Algorithmic and Mechanical Forex Strategies | OneStepRemoved

Working 8 days a week

by 22 Comments

Reaching an all-time high in my equity curve means it’s time to buckle down and keep improving. My Dominari strategy has done very well over the past 7 months and especially this and last month.

Dominari Equity Curve December 6, 2016

Is the party going to continue?

I certainly expect so. Drawdowns are inevitable, but that’s part of trading. Short-term performance is exciting, but my ambitious goal is to turn my starting balance of €8,000 into €50,000 within the next 3 years. As of this writing, I’m at €9,323.

You’re probably wondering how a 16% profit leads me to extrapolate an annual return of nearly 100%. The answer is that I dramatically changed my leverage at the end of September… just in time for that ugly drawdown. If I was trading on my current leverage, the current live return would be around 40% (i.e., right on track to hit my goal).

What really counts is what I’ve really done. So far, I’m up €1,323 with another €40,677 to go by December 6, 2019.

The research for Dominari is effectively finished. It’s been slightly more than a year since I began researching the strategy. Although minor variations of Dominari popped up or came from traders copying my signals, none of them improved the long term outcomes.

One version that improved the risk profile was to trade with limit orders. The original blog post mentioned limit orders, but the variation placed them considerably further from the current market than what I tried previously. I also lacked a system for choosing settings appropriate to every pair, which I’ve more than likely resolved. The problem is that I have a million things on my to-do list and only 8 hours a day. You’ll see some of my top projects when you scroll down.

Pilum: The latest and greatest

Pilum is a strategy based on a statistical process that identifies momentum. One of the scary elements about most quantitative strategies is that most of them are mean-reverting. They buy when the price drops and sell when the price rises. The approach is favorable (i.e., profitable) in the long run, but it takes some psychological fortitude to trade.

Pilum is a major advancement because now I’ll have a strategy that should profit exactly when Dominari is most vulnerable to a drawdown.

dominari trade outcome histogramThe new strategy uses limit orders to enter the market. Something like 90% of these orders never execute. But when they do execute, I win 75% of the time. Additionally, my profile of winners to losers is very comfortable.

Most traders understand the ideas even if the statistical jargon is unfamiliar. Pilum’s biggest winner is larger than its biggest loss. The average winner is bigger than the average loser. And, it wins 77% of the time.

Pilum trade outcome histogram

So far, I’ve done a sort of piecemeal backtest using R. When I finish the Quantilator (see below), I’ll redo the backtest in a fully fledged trading platform. More than likely, I’ll use QuantConnect to test the strategy level approach.

Trading platforms drive me crazy! The biggest problem that I have as a trader is continuously reallocating capital across my portfolio. MetaTrader, NinjaTrader and the likes implicitly assume that I want to trade some percentage of my account balance on every trade. Either that, or that I trade fixed lots.

Trading that way is extremely inefficient. I’m trying to trade 40+ currencies, so I need to be able to decide which ones need the money for trading and which ones don’t have signals. Then, among the ones that do have signals, I need to dish out their allocations proportionately. The allocations aren’t the same for each instrument. If you know of any good FX platforms that meet this requirement, then let me know in the comments section.

Testing Pilum on its own is important. More important than the performance of Pilum is how that performance interacts with Dominari. That means taking the daily equity values of each currency. Does Dominari lose when Pilum wins and vice versa? Should I allocate capital 50-50 between the strategies or does one strategy deserve the lion’s share of the portfolio? Is one strategy so good that it should get all of the money?

The main concern with portfolio allocation is how it relates to leverage. Dominari historically make 96% annual returns, inclusive of trading costs. But, it’s also trading with leverage of roughly 19:1. It’s possible for markets to rip over stops and create significant losses.

USDCHF lost 40% of its value in one hour in January 2015. Say that the scenario was even more extreme and that nobody could place a trade during that time at any price. That 40% move is multiplied by the 19x leverage used. That’s a theoretical 800% loss; more than the money in the account.

Everyone loves leverage because they think of profits. Leveraging losses is a lot less exciting.

So, if you could earn 96% annual returns and only use 5:1 leverage, that is exponentially superior to earning 96% on 19:1 leverage. I need to compare the returns of Pilum to Dominari per unit of risk. That allows me to do cool things like

  • Minimize the negative variance of the returns
  • Increase absolute return
  • Dynamically increase/decrease strategy allocations if I find patterns in their interactions

It’s a lower tech way of averaging strategies, like the litte guy’s version of what Numerai is doing… except that I have to create all of the strategies myself.

Quantilator

I spent the last few months sending surveys to segments of my subscribers asking how I can better serve you. Articles about indicators are overwhelmingly my most popular content. The trouble with that content is that I can’t complete the research fast enough to keep up.

The most valuable thing I’ve learned from the developing algorithms for the past 11 years is my process of deciding whether or not an indicator offers predictive value.

Moving Scale

Say that you’re interested in gaps: do gaps predict future returns? What I normally do is code a gap indicator in R, implement my analysis methodology and give a verdict. My methodology is kind of like a system for building systems. Using my approach usually takes an hour for every new idea that comes along.

An hour is pretty short. An hour is REALLY short compared to when I didn’t have a research methodology. I used to waste months on junk strategies.

