Buy Low, Sell High. The Objective Wave Pattern Edge. Michael Gould. A Presentation for the Market Technicians Association

Transcription

1 Buy Low, Sell High The Objective Wave Pattern Edge A Presentation for the Market Technicians Association Michael Gould

2 Arthur A. Merrill One of the tools he gave us: Filtered Waves % Swing Filter Volatility instead of Price Renamed Zig-Zag by someone..

3

4

5

6

7 3 Pairs of MW Wave Patterns: M12:W5, M13:W4, M7:W10 M Bullish W - Bearish 3 Sources: 1 Larry Pesavento s work on Gartley and Butterfly patterns 2 Observation 3- Elliott Wave

8 Edge?

9 Edge

10 Reward to Risk One of the key areas where chart patterns help quite a lot. 1. Can be understood in terms of chart pattern 2. Risk can be defined, Reward can only be estimated 3. Only viable information when win rate is considered (as win rate also requires Reward to Risk information to be a valuable metric)

11 Probability or Win Rate of M&W Patterns How is the test designed? - Merrill tested the 5% M and W patterns on a large number of stocks. He considered the patterns in pairs and ranked bullish or bearish by second pattern s final point being higher or lower to the first pattern s final point. How is environment defined? (what is the market model?)

12

13

14

15

16

17 Pattern Only

18 Percent Retracement Entry Method Used in all tests presented M used for Bullish Retracement Entries Exact entry is using the 20% A-D pattern retracement in all tests that follow. W used for Bearish Retracement Entries All kinds of entries are possible (momentum, volatility, buying early, candle/bar pattern, P&F pattern, volume, etc) each with different performance metrics Again Only 3 pairs of 16 are being considered in the following examples

19

20

21

22

23

24

25 3-day Swing Trade Single Pattern 1 minute chart

26

27 3-day Swing Trade Several Patterns 1 minute chart

28

29

30

31

32 Do Pattern Entries in a trend give us an advantage over random?

33 Comparison of Pattern Entries in Intraday Trends with Random Entries 5 pip Stop Loss Examples - 1m data Trade Date_Trend_Began Pattern WinRate NumPatternTradesTotal Random WinRate NumRandomTradesTotal GBPJPY 3/8/ % % 13 GBPJPY 1/26/ % 45 31% 13 EURUSD 3/8/ % 4 33% 6 EURJPY 3/7/ % % 68 AUDUSD 2/5/ % 45 39% 51 AUDUSD 1/23/ % % 61 USDCAD 3/6/ % % 20 USDCAD 3/5/ % 24 32% 25 GBPUSD 6/10/ % % 12 GBPUSD 6/9/ % 10 27% 11 GBPUSD 6/6/ % 6 20% 5 GBPUSD 6/5/ % 13 40% 5 GBPUSD 6/3/ % % 8 GBPUSD 6/2/ % 8 0% 10 GBPUSD 6/8/ % 14 20% 10 WeightedAvgWavePatterns = 52% WeightedAvgRandom = 36% Caveats: 1) This test is not exhaustive even though it is supportive 2) Pattern Win Rate vs Random Entry can be compared on basis of whether a tight stop loss was hit or not however pattern has SIZE as a variable (which influences win rate also) this test considered constant limited number of total size possibilities. 3) The exact pattern entry method used is only one of many possibilities (which could influence win rate)

34 Pattern Win Rate with respect to Practical Strategy SL 5 pips is a very close stop loss.. What happens when we are a little more generous, 10, 15, 30 pip stop losses? What happens in a random entry approach when we do the same? Snapshot The worst pattern result from sample: AUDUSD February 5 th, pip SL had 32% win rate Pattern Win Rate 10 pip SL 40.7% 15 pip SL 46.7% 30 pip SL 80.0% Random 10 pip SL 26.8% 15 pip SL 43.8% 30 pip SL 65.0%

35

36 Wave Pattern Random

37 Further Concepts For One Type (e.g. M only for 5 point) 3 Point Patterns 2 Forms 4 Point Patterns 5 Forms 5 Point Patterns 16 Forms 6 Point Patterns?

38

39

40 3,4,5 2,5,16

41 4x 2-25x+41 is this curve fit? Definitely. Will it hold? Possibly. Going back to this formula, if it correctly describes the relation between Number of Pattern Points and Total Number of Patterns we can extrapolate: >>> def patpivot(x): return 4*x**2-25*x+41 >>> patpivot(3) 2 >>> patpivot(4) 5 >>> patpivot(5) 16 >>> patpivot(6) 35 >>> patpivot(7) 62 >>> patpivot(8) 97 >>> patpivot(9) 140 >>> patpivot(10) 191 >>> patpivot(2) gives 7, which is nonsense from an Objective Wave Pattern analysis perspective.. Just as Merrill s testing showed certain patterns to be more reliable than others, its quite possible that some of the more complex Objective Wave Patterns could hold probability edges. Even if probability edges are limited, the reward to risk skewing can be taken advantage of inside other probability contexts.

42 Michael Gould patterntrading.com Chart Pattern Indicators, Strategy for TradeStation & MultiCharts