Algorithmic Blues – Past and Present, Part 1

I’ve been through many, many systems and algorithms in my trading career so far. Some I’ve chopped too early, some I’ve made some profit on but not understood either the risk or plan itself and have resulted in some swings so intense that I’ve ditched them in a confused state, and some just straight haven’t worked.

Suffice to say, I’ve always felt on the edge of glory, but something has stopped me from crossing the line for whatever reason.

At this point, I’ve gone so far down the road that I’m not sure what failed where historically, but I did want to highlight some of my past algorithms which have come close but no cigar, and my interpretation of why they failed.

Let’s start with the rather unhelpfully named “Age of Destruction”. Yeah, I have naming convention issues. Live with it.

Indicators Used:

  • NihilistRSI (Specially developed indicator)
  • JB_Center_of_Gravity (a non-repaint version of the Center of Gravity indicator)
  • MTF ATR (1D over 10 periods, used for setting P/L targets)

This was created around about March-April 2019 time, for use on a 1H chart. I was still learning NNFX, but obviously hadn’t taken enough information on board at this time, hence my use of RSI (which I agree today, isn’t particularly helpful). Simple enough approach with minimal indicators.

NihilistRSI is a combination of two indicators – RSI (durr) and Nihilist Ultra Trend. Nihilist Ultra Trend is a propitiatory indicator and I have absolutely no information on how it works, but I saw a link between how RSI reversed in certain territory when Nihilist Ultra Trend changed from red to green and vice versa. It seems to give fairly keen reversal signals, and when accompanied with these signals appearing outside of the CoG bands, these elements together made up a reversion to the mean strategy that seemed to be pretty solid.

In trading, it didn’t work. At all. Why?

Well, here’s your super dumbass moment from yours truly – Nihilist Ultra Trend repaints, quite significantly far back too. The very nice looking arrows on this chart would never appear in live trading, and with the myriad of colours on the Nihilist Ultra Trend indicator it’s quite hard to spot this repainting.

I would’ve got away with it if it wasn’t for those meddling arrows

I went all in on backtesting this particular indicator and seeing the number of trade opportunities that would come out of it. It was only after I’d spent quite a bit of effort on crafting the combination that I went more into detail to investigate the problem.

From this, I learned the following:

  • Test, retest, triple test that the indicator you’re intending to use doesn’t repaint. If it does repaint but you still think it’s going to be useful, be sure this is factored into your system and additionally, be aware that backtesting your system is going to be largely impossible.
  • Have a concept of how your main entry indicator works even if it’s a remote one. This is just stupid. Even if you don’t know the ins and outs you need the know the loose concept.
  • Combining indicators is actually really powerful providing they are the right indicators. RSI on its own, as NNFX has been very clear, is not a great indicator for Forex. However, if you are using it to measure the strength of another signal, it can turn into a very powerful separate confirmation of a single indicator. Don’t be afraid to throw these things together as an experiment, and don’t think you always have to look at them as oversold/overbought indicators. It can be as simple as “my confirmation indicator is giving me a signal, and also RSI has reversed past a certain level on the same candle. Might be worth a look”
  • Center of Gravity is a great indicator for both trend following and mean reversion strategies, but it does not fit as any kind of primary confirmation tool. Use it as your last defining tool.

Another variant of this I have in my old template list is “BreadOfJustice”. Must’ve had toast that day. This was exactly the same as the above but with an added TEMA baseline with a 34 period in order to try and fight the repainting issue after I’d discovered it. It relied on a NihilistRSI signal popping outside of the CoG lines followed by a change in the TEMA direction. The two problems were, the repainting of NihilistRSI can stretch back quite a significant distance, and the change in the TEMA would always pop too late to be able to get onto any mean reversion move. It was quickly binned.

The last template I’ll look at today is a variant of Oma Ally’s BBMA strategy. If you haven’t come across BBMA before, it’s worth looking at as a study. Conceptually it’s a solid system that didn’t particularly work hand in hand with my own approach to trading.

BBMA stands for Bollinger Bands Moving Averages. At its core, the approach is to observe when moving averages pop out of a Bollinger Band, and other conditions apply (such as reversal candles) you get into the market. I made several modifications to the idea to see if I could get something together I found more tradable, but obviously the key issue here is trying to trade a reversal in Forex is just not a smart idea.

The closest I got was “BBMA-Xtreme-Confirm”. This would’ve been during Summer 2019.

Indicators used:

  • Bollinger Bands
  • X-Treme_MAYOR PAIR_v2.4.1_nrp
  • X-Treme (these two do combine to give a slightly clearer picture than using one of their own, but they are essentially modified Zig Zags)
  • Powered_Trend_Signal_Arrow_Alert – 2
  • Powered_Trend_Signal_Histo_Alert (these two have different HL settings. I’m not sure why)

The “Extreme” part of the system remained the same – you’d need the X-Treme indicator to show an initial signal outside of the Bollinger Bands. At that point, you’re playing reversal signals. The Trend histo and the Trend arrow then need to match up to take an entry. Very simple to apply, very simple to use.

That green line represents the perfect signal. Only problem is it wouldn’t have been backed up with confirmation for 10-20 bars after it appeared.

The one massive problem is Powered Trend Signal repaints. Badly, too. There’s a bit of lag on the X-Treme indicators too. That’s the nature of Zig Zag based indicators. That said, having looked at it again now for this post, there are big chunks I would like to incorporate into my current algorithm. I wonder if taking it into the MTF domain would improve its consistency and ability to pull out great trades. It’s certainly a concept I’d like to revisit, and it’s something going to add to my worklist!

That’ll wrap this post up. I think I’ll try doing this on a semi-regular basis. It’s been a great learning experience for me, and hopefully for anyone reading this struggling for ideas it’s given you some base to start from too.