When the opening rule switches to the opposite direction, we close o ur position.
We then immediately open a new position in the new tren d direction.
This type of trading is sometimes called stop and reverse .
However, I prefer the term continuous trading , because we always have a tra de in place.
What if you have added multiple opening rules, as I advised in chapter eight? The procedure for continuous trading with multiple rules is equally simple. Take a weighted average of your rules, using the method described in chapter eight, to determine which direction your initial position should be. You then continue to monitor all of your rules and close the trade when the average has changed its mind. Then it’s time to open a new trade in the opposit e direction.
Advantages of continuous trading
Figure 24: effect of dropping the stop loss from the Starter System
I tested the effect of dropping the stop loss from the Starter System trading rule, plus two other moving average crossovers that I use in my own system. The results are shown in figure 24, using the usual box and whiskers plots that show how statistically significant the results are. There are three pairs of boxes, each covering a particular trading rule. The first box in each pair shows the Sharpe ratio (SR), with error bars, for a particular trading rule using the stop loss. The second box (suffixed NS – no stop) shows the same data with the stop l oss dropped.
You can see that in all cases the second box is at least as good as, or better than, the first. Admittedly, none of the differences are statistically significant, as the boxes all overlap. But taken together there is good evidence that dropping the stop loss improves the performance of the Starter System by around 13% (from a SR of 0.24 to SR 0.27).
This is a nice improvement, although it is not as high as the gains we saw from diversifying across instruments and opening rules in part three. But there is another advantage of dropping the stop loss. It makes designing your trading system easier. If you want to change to a different set of trading rules, you do
not need to worry about recalibrating your stop loss so that it matches up.
Finally, do not use these higher SR expectations to increase your risk target as you would do if you were adding instruments or trading rules. The improvement is not statistically significant, so it is better to be conservative and not increase your leverage to match a higher risk target.
When you should keep a stop loss
There are three reasons why dropping the stop loss from the Starter System might not make sense. Firstly, the stop loss is a comfort blanket , which makes you feel like you are more in control. If getting rid of stop losses worries you, then you should keep them. The slim benefits of removing the stop losses are not worth the loss in sleep. ¹²³ Secondly, continuous trading might involve more work , especially if you have multiple trading rules. However, if you have set up your trading system on a spreadsheet this should not be a se rious issue.
Finally, for some styles of trading, dropping the stop loss would be dangerous . Suppose you’re not using a trend following rule, but instead using a trading rule which doesn’t get out of positions when the price reverses. Then you’d be stuck in a loss-making position as it got worse and worse. Here it would be better to use a stop loss. I would advise keeping the stop loss if significantly more than 50% of your opening rule weighting is in non-trend following rules (like the carry rule).
In my own trading I take a mix-and-match approach to stop losses.
There are three components to my portfolio. Firstly, I have a fully automated futures trading system with no stop loss. This is mostly a trend following system, although it does include a 40%
weighting to carry. So, the actual trading rules do the duty of closing loss-making positions. Additionally, as it is automated there is no extra work involved in recalculating the forecasts fo r each rule.
The second part of my portfolio trades UK equities, picking cheap shares based on valuation metrics like dividend yields. Here I use an explicit stop loss (this sub-portfolio is where I bought those Royal Mail shares). Because I’m buying cheap shares my trading rule won’t close positions when prices fall: it will usually want to buy more. Removing the stop loss would be dangerous. Also, I am running this portfolio manually, and I don’t have the time to do the extra calculations required for continuously trading without a stop loss.
Finally, I have a buy and hold sub-portfolio of exchange traded funds. Because I’m investing, and not trading, these instruments don’t need a sto p loss rule.
Practical trading without a stop loss
Here is the updated trading plan for trading without a stop loss: How would I have traded differently with the examples from chapter six, if I hadn’t been using a stop loss?
Let’s begin with corn.
Corn
You might remember that I originally opened a short, and I closed this trade on 10 October. A new long trade was then opened on 26
October. Now look at figure 25. I would have kept the original trade open longer, until 26 October when the opening rule changed its mind (though it comes close to being closed in mid-August).
Figure 25: Corn trade would have been closed slightly later with no stop loss
Comparing the original trade log (table 56) to the new one (table 57), you can see there is an additional 2.2 point loss when the trade is closed on 26 October. We are trading two contracts at $50 a point, so this costs:
2.2 × $50 × 2 = $220
The original profit was $1,165; using a stop loss would reduce it sligh tly to $945.
Table 56: Corn future trade log with stop loss
‘Comm’: commissions. Excludes spread cost, since the price shown reflects the effect of paying the spread. P&L is profit or loss in price points for closing trades only.
Table 57: Corn future trade log without stop loss
‘Comm’: commissions. Excludes spread cost, since the price shown reflects the effect of paying the spread. P&L is profit or loss in price points for closing trades only.
No w for gold.
