One can always go directly to Github to view the actual code for any factors. In this case, the code for the RSI factor can be found here and is

class RSI(SingleInputMixin, CustomFactor):  
    """  
    Relative Strength Index  
    **Default Inputs**: :data:`zipline.pipeline.data.EquityPricing.close`  
    **Default Window Length**: 15  
    """  
    window_length = 15  
    inputs = (EquityPricing.close,)  
    window_safe = True

    def compute(self, today, assets, out, closes):  
        diffs = diff(closes, axis=0)  
        ups = nanmean(clip(diffs, 0, inf), axis=0)  
        downs = abs(nanmean(clip(diffs, -inf, 0), axis=0))  
        return evaluate(  
            "100 - (100 / (1 + (ups / downs)))",  
            local_dict={'ups': ups, 'downs': downs},  
            global_dict={},  
            out=out,  
        )

The discrepancy you are seeing probably stems from using a window_length=2. The common RSI definition uses the past 14 days price differences or a window_length=15 (need one more day to get 14 days of differences). This is the default. The other small issue is the Quantopian implementation uses the mean of the differences while some sources use the exponentially weighted mean.

Hope that helps.

Hi Dan,

Thanks again for your kindle support.

I didnt response because i have been thinking in your answer these days.

For sure, you know RSI is a classic factor and RSI 2 period is a proven rsi lenght in several strategies. I have been witness of the good performance using 2 periods.

Im really confused. It seems that RSI here in quantopian works different than others platforms . All of them are based on the same formula and shows very similiar values. So strategies reach the same results. Why not in Quantopian? :(

The github formula seems very similar of the classic one but im not an expert to confirm this.

Im an amibroker user moving here to quantopian looking to extend my possibilities with your good fundamentals datasets and others features.

I think the solution is to build my own rsi through custom factor, will learn for.

Thanks again!