It’s been a full 12 months since I created a new live account and started publicly sharing the results of trading with the Icarus method.
Let’s take a look at how things have been coming along.
Here, we see that most of the gains were made in the first 7 months.
There was some under-performance in the last few months due to the market crash in late-Feb that led to some of the most volatile price swings I’ve ever had to trade through.
Thankfully, the account managed to make it to the end of 12 months with a positive return.
All in all, it achieved a net 22.76% gain with a 13.02% drawdown. The drawdown figure is larger than I would have liked, but given the market crash in February I suppose I should be glad the number isn’t twice as large.
Overall, I’m satisfied with the performance on this account, especially since it doesn’t take a lot of screen time to implement the Icarus method. Hopefully, price volatility will return to normal levels soon.
TF Strategy
Given how choppy the currency markets had been in May, there were virtually no trade-able opportunities with the TF strategy.
If this continues over the next few weeks, I’ll be shelving this strategy until the situation changes.
Algo Trading
Learning about algorithmic trading has broadened my view on how one might approach the markets.
In my usual (non algo) trading routine, I’m usually tuned in to the market sentiment. I spend time every week estimating the “mood” of the market by keeping up with the news and observing the ebb and flow of market prices.
With algo trading however, there’s no room for such subjective approximations. Everything works in black and white, which leaves out any appreciation of the “flavor” of the market.
The good thing about algo trading however, is that I can backtest multiple versions of a strategy over ten years of market price data, across 28 currency pairs, in less than a week (including the time it takes to code the algo).
What’s more, I can backtest combinations of different strategies simultaneously. This not only saves me a tremendous amount of time, it also allows me to approach the market in ways that I previously couldn’t have.
While algo trading isn’t great for detecting nuance (with my current understanding of it, at least), it opens up a number of trading paths I otherwise wouldn’t have access to.
Here are some of the backtest results of the strategies I’ve been working on:
Whether these results can be replicated in live trading, remains to be seen.
To that end, I’ve recently begun forward-testing a couple of algos on a demo account, and will be reporting on their results in future posts.
Is algo trading the future?
The more I think about it, the more it occurs to me that I should be dedicating more resources to algo trading.
It takes more time and effort to get started, but – assuming I can get the algos to be profitable – it’s a more scalable approach to trading with larger amounts of capital.
What’s your opinion of algo trading? Have you tried it? If not, would you like to?
Let me know your thoughts in the comments section below.
I installed four robots on my demo. One was given to me by a trader and the other three I found on the internet. The one sent to me was taking trades and was profitable but at some point it stopped taking trades so I removed it from the charts. Of the other three, one was just a money losing robot so I removed that one too. The other two have been incredibly profitable in the six months since they have been trading. One of them even adjusts the lotsize automatically based on the balance on your account.
Like I said this is a six month result from a demo. And I understand some brokers are in a hissy fit over robots and might want to cause them to malfunction.
I want to learn how to create a forex robot. There’s a strategy I came up with that I want to see if I can code into an algo.
Hi Okeke, from what I understand many brokers are fine with algos as long as they are not entering/exiting trades every few seconds. I might be wrong though.
Chris,
Can you share where can get the algo trading template?
Thanks,
Edwin
Hi Edwin, what do you mean by ‘algo trading template’?
Hi Chris,
I mean any program you used to back test. Is it you mean algo trading?
Thanks
Hi Edwin, I’m using python, and the strategy tester on MetaTrader 4.
Hi Chris, algo method I’m currently using includes MACD candles V3 / ASH strength histogram (entries and exits and just the red and green lines) / Rex oscillator – exit option) / and the ATR for TP & SL..either multiplying x1.5 or 2 pending on dollar values to keep it managable with the trading currency (ie. $AUD). Usually 2x lots (first lot with the ASH new direction and the V3 candle though in the ranging colour but strong, and the 2nd lot when the candle colour changes to trend). Also no TP on 2nd lot and follow the trade up or down with the SL. (As you can imagine, first week of June delivered a bumper crop!) Re May 2020 AUDJPY delivered 150 pips. Just 1 example. Trade the Dly to avoid most of the market noise – once on profit with the 2nd lot have traded through the bigger news ie cash rates, GDPs and the NFP.
Hi Mike, a nice result! What does the backtest of your algo look like?
Forgot to mention using a 50MA simple as a baseline on top of the MACD candles – re AUDJPY – counter trends can be positive when between 150 – 300 pips away from 50MA – as Aug 2019 demonstrates. Otherwise close to the baseline they tend to range or stall more often than not. As you would know some pairs backtest better than others..then there’s the EURJPY this year…chaos…and the EURCHF …it’s been below the the 50MA since Dec 2019…a good run of 200+ pips from late Dec to late Jan 2020….(but who trades then?). But as the backtest reveals, the EURCHF with the MACD candles and when red (sell) and under the 50ma, have returned roughly 20 to 30 pips a time ..May 18 2020 the exception as we’ve witnessed. There’s better pairs of course.
And thanks Lee for your great line – FTPESD….the acronym. (One of my mortal weaknesses). So timely. Also for exposing the 16+ tech indicator ‘no nonsense’ smorgasbord that had me in melt down mode. But I think I have plucked some knowledge from it all. If that’s the site you’re referring to. Regards, Mike
Hi Mike, I’d be careful exrapolating results from the ‘late Dec to early Jan’ period, as liquidity during that period is low.
Other than that, I’d say the best we can do is to try and make sense of the market tendency we’re trying to profit from, and to estimate if that tendency can persist over time.
I do a lot of algo trading having my own autotrader that I can program/set parameters at will. For me there are two key factors to consider.
1. What tick data are you going to use?
Your broker’s data, only about a year’s worth normally available unless you have had the account running for however many years. But at least it will be the type of (clunky) data you will trade on.
History Centre data, from Metaquotes which as it says, cannot be used to resolve any issues with your broker; ie it is different to what your broker has.
Dukascopy data, the most popular and supposedly the most “clean”, but likely to be quite different from the broker’s data you may be trading on; unless you trade with Dukascopy.
Each of these data sets will give you different results for optimisation and Back Tests (BT).
2. Over what period are you going to optimise and how much Out of Sample (OOS) data will you have for the BT? Showing a BT over the optimisation period used is not a BT. It is the data-fitted performance curve resulting from the data you used to optimise your strategy. It looks good (because you chose the best Pass), but is totally misleading as a guide to future performance. Unless you strike a lucky patch. Pure chance.
The notion that optimising over a long period must result in a more robust strategy seems logical and is appealing; but often not so (spoken from experience having done 19 year optimisations). You should aim for using market periods which have a similar “mood” to what you expect the next year or so to be.
3. Trap for new players. The Spread setting in Strategy Tester is a value to be entered in Points, not Pips (1 pip = 10 points). So enter at least 20 (ie 2 pips). Otherwise your BT will look great but your forward test will be a disaster.
Hi Andrew, all good points. Thanks for shaing your experience with us.
Hi Chris Lee,
I have not used algo, and I want to use it am interested!!
Await ur response
Hi Apata, thanks for the feedback. I’ve still relatively new to algo trading but as I delve deeper into it I’ll be sharing what I learn.