Munich Personal RePEc Archive

A New Adaptive Moving Average

Pierrefeu, Alex

30 May 2019

Online at https://mpra.ub.uni-muenchen.de/94323/
MPRA Paper No. 94323, posted 11 Sep 2019 05:42 UTC
 A NEW ADAPTIVE MOVING AVERAGE (VAMA) TECHNICAL
INDICATOR FOR FINANCIAL DATA SMOOTHING

A P REPRINT

Alex Pierrefeu
Sevilla, Dos Hermanas, 41700
[email protected]

May 30, 2019

A BSTRACT
The separation of the trend from random fluctuations (noise) is a major objective in technical analysis
and for a long time two commons filters, the simple moving average and the exponential moving
average have been used to achieve this goal, those two filters use one parameter to control this degree
of separation, higher degree of separation involve smoother results but also more lag. Lag is defined
as the effect of a moving average to show past trends instead of new ones, this effect his unavoidable
with causal filters and is a major drawback in decision timing . In this article I will introduce a new
adaptive moving average technical indicator (VAMA) who aim to provide smooth results as well as
providing fast decision timing. This new method will be used for the construction of a simple MA
crossover strategy in EURUSD, the results of this strategy will then be compared to the results of the
same strategy using other adaptive moving averages to provide a comparison of the profitability of
this indicator.

Keywords Moving Average · Adaptive Moving Average · Smoothing · Filters · Technical indicator · Technical
Analysis · Volatility

1 Introduction
One of the main characteristic of the financial market is its dynamic condition, so in order to respond to this dynamical
behaviour as well as addressing solutions to the lag induced by classical filters various adaptive filtering methods have
been proposed. Adaptive filters aim to adapt to characteristics of the market price, they might tend to adapt such that
smoothness increase or decrease when market act in a certain way, some methods adapt to market trend strength [1],
to market shape complexity [2], to the rate of change of the phase [3] or to market volatility [4]. Adapting to market
characteristics require to quantify those characteristics in a smoothing variable αn who will control the smoothness of
the filter. The smoothing variable used for the calculation of VAMA is a modified version of the ratio of open-close to
high-low range indicator. This indicator originally measures the dispersion of the absolute difference of the close price
with the open price relative to the difference of the high price with the low price and is calculated as follows : |C n −On |
Hn −Ln
where Cn is the closing price, On is the open price, Hn the high price and Ln is the low price.

2 Moving Average Construction

|SM Ap (Cn ) − SM Ap (On )|

A simple moving average of period p defined as SM Ap is used to smooth each price components in order to increase
smoothness (decrease αn ) when p increase.

3 Strategy Construction and Testing

In order to test the profitability of the VAMA indicator, a simple MA crossover strategy will be tested with EURUSD,
results will then be compared to the same strategy using other adaptive moving average instead of VAMA. The MA
crossover strategy is a simple strategy who use the crosses of a fast moving average with a slow moving average to
generate signals, a common period is 50 for the fast moving average (SM A50 ) and 200 for the slow moving average
(SM A200 ), when SM A50 cross over SM A200 a buy signal is generated and when SM A50 cross under SM A200 a
sell signal is generated. For this strategy SM A50 will be replaced by VAMA with p = 50. The other adaptive moving
averages used for comparison will include the Kaufman adaptive moving average[1] (KAMA) with period = 50, Fractal
adaptive moving average[2] (FRAMA) with period = 50, Mesa Adaptive Moving Average[3] (MAMA) with fast limit
= 0.1 and slow limit = 0.02, Variable Index Dynamic Average[4] (VIDYA) with period = 25, Parameters have been
selected to reduce the difference between each moving average in order to have more accurate testing results.

3.2 Testing

The strategy is tested from 2018-02-06 to 2019-05-06 in 1H time frame and dollar as base currency, no commis-
sion/spread have been applied to the test. Previous positions are exited when a new position is opened.
Table 1: Statistics of the strategy applied to EURUSD with order size of 1000 contracts
V AM A50 KAM A25 F RAM A50 M AM A0.1,0.02 V IDY A10
Net Profits -57.34 -132.5 -38.28 -74.48 -84.27
Gross Profit 135.21 123.39 174.44 143.96 133.72
Gross Loss 192.55 255.64 212.72 218.44 217.99
Max Drawdown -78.76 -132.25 -83.64 -128.11 -98.46
Total Closed Trades 59 61 72 58 61
Number Winning Trades 15 12 18 17 18
Number Loosing Trades 44 49 54 41 41
Average Profit per Trade -0.97 -2.17 -0.53 -1.28 -1.38

4 Conclusion
I have presented a new adaptive moving average that adapt to a modified version of the ratio of open-close to high-low
range indicator. Testing results show that the classic MA cross strategy is not profitable, however VAMA proved to have
good results among other adaptive moving averages, this show that it can be interesting to adapt to market price based
on high/low/open and not only closing price information, thus making the modified ratio of open-close to high-low
range indicator an interesting smoothing variable for other adaptive indicators.

5 Pinescript Code
//@version=2
study("VAMA",overlay=true)
length = input(14)
//----
c = sma(close,length)
o = sma(open,length)
h = sma(high,length)
l = sma(low,length)
lv = abs(c-o)/(h - l)
//----
ma = lv*close+(1-lv)*nz(ma[1],close)
plot(ma,color=#FF0000,transp=0)

2
 A PREPRINT - M AY 30, 2019

References
[1] P. Kaufman, Trading Systems and Methods, John Wiley Sons, Third Edition (2008).
[2] J. Ehlers, "Fractal Adaptive Moving Average", Technical Analysis of Stock Commodities”, October 2005.
[3] MAMA, Ehlers, John F., “Rocket Science for Traders”, New York, John Wiley Sons, 2001
[4] VIDYA, Chande, Tushar S. and Stanley Kroll, “The New Technical Trader”, New York, John Wiley Sons, 1994