It was 23 January 2009 and I was in my London office... Data was about to be released indicating how the UK economy had performed in the last three months of 2008. It would be bad news - the official confirmation that we were in recession - but nobody knew how bad. This didn't mean extra work for me however, since a bank of computers would adjust our clients' portfolios automatically when the news arrived.... With a stressful full-time job, I was not a particularly active trader but very occasionally an opportunity came up that was too good to miss. This was one of them. In my research I found that historically when people's fears were confirmed by terrible economic numbers was often the best time to buy; and this was potentially the worst news I'd seen in my lifetime.
(page 1)
The author's description of the opportunity in January 2009 reminds me of my own experience with NVDA in March 2020. Although I went all in on NVDA calls near the bottom, I couldn't hold onto the position. If I held for 5 days (instead of a day), it would have been my biggest single trade to date. After months of painful reflection, I gradually realized that even if given the same opportunity again, I might profit a bit more than before, but I still wouldn't be able to hold until the end. There's a right path in investing. When entering, one should decisively seize opportunities and be willing to increase the position size. When exiting, especially in situations without using day trading or high leverage, it's necessary to be more "obtuse". So, how can you compel yourself to become more "obtuse"? I think you can approach it from two angles.
On one hand, if your disposition leans toward reducing positions and you're uncomfortable with adding to positions when the trend is in your favor, you can force yourself to only sell a portion of your position that's currently profitable when you decide to exit. This way, most of your positions remain if the trend persists. On the other hand, you can employ multiple timeframes to better time your entries and constrain your exits.
Generally, entry and exit signals on lower timeframes precede those on higher timeframes. However, it's important to note that a signal on a lower timeframe doesn't guarantee that the stock's trend will align with your desired direction on a larger timeframe. Many traders label this phenomenon as false signals, but I see it differently. Any issued long signal is a valid signal, not noise. Let me use a long signal as an example to explain my perspective.
The market's opinion of an asset can be represented on a spectrum, ranging from extreme bearish (-1), moderately bearish (-0.8), neutral (0), moderately bullish (0.8), to extremely bullish (1.0). Let's call this trader's opinion distribution "opt." In addition, there's the trader's capital distribution, denoted as "cap." These two distributions ultimately result in a market's capital opinion distribution ("capital's opinion spectrum"). The average of this distribution represents the market's overall sentiment. Any opinion diverging from 0 can alter the market's direction, even if momentarily. However, it's important to note that this distribution itself possesses self-catalytic properties (what Soros refers to as reflexivity): capital originally intended for long positions may degrade to neutral positions (weakening the counter-trend force) due to bearish market sentiment or even turn into short positions (strengthening the existing trend force). This self-reinforcing trend is the underlying cause of major market movements. Therefore, an immature "false long signal" might have a reasonable nucleus but lacks a complete growth process. In other words, the followers of this bullish signal might not be numerous enough to foster a robust trend. A long signal on a 30-minute timeframe may not always evolve into a long signal on a 1-hour timeframe.
Likewise, a closing signal on a 30-minute timeframe might not smoothly transition to a closing signal on a 1-hour timeframe. However, multiple valid 30-minute long/closing signals can eventually lead to the appearance of 1-hour long/closing signals. Hence, one possible way to become more prudent is to use a 30-minute signal for entry (with a corresponding risk stop at this timeframe) and a 1-day signal for exit.
Where an asset has a higher chance of a large down move than an equivalent up move, it is said to have a negative skew. If large up moves are more likely then it has positive skew. Assuming they have the same Sharpe ratio, the returns from a positively skewed asset will contain more losing days than for those of a negatively skewed asset. But the losing days will be relatively small in magnitude. A negatively skewed asset will have fewer down days, but the losses on those days will be larger.
(page 33)
Executing an operation where you enter on the 30-minute timeframe with a risk stop set at this level and exit based on a 1-day signal seems to be a strategy with a positive skew. This approach is geared towards the hope that the 30-minute trend can evolve into a weekly trend. It's worthwhile to take the possible loss within the price range of a 30-minute stock movement to potentially capture a significant weekly trend, given the potential return. However, this approach should be validated through backtesting.
Equities normally have mildly negative skew. 'Safe haven' assets like gold and Swiss francs tend to have positive skew. However the skew of these assets is relatively mild compared to owning options....Buying the VIX futures gives you highly positive skew, but as you are effectively purchasing insurance against unexpectedly high equity volatility it also tends to have a negative Sharpe ratio. Similarly selling the futures gives a positive Sharpe ratio, but with an extremely negative skew.
