Documentation of crossover

[prRZ5F4LXZ]

This is about the documentation of crossover.

https://kernc.github.io/backtesting.py/doc/backtesting/lib.html#backtesting.lib.crossover

It just says Return True if series1 just crossed over series2. Does it mean series1 crosses series2 from below or from above?

https://www.investopedia.com/terms/c/crossover.asp

Investopedia says the following.

The crossover is a point on the trading chart in which a security's price and a technical indicator line intersect, or when two indicators themselves cross.

It does not mention whether it is crossing from above or from below. So I'd assume crossover means from below and above.

https://github.com/kernc/backtesting.py

But the example on the README seems to mean crossover is from below. If this is the case, should the function crossover() be renamed to crossoverfrombelow()?

Expected Behavior

NA

Actual Behavior

NA

Steps to Reproduce

NA

Additional info

NA

[zlpatel]

Here's another snip from investopedia.(https://www.investopedia.com/terms/g/goldencross.asp)

How do I identify a golden cross on a chart?

The golden cross occurs when a short-term moving average crosses over a major long-term moving average to the upside and is interpreted by analysts and traders as signaling a definitive upward turn in a market. Some analysts define it as a crossover of the 100-day moving average by the 50-day moving average; others define it as the crossover of the 200-day average by the 50-day average. Basically, the short-term average trends up faster than the long-term average, until they cross.

Crossover is used in reference to series 1 and series 2. you switch their position in the crossover function when you need to detect the opposite crossover. But the function stays the same.

Crossover(A,B) where A crosses over B. Meaning of crossover will change depending on whether A is faster moving data or slower.

For direction agnostic cross there's a separate function: https://kernc.github.io/backtesting.py/doc/backtesting/lib.html#backtesting.lib.cross

[prRZ5F4LXZ]

Then, an alternative name could be goldencross.

But the name of crossover is still ambiguous as it does not say the direction of cross. But the implementation of the crossover function clearly means from below. Therefore, I think the name of the crossover() function should be renamed to reflect what it actually does.

[rbscholtus]

One indicator goes over (/above) another so it crosses over, implying it's from below to above. It's not terribly black and white in it's meaning but not really a problem either. I'll try to make a pull request for improved documentation if I get a chance.

We could potentially add crossunder() except this word isn't backed by proper English.

Golden crosses and death crosses are more specific than just one indicator crossing another in general.

[prRZ5F4LXZ]

https://www.dictionary.com/browse/crossover

I don't think that is a correct interpretation of 'crossover'. 'over' just means that one line is crossed on top of another one ('top' in the direction perpendicular to the plane in which the lines are drawn).

This is clearly demonstrated in the context of a figure in genetics in which there is no above or below implied by over.

https://www.genome.gov/genetics-glossary/Crossing-Over

[rbscholtus]

Genetics is a little different. We're talking about one (financial) value going over another, so, not under.
I know there could be interpretations, but this is mine.

[prRZ5F4LXZ]

I know that is for genetics. But I am talking about English. The genetics example just proves that 'over' does not mean 'above'.

[eervin123]

To me, a native English speaker and someone who has been in finance for over 20 years. See what I did there. ;)

I read this perfectly as over meaning above. Every chart that I think of in finance is a plane where up/over/above/rising etc. are all directionally relevant. These are not floating objects in a spatial void where the concept of up down, left and right are absent. When I hear, "When the fast moving average crosses over the slow moving average, you do X" I instantly picture the fast moving average going higher, or above the slow moving average. So, when A crosses over B is perfectly understandable in the context of financial assets rising and falling.

I suppose, with no context, as in the dictionary above for railroad tracks, on the horizontal plane, crossover means left to right and that would indeed be ambiguous, but this is just one person's opinion.

I'm not sure this needs any further explanation, but to clear any doubt, I suppose you could add a statement to the documentation.

[fortesantonio]

Two ideas that might work for Backtesting.py crossovers Libraries ;

def crossesOver(stream1, stream2):
if isinstance(stream2, int) or isinstance(stream2, float):
if stream1[len(stream1)-1] <= stream2:
return False
else:
if stream1[len(stream1)-2] > stream2:
return False
elif stream1[len(stream1)-2] < stream2:
return True
else:
x = 2
while stream1[len(stream1)-x] == stream2:
x = x + 1
if stream1[len(stream1)-x] < stream2:
return True
else:
return False
else:
if stream1[len(stream1)-1] <= stream2[len(stream2)-1]:
return False
else:
if stream1[len(stream1)-2] > stream2[len(stream2)-2]:
return False
elif stream1[len(stream1)-2] < stream2[len(stream2)-2]:
return True
else:
x = 2
while stream1[len(stream1)-x] == stream2[len(stream2)-x]:
x = x + 1
if stream1[len(stream1)-x] < stream2[len(stream2)-x]:
return True
else:
return False

def crossesUnder(stream1, stream2):
if isinstance(stream2, int) or isinstance(stream2, float):
if stream1[len(stream1)-1] >= stream2:
return False
else:
if stream1[len(stream1)-2] < stream2:
return False
elif stream1[len(stream1)-2] > stream2:
return True
else:
x = 2
while stream1[len(stream1)-x] == stream2:
x = x + 1
if stream1[len(stream1)-x] > stream2:
return True
else:
return False
else:
if stream1[len(stream1)-1] >= stream2[len(stream2)-1]:
return False
else:
if stream1[len(stream1)-2] < stream2[len(stream2)-2]:
return False
elif stream1[len(stream1)-2] > stream2[len(stream2)-2]:
return True
else:
x = 2
while stream1[len(stream1)-x] == stream2[len(stream2)-x]:
x = x + 1
if stream1[len(stream1)-x] > stream2[len(stream2)-x]:
return True
else:
return False