With Quantilator, I’ll be able to analyze anything in under 5 minutes. I’ll only have to follow 3 steps:

  1. Run a script in MT4 to export price data and indicator data
  2. Upload the exported data to Quantilator
  3. Analyze the results

This tool will be 100% free. I’m planning to go through the most popular indicators in MetaTrader to analyze whether or not they offer an edge. I’m building a library of small edges that can be combined into powerful strategies like Dominari and Pilum. And, in the spirit of open source, I plan to make that library available to you for free.

I’m writing this tool in Django, which is a Python framework for displaying web content. The initial version will do the analysis. I’m hoping to use this as an education tool. A bit of momentum can justify make it a fully fledged library with sample data, indicators and training videos and more.

Quantilator’s name comes from a key concept in my system analysis methodology; I break data into quantiles. These quantiles calculate average market returns for a given period of time. The quant in Quantilator refers to quantiles, but I really like the implied double entendre of making you a quant. After all, that is what I’m doing for you!

Update: The Quantilator is now freely available at http://q.onestepremoved.com/

How badly do I want in?

by 10 Comments

You absolutely must check your trading system’s performance on a regular basis. You’re going to miss most of the problems from watching your equity curve alone.

That almost happened to me a few weeks ago. When I observed my account, I noticed that the real results had dramatically underperformed the hypothetical results. A quick review showed me that I only took 271 trades over the prior week, whereas my backtest expected to find 360.

I was only trading 75% of the setups! What could explain the missing trades?

Finding the flaw

One feature that I wrote into the MetaTrader version of the Dominari was a maximum spread feature. I’m paying commissions, so the idea of the rare but possible scenario of paying a 10 pip spread to enter a trade seemed intolerable. I added a maximum spread feature to prevent getting ripped off.

I also didn’t put much thought into what happens if the spread is too wide. My initial instinct was to put the EA into hibernation for a few seconds. It would then wake up and check the spread. If the spread narrowed enough, it would send a market order. But in my haste to start trading, I forgot to also require that the price be near my original requested price. That design would have allowed the market to drift up 10 pips and then, if the spread narrowed, dramatically overpay to get in the trade.

The new method for capping the spread paid uses limit orders if the spread is too wide. The advantage to this method is that it solves two simultaneous problems. The first one is easy to understand. A limit order has a limited price. It’s not possible for the price drift described in the above paragraph to occur. I either get the price I want or the market moves without me and I miss the trade.

Equity curve since I made the execution changes on March 16.

Equity curve since I made the execution changes on March 16.

The second advantage to using limit orders on entry is the fact that a limit order rests on the broker’s server. The hibernating method could potentially miss fractions of a second where the spread temporarily narrows to an acceptable price. Limit orders catch all price quotes, improving my theoretical likelihood of a fill.

Reality proved the theory after a week of trading. Instead of taking 75% of all possible signals, I’m now taking 87.5% of signals. That’s a result of the new limit method and my willingness to pay a wider spread to enter a trade.

More improvement

The question at the top of my mind was, “Should I be willing to pay even more to enter these trades?” Like a good quant, I immediately decided to calculate the question instead of haphazardly guessing.

I wrote a script in MetaTrader to search for every limit order in my account which was cancelled. I then looked at what the hypothetical performance of those trades would have been if I had simply paid the exorbitant spread.

It turns out that I should be willing to pay a lot more money to enter these trades.

There have been 50 cancelled limit orders within the past week, 44 of which were theoretically profitable. The average theoretical profit per trade was €1.28 compared to €0.33 for all executed trades. That’s a massive 287% difference in profitability!

The other shocker was the percent accuracy. 44 out of 50 implies an accuracy of 88%, compared to 64% accuracy on executed trades. 50 signals isn’t a lot. Am I getting too excited about missed profits or is that bad luck?

Basic statistics gives an answer with a high degree of precision. If the real accuracy is 64%, then you would expect to see 50 * 0.64 = 32 winning trades in a random sampling. My observed, theoretical accuracy with these limit orders was 44 orders out of 50, which is 88% accurate.

It turns out that I should be willing to pay a lot more money to enter these trades.

The standard deviation for 64% accuracy on 50 orders is 0.48, which we can then use to calculate the standard error. The standard error on 50 orders is sqrt(50) * 0.48 = 3.42 orders.

And finally, the standard error gives us enough information to compute the z-score. The z-score is the observed values-expected values/standard error, which is (44-32) / 3.42 = 3.5. A z-score of 3.5 has a probability of 0.000233 occurring due to random chance, or about 1 in 4,299 tests.

Conclusion: The statistics say with high confidence that my non-executed orders are substantially more accurate than my executed orders.

With the orders being both more accurate and having a higher per trade value, I increased the maximum spread that I’m willing to pay by 53%. While that sounds oddly precise, the per trade value might be substantially overestimated. I ball parked a guess that paying 40% in trade costs for a high quality trade seems reasonable. That number may have to go higher in order for me to measure the details.

Ideas for exploration

The amazing extrapolation from the live order analysis is that the spread seems to predict my likelihood of success. Wider spreads make me more likely to succeed and with a better risk:reward ratio. My project over the next few days will be to start logging my spreads at signal generation time to evaluate whether the spread predicts the profitability of my signals.

Oddly enough, there might even be a paradoxical outcome where narrow spreads predict my failure. More on that when I have enough data to answer the question.