[IMPROVE] NaN values in PVGO

[giraffeingreen]

For the following code snippet:

from financetoolkit import Toolkit

toolkit = Toolkit(["NVDA", "AAPL","BRK.B"], api_key=API_KEY, start_date="2000-01-01")

pvgo = toolkit.models.get_present_value_of_growth_opportunities()

I get the following result:

	AAPL	BRK.B	NVDA
2000	0.4451	207.172	1.3462
2001	0.3214	98.699	5.283
2002	0.2313	114.778	1.0905
2003	0.3125	20.3332	1.677
2004	nan	5.1102	1.6264
2005	nan	-25.3974	1.8248
2006	nan	3.8913	nan
2007	nan	-39.0508	nan
2008	nan	81.6841	nan
2009	nan	48.8492	4.3185
2010	nan	31.844	3.6831
2011	nan	-113.76	-2.3902
2012	nan	32.7019	1.6914
2013	10.6612	91.6904	3.0434
2014	10.24	75.3026	3.5505
2015	-197.63	-482.481	-16.096
2016	7.5209	57.8841	24.4573
2017	30.7936	110.427	46.3468
2018	93.9601	248.859	44.3535
2019	64.6675	90.3009	56.1368
2020	113.997	109.681	125.633
2021	159.81	89.0447	290.602
2022	155.176	209.412	154.967
2023	170.215	102.937	491.612
2024	nan	nan	815.209

Does the FMP API miss something?

[JerBouma]

Solved! Total Debt was 0 for Apple which is why you see the NaNs. Please update to v1.8.5.

[giraffeingreen]

NVDA still exhibit a NaN value for 2008? Is this the anticipated behavior?

[JerBouma]

Ah I see, probably because interest expense is NaN in that case. I'll adjust this later.

[giraffeingreen]

Is that because we use WACC instead of the cost of equity, I can implement a flag to swap between using WACC or cost of equity.

[JerBouma]

Feel free to do so! Makes sense to have both.

[giraffeingreen]

I also talked to one of my finance teachers, and we figured it'd be better to go with the estimatedNetIncomeAvg from https://site.financialmodelingprep.com/developer/docs/analyst-estimates-api instead of using earnings from current period.

[JerBouma]

That's indeed also a good option, note that this is obtainable directly from the Finance Toolkit as well: https://www.jeroenbouma.com/projects/financetoolkit/docs#get_analyst_estimates (but I think you are aware of that 😉 )