Trevor Neil has been a professional trader for more than 40 years. He started out at 18 working the floor of the London Coffee Exchange for Merrill Lynch where he was taught to carry out his technical anlaysis using graph paper, a pencil, a ruler and a compass. He later became an institutional broker, during which time the invention of the personal computer transformed charting, and technical analysis at large. “Computers enabled the indicators that are so popular today to emerge – the RSI [Relative Strength Index], MACD [Moving Average Convergence Divergence], on balance volume – they’re not complicated, but they are tricky if you don’t have a computer,” he says.
Neil spent the 1990s as a systematic fund manager before serving as Bloomberg’s global head of technical analysis from 1999-2003. He then co-founded an active long short hedge fund in South Africa, which quickly became the top performing in the country. In 2006 Neil returned to London to launch his own smaller fund, which he continues to manage today.
“We were in South Africa to keep the institutional investors happy and make sure we had low deviation or returns, which I found very restrictive,” he says. “But this fund is to make money. It’s mostly my own money and I’m able to accept larger draw-downs than would be feasible for a large hedge fund,” he adds. Here are his tricks of the trade, a definitve guide to the merits of technical analysis.
I’m a 100% systematic trader. I set up computer programs that make all of my trading decisions by analysing stocks using rule-based technical analysis. The big problem for everyone is oneself; it’s fear, greed and maintaining discipline, all those childish things. The secret of success is to become like a machine and get rid of all that. That’s a hard thing to do; learning technical analysis is relatively easy. The only way I can do that is to get rid of myself completely and let the computer programs I create make all of the decisions. To assert that I’ve tested this system and I know it works. All I have to do is let it run. The only reason I execute the final trades myself, rather than let the computer do it, is so I have something to do.
“The big problem for everyone is oneself; it’s fear, greed and maintaining discipline, all those childish things”
I quite often strongly disagree with what I’m doing. But I still do it. If you ask me for my opinion on the pound, I’m bearish. But my money happens to be long. I don’t let one interfere with the other; I know from experience that this is the best thing to do, that the machine is better than me.
The machine still needs a human to program the system first. The company I was working at 25 years ago wasted a lot of time and money on machine learning. You can do the things we were working on a lot faster now, but I still believe that, like back then, it doesn’t work. The markets are far too complicated for a computer to master without being told how to react; a computer can’t even completely drive a car yet because it can’t account for the unpredictability of people. A computer can just about beat the best person in chess, but chess is a game with a certain amount of squares, pieces and moves and it’s against just one person. The market is thousands and thousands of people, all of whom do unpredictable things – it’s a fool’s errand to think a computer can judge and respond to all these people.
“I quite often strongly disagree with what I’m doing”
Technical analysis works because everything is in the price. Supply and demand, whether the sentiment is bullish or bearish, plus the trading behaviour of its participants. The price is all of these elements weighed up accurately and given a value. I follow the fundamentals, but I never make a decision on them as I can’t judge whether the fundamentals are important or not, it’s the market that will decide.
My fund is divided into two portfolios that make money in different environments. One follows longer-term trends and the other uses pattern recognition to ride ups or down turns The setup uses the modern portfolio theory; the idea that you strive to have two trading systems that are both ultimately profitable, but zig-zag up as opposites to each other, cancelling out the draw downs of one another and reducing volatility of returns.
“The market is thousands and thousands of people, all of whom do unpredictable things – it’s a fool’s errand to think a computer can judge and respond to all these people”
My pattern recognition system requires very specific, high probability conditions to enter a trade. In order to find enough of these opportunities, I scan the equities listed in the STOXX Europe 600 to find patterns, which I then trade using CFDs to get leverage, and go short.
I buy or short during a weak, low-volume retracement in a stock that is otherwise experiencing a strong upwards or downwards trend. To find buy opportunities the first scan identifies the stocks that are showing signs of an emerging uptrend – those that have both a positive weekly MACD and an Average Directional Index [ADX] above 35. Out of these stocks, the program then identifies those that are pulling back due to experiencing a weak sell-off. This is assessed by identifying those that have for at least three days had a RSI below 50, and which have a three-day average volume lower than the five-day average volume. This tells me that the stock is drifting, rather than being completely oversold and that the volume is declining; it’s losing power on the downside.
The final decision to buy is made by assigning the security with a risk/reward score; I buy those with the best score. To come up with this score I first calculate the difference between the point that the stock will be bought at – just above the highest points of the last two bars, and the point that the stop will be set – just below the five-bar low. This gives me the risk number. The computer then uses Fibonacci to set a minimum price target for the trade – the lowest level the stock is likely to go if the trade is successful. This is my reward number. The risk is then divided into the reward to provide an attractiveness score. I will not take any trade that is less than 3/1. The risk/reward score also sets the position sizing of a trade.
I’ll never be more than 20 long and 20 short in this system. The trades are highly correlated – they all make money together or they all lose together – so there’s no point having more positions, which would generally have a less attractive risk/reward ratio than the ones I’m in.
“I’ll never be more than 20 long and 20 short in this system”
When a trade passes my minimum price objective, all I do is tighten the stop. To exit the trade I use a trailing five-bar low stop, until the trade breaks the price target, at which point I switch to a trailing three-bar low stop. I let the market run until it changes direction.
I’m looking for truisms, not to exploit mathematical anomalies or distortions. It’s a statistical fact of the market that if a security has a strong uptrend and a weak deep, it very, very often reasserts the trend and goes on to make a new high; that’s the nature of a bull market. In a short position, I look for exactly the opposite. I’m using very familiar technical analysis indicators, there’s nothing special about them individually, it’s how they are paired together which creates the effectiveness.
“I’m looking for truisms, not to exploit mathematical anomalies or distortions”
My second portfolio uses a trend following a technique similar to the Turtle method. It’s a diversified range breakout system using the futures market and cash forex in order to get as much non-correlation through the portfolio as possible and create a barrier against shocks. If there are trends in the markets, it makes money. It’s all about catching those big moves, building the position as the move develops and participating in it from beginning to end, up or down.
This means longer holding periods, with successful trades lasting 4-6 months, with the sell signal coming when the lowest low of a lower period emerges. It requires a fairly large amount of capital to operate and I find that many traders simply do not have the patience to follow this method, despite it being tried and tested. The style was first developed by Richard Dennis and William Eckhardt back in 1983.
Placing a trade or investment is a choice; you have many choices of where to put your money. How to make this decision is what I’ve been concerned with in helping to develop a new piece of technical analysis led by my colleague Julius de Kempenaer. It’s called Relative Rotation Graphs [RRG] and is a way of visualising a stock’s relative performance, as well as the momentum of the relative performance in one graph.
It solves the problem that if you wanted to, say, compare one FTSE 100 stock against everything else in the FTSE 100, it would require you to study thousands of charts. I use it not to find the best performers – the top-performing tend to have already peaked and are fairly stagnant – but those that are coming up under the radar at speed and will eventually become the best performing. We have licensed RRG to Bloomberg and Refinitiv's Eikon platforms. It is proving popular among institutional managers and traders alike.
“Placing a trade or investment is a choice; you have many choices of where to put your money”
One of the biggest mistakes I’ve made is getting too complicated. Years ago I thought it was an arms race of computer power. I’ve wasted time and money on things like machine learning and chaos theory. But it’s still the simple things that work best; I know a lot of technical analysis, yet the technical analysis that I use is simple – and that’s deliberate.