(page 34)
One of the most powerful techniques I use in my trading system framework is volatility standardization. This is adjusting the returns of different assets so that they have the same expected risk. As I discussed above, my standard definition of expected risk is to use an estimate of recent standard deviation. This has a number of benefits. It allows you to have portfolios where each component contributes an equal amount of risk. Furthermore, as you will see later in the book it means that you can apply the same trading rule to different assets, if the trading rule is applied to volatility-standardized returns.
Be aware that if you are making steady profits nearly every day, and most of your trades are winners, then there is a good chance you are engaged in negative skew trading. It's just that you haven't yet seen any rare large losses.
(page 40)
Selling naked put is also a negative skew strategy.
The law of active management, first articulated by Richard Kahn in 1989, states that the annualized Sharpe ratio of a trading strategy will be proportional to the square root of the number of independent bets made per year.
(page 42)
Higher trading frequency tends to lead to a higher Sharpe Ratio for a strategy.
Mean reversion and relative value traders act as contrarians -- they seek to take advantage of mis-pricing which means buying low after falls and selling when the price has risen. Other styles of trading involve following the market; notably the various forms of trend following. Contrarian traders like to catch falling knives and buy more as prices fall. Trend followers will close positions that have started to lose money, like the early loss taker trading rule. Market followers tend to see positive skew from taking small losses as the trend moves against them, with an occasional large profit from a significant move in their favor. Conversely contrarians see negative skew, which many small profits as each mispricing is corrected, then occasional large losses when prices jump away from their equilibrium.
(page 45)
This statement is indeed brilliantly articulated and crystal clear. For speculators, it's crucial to discern their style and personality inclinations and then decide whether to pursue mean reversion or trend following.
There are two apparently easy ways to try and increase them, both of which are incredibly dangerous. Firstly you could trade negative skew strategies. Very high Sharpe ratio is often a result of hidden negative skew. The second path to the mirage of higher Sharpe ratio is to trade more quickly.
(page 47)
There are two direct methods to enhance the Sharpe ratio: employing strategies with negative skewness and engaging in high-frequency trading. However, both of these approaches also increase one's corresponding level of risk.
The solution (to make a good trading system) is to separate out the components of your system: trading rule (including explicit or implicit stop losses), position sizing, and the calculation of your volatility target (the average amount of cash you are willing to risk). You can then design each component independently of the other moving parts. Trading rules and stop losses should be based only on expected market price volatility, and should never take your account size into consideration.
(page 95)
Modularizing a trading system makes it easier to conduct backtesting and identify issues.
For two stocks each with identical return volatility of 10%, with half money in each, then the volatility of the whole portfolio will depend on how correlated the two assets are. If they are perfectly correlated then the portfolio will have a return standard deviation of 10%; the same as the individual assets. But if the correlation between the two assets was 0.5, the portfolio volatility would come out at 8.66%. Similarly a correlation of 0 gives a volatility of 7.07%. More diversified portfolios have lower volatility.
(page 129)
Exploring the correlation of assets and the overall portfolio's volatility is a valuable subject. However, it's crucial to bear in mind that asset correlations can change over time. In certain extreme situations, two stocks with a correlation of 0 might end up having a correlation of 1. It's essential to fully understand your tools before applying them.
Diversification really is the only free lunch in investment. Allocating across different asset classes can easily double your expected Sharpe ratio. It is best to simultaneously run a portfolio of as many trading subsystems and instruments as possible and allocate your trading capital between them.
(page 165)
Diversification across asset classes is beneficial, but diversifying across various strategies is an even more effective approach.
1. A systematic trader should be humble, and underestimate their intelligence, skill, and luck. Assume your trading will go badly, be prepared for that eventually, and be pleasantly surprised if it doesn't. Don't try anything too clever; it is probably unnecessary and it's more likely to go wrong.
2. You should be skeptical.
3. Be pessimistic. Do not trust backtests, even if you haven't over-fitted them, and even if they have been done on a rolling out of sample basis. The future is unlikely to be quite as good as the past. In my opinion a highly diversified system of systematic trading rule is unlikely to beat a Sharpe ratio of 1.0.
4. A good systematic trader will be thoughtful. You should know why you might be making money and why you might not. Understand your markets and your trading rules.
5. Thriftiness is another virtue. Know your trading costs.
6. You should be nervous.
7. The best systematic traders will be diligent when creating their systems, but lazy when running them. Put the hard work into designing a safe system that you are comfortable with and then do not change it.
8. Finally to make money you need to be lucky.
(page 259-260 "What makes a good systematic trader”)
Each trader has their own guiding principles. To make them feasible, it's advisable to maintain a concise list.