Gold
In chapter six this particular trade was kept open until I’d finished in early November, for a 53.1 point profit. Figure 26
shows that using the opening rule rather than a stop loss would have closed the position in October, and I would then have gone long. Because volatility has risen the second trade is smaller: a buy at £1 a point.
Figure 26: Gold trade would have been closed in mid-October
Table 58: Gold trade log with stop loss There are no commissions. Excludes spread cost, since the price shown reflects the effect of paying the spread. P&L is profit or loss in price points for closing trades only.
Table 59: Gold trade log without stop loss There are no commissions. Excludes spread cost, since the price shown reflects the effect of paying the spread. P&L is profit or loss in price points for closing trades only.
I made a total of 39.9 + 9.8 = 49.7 points at £1.03 a point, and then lost a further 2.8 points at £1 point. With no commissions or holding costs th e profit is:
(£1.03 × 49.7) + (£1 × – 2.8) = £48.39
For comparison, in chapter six the original profit using a stop loss was £54.69.
The next trade we’ll conside r is AUDUSD.
AUDUSD
In chapter six the initial short trade was kept open for the entire period. Figure 27 shows that without a stop loss the opening rule does exactly the same thin g as before.
Figure 27: AUDUSD stays short
Now let’s look at Euro Stoxx.
Euro Stoxx 50
This was a pretty disastrous trade in chapter six; I went long and was then stopped out at a loss on 10 October. My short then lost money when the market rebounded in early November. Figure 28
shows a completely diff erent story.
Figure 28: Euro Stoxx long is short lived, and eventually goes short quite early
The initial long is short lived and gets closed after onl
y a few weeks. Then I go short for a month or so, again on the wrong side of the market and this trade is closed in mid-July. A short-lived, unprofitable long follows. Finally, I go short, and this time it pays off. By 8 November this final short has made me nearly 200 points.
The original and new trade logs are in table s 60 and 61.
Table 60: Final trade log for Euro Stoxx CFD, with stop loss There are no commissions. Excludes spread cost, since the price shown reflects the effect of paying the spread. P&L is profit or loss in price points for closing trades only.
Table 61: Final trade log for Euro Stoxx CFD, without stop loss There are no commissions. Excludes spread cost, since the price shown reflects the effect of paying the spread. P&L is profit or loss in price points for closing trades only.
I’m using the bet per contract CFD here, and the position size is always two contracts at €2 per contract, so the to tal loss is:
(–31–133–72–54+143.5) × 2 × €2 = – €154 = –$174
This is a slightly larger loss than before – when using a stop loss, I lost $131. The numbers would be very similar for the bet pe r point CFD.
S&P 500
Finally, have a look at figure 29 which shows the S&P 500.
Originally the stop loss got out of this trade on 24 October. I then opened a short, but this was at the worst possible level as the market then rebounded to $280. Using the opening rule instead this trade is closed a few days earlier on 16 October at a much better level. The short that follows is kept open until 8
November and ekes out a small profit. This is a much better outcome than in chapter six.
Figure 29: S&P 500 gets taken out earlier The original and new trade logs are shown in table s 62 and 63.
Table 62: Trade log for S&P 500 margin trading, with stop loss
‘Comm’: commissions. Excludes spread cost, since the price shown reflects the effect of paying the spread. P&L is profit or loss in price points for closing trades only.
Table 63: Trade log for S&P 500 margin trading, without stop loss
‘Comm’: commissions. Excludes spread cost, since the price shown reflects the effect of paying the spread. P&L is profit or loss in price points for closing trades only.
The first trade made $12.2 0 per share: 12.20 × 19 = $231.80
Commissions cost me another $2, but there was no funding cost as my leverage factor was less than 1. The net profit was 231.80 – $
2 = $229.80.
The second trade was also closed at a profit of $2.1 9 per share: 2.19 × 17 = $37.23
Commissions cost me another $2, but as I was short I also had to pay a borrowing fee of 0.25% a year; this came to $0.49. Sadly, my broker didn’t pay me any interest on the cash I realised from selling short. The profit on the second trade w orks out to: $37.23 – $2 – $0 .49 = $34.74
The total profit was 229.80 + 34.7 4 = $264.54.
Table 64: Summary of trading profit and loss, with and without stop loss
In three of the five examples profits are slightly lower with a stop loss (gold, corn, and Euro Stoxx), with no difference in AUDUSD, and a substantial improvement in S&P 500.
Of course, a handful of trades don’t by themselves establish a clear case for removing stop losses. Looking at a few isolated case studies like these is meaningless: the large improvement in S&P 500 in particular was pretty fluky. Decisions should be made on back-tests done with many years of history, and over large numbers of instruments. But they should reassure you that it’s safe to trade without a stop loss.
¹²¹ If you have been paying attention you may be thinking that is quite a wide stop: it equates to 30% of the price. The volatility of Royal Mail shares was about 25% a year at this point, implying a stop loss of 12.5% of the price using the normal volatility fraction of 0.5. In this particular trading account, I have opted for wider stops to increase my expected holding period, partly because I am trading the actual shares in this account rather than CFDs or spread bets, and actual UK shares are quite
expensive to trade because of the 0.5% stamp duty tax. This is an example of adapting a system to your own requirements.
