A conversation on testing, analytical thinking and randomness with a fellow trader over a cup of coffee on Wednesday afternoon.
Yesterday afternoon I had a great time catching up with a fellow trader I had met on the VTP program in 2016. Thanks to being back in the previous trading office, it’s only a couple of hundred yards from the Goswell Road coffee joint which, in my humble opinion, serves awesome coffee, good music and has several comfortable couches!
This blog post recounts some points of our conversation.
Advantages of trading on higher timeframes
About a month ago, Ted [not his real name!] switched to trading on the 4-hour and Daily timeframes, as opposed to the lower timeframes that he was using previously. He explained that this has decreased the amount of stress and pressure in his trading, lowered the amount of time it takes to check for setups, has allowed him to manage his positions better and effectively removed the possibility of emotionally-induced poor trading behavior. We agreed that trading on the higher timeframes also reduces the impact of transaction costs. Good food for thought for anyone considering switching timeframes.
Next I subjected Ted to listening to some of my recent learnings.
Breaking down my thought process & coding it
Learning to code in EasyLanguage on the TradeStation platform has allowed me to rapidly test ideas, be they indicators or trading strategies. What would have taken me weeks or months to figure out previously, via testing in ForexTester, on my own charts or in an Excel spreadsheet, I am now often able to complete in days.
I can get to these insights fairly quickly and back it up with large sample sizes – I can quickly state how often a particular situation actually occurs and how often it results in subsequent price action. It’s a form of statistical or probability analysis. I am starting to build my own methods for collecting statistics on how the market behaves.
Additionally, am quickly building a reservoir of code that I can quickly insert into new trading strategies – position sizing, order generation and trailing stops to name a few.
Attempting to code my trading thoughts is forcing me to carefully articulate what it is that I am looking at on the price chart in forming my bias. And next, how can I formulate these impressions into an algorithm? This process is helpful and insightful, though sometimes painstakingly difficult.
Incidentally, much of this testing – say looking at engulfing candles or a pinbars as a simple example – is not promising any value (or ‘edge’). Alas, many of the ideas or concepts are not helping me to predict subsequent price behaviour. It’s as if I was flipping a coin.
What’s the deal with that? Is it a case that I need to get better at assessing the context? By attempting to judge the trend, volatility levels, support & resistance levels? In other words, needing to add more parameters to the code?
Or does it require exercising good discretionary judgement on concepts that can’t be coded? The more I program, the more random the market seems to me. Looking at a chart and pinpointing a certain combination of price action initially tricks my brain into creating a totally logical explanation for what happened. But once I successfully articulate that situation into code, I realize that the setup works half the time and not the other half – basically it’s just random behavior and the price action cannot be used to predict price movement. Irrespective of how I set the stops and targets.
The above thoughts could be good insights, or they could simply highlight my inability to analyse the markets in an effective manner. Maybe others can, but I can’t. Which one of these two is it?
My natural (somewhat pessimistic inclination) says it must clearly be the latter as other traders have the ability to produce outstanding equity curves with just a few lines of code! Just check out the twitter feeds of Dave Bergstrom, Gustav or PipBandit. How on earth do these guys do it?
Tomasini & Jaekle on automated trading strategies
Fred [not his real name!], a systematic trader at the hedge fund I left two weeks ago, said that this book contained pretty much most of what I need to be aware of in becoming a systematic trader. 200 pager textbook – Easy to buy on Amazon (new or used) £20-£25. Fred strongly emphasized that the ability to verify that an apparent edge is real, is as important as finding the edge in the first place. This book, he says, will be a great help for me to build this process.
I’ve read through 60% of it thus far, yielded some good practical tips and started applying it. The following diagram simplifies and summarizes a good chunk of it lot of it.
Still to read and study are the topics of walk-forward analysis, Monte Carlo simulations, position sizing and portfolio trading.
Fred also said that it’s better to be in a position of having a lot of market experience with limited coding skills than the other way around. The latter is much more easily learned than the former. The 5-page conclusion at page 193 in the Tomasini text discusses thesame idea – 5 pages with outstanding thoughts and ideas.
My search for non-randomness continues.
Comments and discussion welcome.