Automating Al Brooks price action

Hi,

I am trying to automate the way a price action trader sees a chart. I mostly rely on Al Brooks descriptions. An automated analyzer should recognize 2 basic patterns of price action: Trend and Trading Range and identify their parameters.

Al Brook mentions quite often that trades of big institutions are mostly (if not entirely) are done by computers. He often says something like “many algorithms that institutions use will see this as a major high…”. Such statements raised my curiosity if these algorithms could be somehow replicated.

I’ve been working on this task for a while. I can admit that developing such an analyzer takes enormous amount of time and efforts. The level of complexity is testing my limits as an engineer (I am quite good at software engineering). At this point I do have a prototype, but it’s nowhere near production quality.

Below are normal distributions for trends for 2019 and 2018, EURUSD, M5 which I am getting from my prototype.

If you trade EURUSD M5 price action - does this statistics make sense to you?

This data is for 24/5 analysis. Filtering it to New York session reduces amount of trends roughly by 2.5, trading ranges - by almost 4 (for 2019), and increases pips size, but not dramatically (38->48). Other parameters stay similar.

Description:

Bars - trend length in bars
Pips - trend size in pips
Pushes - number of pushes (breakouts) in a trend
BO Bars - breakout length in bars
BO Pips - breakout size in pips
PB Bars - pullback length in bars
PB Percent - pullback size as a percentage of previous breakout
The number # at the end of each line is N - the number of items (trends, breakouts, pullbacks) to calculate statistics.

2019, EURUSD, M5
Trends: 535, up: 254, down: 281; TRs: 253

Bars {mean=68.92; gmean=50.44; stdev=52.46; cv=0.76; var=2752.42; kurt=3.47; skew=1.54; min=4.00; max=354.00; #535}
Pips {mean=38.62; gmean=32.76; stdev=23.48; cv=0.61; var=0.00; kurt=4.16; skew=1.68; min=4.10; max=168.80; #535}
Pushes {mean=4.81; gmean=4.06; stdev=2.85; cv=0.59; var=8.10; kurt=2.43; skew=1.35; min=1.00; max=18.00; #535}
BO Bars {mean=7.13; gmean=5.94; stdev=4.96; cv=0.69; var=24.55; kurt=19.29; skew=3.22; min=1.00; max=55.00; #2358}
BO Pips {mean=13.35; gmean=10.97; stdev=9.80; cv=0.73; var=0.00; kurt=15.61; skew=2.97; min=2.00; max=114.90; #2358}
PB Bars {mean=13.05; gmean=8.69; stdev=12.08; cv=0.93; var=145.93; kurt=5.57; skew=1.93; min=1.00; max=89.00; #1816}
PB Percent {mean=0.63; gmean=0.57; stdev=0.28; cv=0.44; var=0.08; kurt=2.12; skew=0.91; min=0.08; max=2.24; #1816}

Longest trend bars: 354 2019-12-26 21:55:00 … 2019-12-30 03:20:00
Largest trend pips: 168.8 2019-01-02 08:05:00 … 2019-01-02 22:55:00
Trend max pushes: 18 2019-06-19 14:05:00 … 2019-06-20 12:35:00

2018, EURUSD, M5
Trends: 626, up: 301, down: 325; TRs: 254

Bars {mean=73.17; gmean=55.59; stdev=50.96; cv=0.70; var=2597.12; kurt=1.49; skew=1.14; min=4.00; max=324.00; #626}
Pips {mean=57.93; gmean=50.05; stdev=31.80; cv=0.55; var=0.00; kurt=4.87; skew=1.47; min=10.90; max=275.70; #626}
Pushes {mean=5.66; gmean=4.82; stdev=3.19; cv=0.56; var=10.17; kurt=1.86; skew=1.15; min=1.00; max=21.00; #626}
BO Bars {mean=6.40; gmean=5.39; stdev=4.46; cv=0.70; var=19.86; kurt=30.69; skew=4.02; min=1.00; max=60.00; #3200}
BO Pips {mean=18.17; gmean=15.15; stdev=12.12; cv=0.67; var=0.00; kurt=10.23; skew=2.36; min=1.70; max=134.30; #3200}
PB Bars {mean=11.94; gmean=7.98; stdev=11.31; cv=0.95; var=128.02; kurt=10.29; skew=2.30; min=1.00; max=118.00; #2549}
PB Percent {mean=0.65; gmean=0.58; stdev=0.29; cv=0.45; var=0.08; kurt=1.35; skew=0.87; min=0.06; max=1.97; #2549}

Notice that the number of trends and trading ranges in both years are comparable.
Most of trend parameters stay similar except for pips size, which dropped in 2019.
The number of pushes per average trend has reduced as well.

I’d like to know if someone here tried to develop such price action analyzer and what are your conclusions?
Or maybe you’ve heard of someone’s else experience?
Is this analysis really so complicated to automate or it’s just me?
And then - are there already known tips on using such a tool?

Al Brooks has been trading his style for about 20 years. If it could have been automated do you think he would have done it by now?? Also have you read any of his books or completed his course?? The reason I ask is that the algorithms companies use are short tom algos that usually work of support and reistance levels

Just a thought.

Cheers

Blackduck

Really appreciate this research am working on it seriously.

I’ve read his first book, then took his course (new version).
In terms of price action trading I have not seen such level of details from other authors.
Hence - I can say that this part of my work is mostly inspired by Al Brooks.

However, I can’t say that I am automating his trading system. I’d like to be able to programmatically recognize trends and trading ranges as they form. There are probably better ways of using such automated analysis than trying to mimic discretionary trading style.

the algorithms companies use are short tom algos that usually work of support and reistance levels

could you refer me to sources where I could read more on this?

I hope it goes well for you. Unfortunately I haven’t read any books on algo trading.

I don’t trust machines to trade for me. I don’t even own a dishwasher because I don’t trust them to clean the dishes. :joy:

Cheers

Blackduck

1 Like