¹²² As I’ve mentioned before you can avoid this scenario by using a guaranteed stop loss. But these are too expensive.
¹²³ Traders who are really nervous should consider leaving stop loss orders with their broker, or even using guaranteed stop losses, though I’m not keen on either of these options: if you leave a stop loss order with your broker then they can deliberately push the price or widen the spread to trigger the stop, and guaranteed stop losses cost more.
Chapter Ten
Position Adjustment
Imagine you are playing Texas
hold ‘em poker. Do not worry if you are not familiar with the rules. At the start of the game you are dealt two cards. You then decide whether to play, and how much to bet, based on how strong 20
your hand is. ¹²74 ‘Pocket aces’ – a pair of aces – is the best possible starting hand. Being dealt a pair of nines isn’t nearly as good but is still a hand that’s worth playing. An experienced poker player would bet heavily if they got pocket aces, but with a pair of nines they would place a more m odest stake.
After the initial round of betting the dealer lays three further cards face up on the table. Players can then choose to bet more, match the bets made by other players, or leave the game (known as
‘folding’). Two further rounds follow where more cards are dealt, and more betting or folding is possible.
A smart poker player will modify their betting strategy as the game goes on. As more cards are revealed, and other players make their bets, they will revise their opinion about the likely outcome of the game. They will change their opinion about how much risk they should take, given the informatio n available.
But in the Starter System we always bet the same amount. The same amount of cash is at risk each time, regardless of how confident we are about our trade. Also, we keep the same position on, even if our confidence or the risk in the market changes. This makes no sense in a poker game and, in this chapter, I explain why it makes no sense when you are trading.
Specifically, I w ill discuss:
adjusting position size according to how much confidence you have in your trading rule
adjusting position size as r isk changes.
Adjusting position size for confidence: non-binary trading
Consider figures 30 and 31 (these should look familiar: they are exact copies of figures 11 and 12 back i n part two).
Figure 30: Opening rule goes long Euro Stoxx equity index without much conviction
Figure 31: Opening rule goes short corn with huge enthusiasm
In one situation (Euro Stoxx) we go long two CFD contracts, and in the other (corn) we go short two futures contracts. By coincidence the size of our position is the same (two contracts), but more importantly the risk of the positions we take in the Starter System will be identical (and equal to whatever it takes to get 12% target risk on our trading capital). The rule in the Starter System is a binary trading strategy: there are only two options – lo ng or short.
But on closer inspection the two scenarios are quit e different.
In figure 30 (Euro Stoxx) the two moving averages are really close together on 18 June, and on the verge of re-crossing.
Whereas in figure 31 (corn) on the same day there is a large, and growing, gap between the two moving averages. Euro Stoxx shows a relatively weak uptrend, corn a really strong downtrend. Yet we have the same sized risk for bot h positions.
Does that make sense?
It does not.
Ideally, you should size your positions according to how confident your trading rule is . If the basic Starter System wants to hold two contracts, then with a weak signal as in figure 30 you might only buy a single contract. If the signal is
exceptionally strong as in figure 31 you might go short three, or even four, contracts. This is non-bin ary trading.
Why non-binary trading mak
es sense
Non-binary trading only makes sense when combined with the idea of continuous trading without a stop loss, which I introduced in 20
the last chapter. ¹²75 With these two methods together the trading process now look s like this:
Starting with no position, we check our o pening rule.
We enter into a position in the direction indicated by the opening rule. The magnitude of that position depends on how confident the rule is.
We monitor the opening rule. If the rule gets more confident 20
about the position, then we increase it. ¹²76 If it gets less confident, then w e reduce it.
When the rule has insufficient confidence, and is unsure whether to be long or short, then our position woul d be closed.
Once the rule has some confidence again, we’d open a small new position in the appropriat e direction.
Our position would then get larger if the rule got more confident, and vice versa.
Why does it make sense to increase your position when you are mor e confident?
Consider the trend following rule used in the Starter System. On 18 June in figure 30 the trend is finely balanced, with the moving averages really close together. A relatively small movement in price either way will tip you from long to short. But small price movements are pretty random, and in this tight situation you can’t be sure about the trend direction. Your position should be much smaller than in figure 31 where the trend is well established. I could make a similar argument for the breakout rule I introduced in chapter eight, and any other trend fol lowing rule.
What about the carry rule that I introduced in ch apter eight?
Let’s use an FX example, as that will be intuitively clearer. The carry rule for FX uses the difference between the interest rates of different countries. As I write this chapter, interest rates are at these levels: for USD 1.919%, for AUD 1.898%, and for GBP
0.569%.
The carry rule would want to go long USD vs GBP, as 1.919% is well above 0.569%. It would also go long USD vs AUD, since 1.919%
Leveraged Trading: A professional approach to trading FX, stocks on margin, CFDs, spread bets and futures for all traders Page 20