What I’ve learnt from trading

20170121 Flight

Over the past decade, whilst I’ve always been a quantitative strategist and developed many strategies primarily for research purposes, I’ve had quite a bit involvement in creating models which have also been traded with real cash. At Lehman Brothers, I was involved in the creation MarQCuS product, which had over 2bn USD under management. I created an automated trading model in Java, which traded liquid G10 FX markets and it successfully traded for a year, under the trading book of the head of FX, up until September 15th 2008… Somewhat ironically, it’s best performance (admittedly this was simulated!) was after the Lehman Brothers bankruptcy. Rolling on another few years at Nomura, I developed a number of systematic trading strategies, both intraday and daily, with the spot trading desk, which were worked pretty well. Since I resigned from Nomura over 3 years ago, I’ve also traded my own cash, with a range of intraday models, although recently I’ve cut my risk, given I have less time to devote to trading. Happily, the vast majority of years, my trading strategies have been profitable. So what has the experience of seeing my models trade real money taught me? Whilst none of this constitutes investment advice, I thought it would be useful for me to write. Most importantly, there’s a huge amount I am still yet to learn about trading!


Research does not teach you how to manage risk. Research is necessary to build a trading strategy. It tells you what the historical returns of a trading strategy is and in particular how they are distributed. It can help to tell you how robust your trading strategy is, with respect to its parameters. Research can help to validate a hypothesis. However, as I’ve written often before, you simply cannot backtest pain and how you’d emotionally cope with that. Those periods when a trading strategy I’ve developed has been down several hundred thousand dollars or even over a million in a day, are painful. It simply isn’t something you can understand when you look at a backtested plot of returns. Sure, it just looks like a small dip on the chart, but if you were there trading that, it would have been very painful. Whilst I’m a quant, part mathematician and coder, ultimately, I recognise there are still limits to the quantitative approach to markets, and whilst I find systematic trading can help give you discipline, it cannot totally remove the emotion from markets. Ultimately, you still need to make some decisions, like how much capital should I be trading? You cannot “machine learn” your way to becoming a good trader…!


Trading is not for everyone. It takes time to get comfortable with the idea of managing risk, and I’ve enjoyed it, but I recognise that it isn’t something that everyone will enjoy. Some folks can read the market very well. There are some strategists who make good judgement calls about turns in the market. However, this is a different skill to managing risk. If you have no discipline, you might repeatedly make the right call, but fail to profit from it, because you can’t tolerate mark-to-market losses on your trade. In this context, it’s more profitable for a strategist to work with a trader. The strategist can provide input, and the trader can manage the risk. I would argue that when we delve into the world of retail trading, the evidence shows trading is most definitely not for most people. Most research I’ve seen shows that a large proportion of retail FX traders lose money (something like 90%), perhaps not surprising given the unbelievable amounts of leverage which are given to retail traders. Indeed, when I’ve looked at retail FX positioning data, it is usually has an inverse relationship with spot, eg. when retail traders tend to be buyers, that’s often the time when spot actually falls. Most of these folks would likely have been better off just investing in a passive tracker (without leverage) and holding it. If you are a retail trader, and you do want to trade, do your research to take time to understand the market and most importantly don’t ever be tempted to use the crazy levels of leverage a broker might tempt you with. In fact, that’s pretty much a universal rule, excessive leverage is a the heart of most trader blowups.


Not all trading approaches are suitable for each trader. When I was at Nomura and Lehman Brothers, the capital being used was not my own. When I’m trading for my own account, obviously, it is my own capital. Strategies that I might have been able to run with deep pockets, I simply can’t run with my own account. You need to figure out how much you are able to stomach losing during drawdown. You can then scale your notional accordingly to match what is a tolerable drawdown for your strategy. I try to avoid at it the other way around: how much money can make? This is just an invitation to leverage up too much. A certain trading strategy which I might run, might not be suitable for another trading, because they have a different risk appetite.


Understand your P&L and your transaction costs. As a systematic trader, my models have made the decisions for me and the whole idea is that you do not constantly change a systematic strategy. However, it is still important to understand the P&L, and this takes time. For example, when I was Nomura and the traders were running models I co-developed with them, on a daily basis, I would analyse the P&L. Which trade had made the most money? How did the theoretical model P&L compare with actual executions? Were certain strategies underperforming versus history (or outperforming)? Had there been a special market event which had impacted the P&L? Was liquidity particularly bad in a certain currency pair? How much was I paying to execute? Some of this falls under the umbrella of transaction cost analysis, which regulators are very keen on (in particular MiFID II). However, I’d argue, even without regulation, it is crucial to analyse your transaction costs. Particularly for intraday trading, you can give away a lot of P&L with poor execution!


You learn from losing more than from winning. Whilst my strategies have generally been profitable, I’ve learnt by far the most during those periods when they’ve lost money, whether it was within a bank or in my own personal trading. Whilst losing money is never fun, it is important to learn from these experiences. They’ve taught me how to improve my trading strategies. At the same time, just because you have a period of losses from a trading strategy, it doesn’t mean you should always turn it off, particularly if the losses are actually within the scope of historical losses. The situation is somewhat different, when your strategy starts to lose money, which bears no relationship with historical returns. Does this indicate the trading strategy has “broken”?


It’s all about the capital. This might seem like an obvious point, but in order to make money you need to have a reasonable amount of capital to start with! In my own personal trading, I’ve managed to generate profitable returns. However, the capital I’m allocating is relatively small. This in itself is going to limit the actual amount of money I can make in practice, even if returns are good. At the same time, I’ve often withdrawn cash from P&L I’ve generated, particularly when I started to help pay for my costs (eg. Bloomberg terminal), which has reduced the amount of capital available to trade. If your costs are much smaller proportion of your P&L, you can reinvest more of the P&L back into the strategy. Over the years the cumulative impact can make a big difference. Often, it can often end up being more profitable to develop a trading strategy and partner with someone who has the sufficient capital to trade it, rather than purely trading it myself.


There is no secret sauce. Sorry, there is no secret sauce to trading, whatever anyone says. There is, however, hard work and effort, to find and research new trading strategies! Very often the “best” trading strategies, which might exhibit high Sharpe ratios, can have their own issues. They might try to use certain understanding of certain elements of market microstructure or a particular type of behaviour in the market, which you might have identified. However, by their nature these type of behaviours can be more fleeting. Often these can be strategies of a higher frequency. They might also exhibit more fragility and also be very low capacity. The only way to run a large amount of capital in these type of strategies, is to develop a lot of them, but this of course is very time consuming. There’s no free lunch folks!


Trading can be fun. It can also be very painful. If you do want to start trading, I’d recommend spending the time to do your research thoroughly and also to do a bit of paper trading too, before making a final decision. Trading is never “easy”, but it is possible to do well with a lot of hard work and time.