Now, let's go into some details...
In this test the patterns were found and drawn by MPS. The algorithm of swing identification was set to "prex(12)". This means I was looking at highest highs of past 12 candles and lowest lows of past 12 candles as swing . The FastBacktest utility automatically wrote down all patterns details including time, entry, stop, target, B ratio, C ratio and more into excel spreadsheet. This methodology of market research eliminates subjectiveity, human errors and it compresses the time. But you can easily verify the results manually (old-school) if you wish.
2) Trade plan
I was testing naked pattern. No reversal confirmations by indicators, priceaction or oscillators. Simple set-and-forget. If it works then it works, if not then not.
Entry: Limit order at 0.786XA (always).
Target: Risk distance calculated and then projected to have 1:1RR ratio.
Result: A pattern was considered a winner if priceaction reverses and hits takeprofit before it hits stoploss.
Testing period was 1999-2019 (20 years!). Pair was AUDJPY. Timeframe was 30M. Total amount of patterns found was 746. Left-top corner number is the probability of win. Right-top corner number is the total number of setups. Left-bottom corner number is the losses count. Right-bottom corner number is the wins count.
I think objectivity is the key to any scientific research. Next time, when you read a book, an article, watch a video or attend a pattern trading course ask yourself a question: "Are the results they are talking about really testable, verifiable and repeatable?" Run away from those mentors who don't tell you the swing identification method. Based on my research, different swing identification methods produce different results. If you would ever wish to use this table in your trading make sure that you find swing with prex(12) method. This research is only true for AUDJPY , 30M.
In my trading I go with the pattern (standard way) if the cell is green and I go against the pattern (reverse way) if the cell is red. This way I always stay on a higher probabilities side.
20 years on 30M timeframe is aproximately 250 000 candles. An average prex(12) swing contains 20 candles. Every pattern has 4 swings: xa, ab, bc and cd. 20*4=80. This means that an avarage pattern with a prex(12) algorithm is 80 candles. 250 000/80= 3125 four-swing groups. 746/3125= 0,238. This means that if you randomly pick up four adjucent swings from the chart there's almost 24%(!) probability that you will have a pattern. A random group of four swings is 24% probability a pattern! It's amazing how often they appear if you take into consideration all the ratio requirements for Gartleys/Bats.
Do the backtest before you argue. AUDJPY, 30M, 1999-2019, Gartleys and Bats, prex(12) algorithm.