Chaikin Money Flow indicator: NaN when positive

[RolandGuijt]

The CMF only seems to return a negative number or NaN.

[xmatthias]

CMF is a oscillator, so it is supposed to move to positive and negative.

Also, as almost every indicator, it has a startup period, where values will be NAN.

Testing this on a random dataframe, it does return both <0 and > 0.

What precise version are you using? and what parameters are you calling cmf with?

[RolandGuijt]

I'm using version 1.21 and I'm calling: cmf(dataframe, 20), so 20 periods. Below is a dataset for ETH/BTC with 5 min candles. According to tradingview the CMF on 23:55 should be 0.06, the function says -0.470299. At 0:00 it should be -0.05, the function reports -0.524170. I'm not ruling out that I'm missing something but I can't figure out what. date open high low close volume cmf 269 2020-12-30 19:55:00+00:00 0.026146 0.026180 0.026111 0.026178 4394.747 -0.376058 270 2020-12-30 20:00:00+00:00 0.026165 0.026197 0.026068 0.026100 2765.282 -0.411997 271 2020-12-30 20:05:00+00:00 0.026100 0.026145 0.026090 0.026138 1756.803 -0.390863 272 2020-12-30 20:10:00+00:00 0.026134 0.026144 0.026106 0.026124 756.670 -0.385205 273 2020-12-30 20:15:00+00:00 0.026116 0.026166 0.026103 0.026135 784.604 -0.383641 274 2020-12-30 20:20:00+00:00 0.026135 0.026154 0.026113 0.026140 582.773 -0.362034 275 2020-12-30 20:25:00+00:00 0.026141 0.026150 0.026083 0.026096 701.193 -0.368587 276 2020-12-30 20:30:00+00:00 0.026094 0.026113 0.026042 0.026046 624.730 -0.382809 277 2020-12-30 20:35:00+00:00 0.026047 0.026053 0.025905 0.025968 2579.020 -0.410907 278 2020-12-30 20:40:00+00:00 0.025969 0.026048 0.025964 0.026033 1114.515 -0.370010 279 2020-12-30 20:45:00+00:00 0.026030 0.026071 0.026014 0.026040 826.861 -0.387976 280 2020-12-30 20:50:00+00:00 0.026046 0.026091 0.026020 0.026076 1066.134 -0.434876 281 2020-12-30 20:55:00+00:00 0.026064 0.026139 0.026047 0.026106 1695.059 -0.427412 282 2020-12-30 21:00:00+00:00 0.026117 0.026245 0.026115 0.026219 1584.855 -0.391172 283 2020-12-30 21:05:00+00:00 0.026214 0.026225 0.026167 0.026197 741.950 -0.407562 284 2020-12-30 21:10:00+00:00 0.026198 0.026214 0.026162 0.026189 1095.265 -0.401517 285 2020-12-30 21:15:00+00:00 0.026190 0.026221 0.026164 0.026169 1136.397 -0.426442 286 2020-12-30 21:20:00+00:00 0.026169 0.026185 0.026087 0.026100 1135.345 -0.434778 287 2020-12-30 21:25:00+00:00 0.026103 0.026162 0.026088 0.026099 733.538 -0.446412 288 2020-12-30 21:30:00+00:00 0.026105 0.026108 0.026027 0.026030 774.119 -0.451525 289 2020-12-30 21:35:00+00:00 0.026029 0.026051 0.026001 0.026018 533.381 -0.537151 290 2020-12-30 21:40:00+00:00 0.026018 0.026018 0.025926 0.025926 2020.862 -0.552504 291 2020-12-30 21:45:00+00:00 0.025926 0.025966 0.025916 0.025916 688.260 -0.602331 292 2020-12-30 21:50:00+00:00 0.025920 0.026037 0.025831 0.025881 1747.952 -0.617141 293 2020-12-30 21:55:00+00:00 0.025881 0.025925 0.025865 0.025894 777.365 -0.618045 294 2020-12-30 22:00:00+00:00 0.025892 0.025914 0.025886 0.025896 244.758 -0.625711 295 2020-12-30 22:05:00+00:00 0.025896 0.025909 0.025839 0.025899 1080.082 -0.596523 296 2020-12-30 22:10:00+00:00 0.025900 0.025964 0.025883 0.025942 353.650 -0.581393 297 2020-12-30 22:15:00+00:00 0.025955 0.026008 0.025949 0.026004 437.009 -0.570979 298 2020-12-30 22:20:00+00:00 0.026004 0.026024 0.025951 0.025962 332.510 -0.598864 299 2020-12-30 22:25:00+00:00 0.025958 0.025973 0.025896 0.025896 470.276 -0.611419 300 2020-12-30 22:30:00+00:00 0.025895 0.026003 0.025881 0.025994 1269.457 -0.597840 301 2020-12-30 22:35:00+00:00 0.025994 0.025999 0.025941 0.025944 420.917 -0.629298 302 2020-12-30 22:40:00+00:00 0.025943 0.025976 0.025901 0.025961 495.324 -0.657674 303 2020-12-30 22:45:00+00:00 0.025961 0.026012 0.025957 0.025996 570.047 -0.652814 304 2020-12-30 22:50:00+00:00 0.025996 0.026025 0.025980 0.025980 396.648 -0.673699 305 2020-12-30 22:55:00+00:00 0.025980 0.026003 0.025920 0.025952 561.041 -0.653465 306 2020-12-30 23:00:00+00:00 0.025945 0.025999 0.025943 0.025949 650.017 -0.647474 307 2020-12-30 23:05:00+00:00 0.025947 0.026008 0.025947 0.025980 755.003 -0.627455 308 2020-12-30 23:10:00+00:00 0.025982 0.026048 0.025979 0.026035 764.204 -0.586594 309 2020-12-30 23:15:00+00:00 0.026040 0.026045 0.025992 0.026030 607.163 -0.571331 310 2020-12-30 23:20:00+00:00 0.026030 0.026037 0.025966 0.025991 971.223 -0.513069 311 2020-12-30 23:25:00+00:00 0.025996 0.026016 0.025984 0.025996 427.070 -0.491516 312 2020-12-30 23:30:00+00:00 0.025996 0.026028 0.025986 0.025993 691.709 -0.472941 313 2020-12-30 23:35:00+00:00 0.025989 0.026003 0.025938 0.025983 598.950 -0.461949 314 2020-12-30 23:40:00+00:00 0.025990 0.026070 0.025981 0.026058 1509.201 -0.421678 315 2020-12-30 23:45:00+00:00 0.026054 0.026147 0.026054 0.026113 1821.780 -0.435783 316 2020-12-30 23:50:00+00:00 0.026115 0.026143 0.026092 0.026127 668.995 -0.434147 317 2020-12-30 23:55:00+00:00 0.026131 0.026136 0.026030 0.026046 911.799 -0.470299 318 2020-12-31 00:00:00+00:00 0.026045 0.026093 0.025913 0.025922 2138.084 -0.524170

