int *outNBElement,
double outRealUpperBand[],
double outRealMiddleBand[],
double outRealLowerBand[] );
Okay, resolved it -- these values are default values, but not the expected type.
For some reason, default values are made into integers if they "look like integers":
https://github.com/mrjbq7/ta-lib/blob/master/talib/_abstract.pxi#L662
You can see the expected type by inspecting the info
TA_OptInputParameterType
, which is defined here:
typedef enum
{
TA_OptInput_RealRange,
TA_OptInput_RealList,
TA_OptInput_IntegerRange,
TA_OptInput_IntegerList
} TA_OptInputParameterType;
So for BBANDS, these are type 0
or TA_OptInput_RealRange
, which is correct:
>>> abstract.Function("BBANDS")._Function__opt_inputs
OrderedDict([('timeperiod',
{'default_value': 5,
'display_name': 'Time Period',
'help': 'Number of period',
'name': 'timeperiod',
'type': 2,
'value': None}),
('nbdevup',
{'default_value': 2,
'display_name': 'Deviations up',
'help': 'Deviation multiplier for upper band',
'name': 'nbdevup',
'type': 0,
'value': None}),
('nbdevdn',
{'default_value': 2,
'display_name': 'Deviations down',
'help': 'Deviation multiplier for lower band',
'name': 'nbdevdn',
'type': 0,
'value': None}),
('matype',
{'default_value': 0,
'display_name': 'MA Type',
'help': 'Type of Moving Average',
'name': 'matype',
'type': 3,
'value': None})])
This, I think, sort of resolves your question -- although there may be a valid discussion on why we are int()
on a double value...
Could you explain the original question about input price names?
Hello,
i have found some inconsistencies in abstract function definitions. Inputs are mainly refered like this:
OrderedDict([('prices', ['high', 'low', 'close'])])
But sometimes like this:OrderedDict([('price0', 'high'), ('price1', 'low')])
If this is when functions input is any ndarray:
OrderedDict([('price', 'close')])
then why OBV is like this:OrderedDict([('price', 'close'), ('prices', ['volume'])])