…

On Fri, Jan 15, 2021 at 7:31 AM Matthias ***@***.***> wrote: CMF is a oscillator, so it is supposed to move to positive and negative. Also, as almost every indicator, it has a startup period, where values will be NAN. Testing this on a random dataframe, it does return both <0 and > 0. What precise version are you using? and what parameters are you calling cmf with? [image: 2021-01-15-072947_799x618_scrot] <https://user-images.githubusercontent.com/5024695/104689834-8e5e5f80-5703-11eb-96a8-10c05744ad55.png> — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#144 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADB3GJXPSNVUYVQR6CYSVSDSZ7OMXANCNFSM4WC47VUQ> .

[xmatthias]

Technical is following the definition chaikin_money-flow1 money_flow_2 - investopedia as far as i can see (feel free to correct me if you think differently based on the code below).

technical/technical/indicators/volume_indicators.py

Lines 25 to 32 in 3323c1a

	def chaikin_money_flow(dataframe, period=21):
	mfm = (
	(dataframe['close'] - dataframe['low']) - (dataframe['high'] - dataframe['close'])
	/ (dataframe['high'] - dataframe['low'])
	)
	mfv = mfm * dataframe['volume']
	cmf = mfv.rolling(period).sum() / dataframe['volume'].rolling(period).sum()
	return cmf

If tradingview calculates differently, then there is nothing we can do to align, really, as their code is closed-source as far as i know, so eventual mistakes in their formulas are hidden from analysis.
Tradingview should however not be seen as reference, but the original papers of indicators should be. If Tradingview deviates from the original version, then this doesn't make "their" version the correct one.

You can see the logic for a Tradingview indicator by clicking on the {} button of the indicator, this is just next to the indicator settings. CMF in Tradingview:

//@version=4
study(title="Chaikin Money Flow", shorttitle="CMF", format=format.price, precision=2, resolution="")
length = input(20, minval=1)
ad = close==high and close==low or high==low ? 0 : ((2*close-low-high)/(high-low))*volume
mf = sum(ad, length) / sum(volume, length)
plot(mf, color=#459915, title="MF")
hline(0, color=#606060, title="Zero", linestyle=hline.style_dashed)

This part ((2*close-low-high)/(high-low))*volume looks wrong to me and does not follow specifications. It is quite easy to make your own indicator in Tradingview. After you click that {} button the "Pine editor" is opened, top right of that editor has a button with 3 dots -> "Make a copy".
https://www.tradingview.com/pine-script-docs/en/v4/Quickstart_guide.html