There are other great books if you just want in-depth knowledge of Python, such as Python for Data Analysis (McKinney, 2017) or Python for Finance (Hilpisch, 2018). And if your interest is in a deep dive into systematic trading, you should look at something like the aptly named Systematic Trading (Carver, Systematic Trading, 2015). (Location 228)
Tags: orange
A solid model trades a real market phenomenon, aiming for a certain type of return profile. (Location 380)
Tags: orange
If you just experimented with settings until the results looked good, all you have done is fitted the algorithm to the known data. (Location 388)
Tags: orange
A proper trading model needs to start off with a theory about market behavior. (Location 391)
Tags: orange
The second and perhaps surprising way is based on a perceived need or business opportunity. Someone working full time with developing trading algorithms may not have the luxury of dreaming up anything he or she wants. (Location 406)
Tags: orange
Generally speaking, you should aim for as few rules as possible and as few variations as possible. (Location 425)
Tags: orange
Robust trading models, those that work over the long run, tend to be the ones that keep things simple. (Location 428)
Tags: orange
No, skip the optimization. But make a few variations of the rules, using reasonable, sensible numbers. (Location 451)
Tags: orange
Adding a specific rule to deal with 2008 makes your backtests look great, but it may constitute over-fitting. (Location 467)
Tags: orange
But for systematic, quantitative traders, another definition may be more practical. When looking at the various markets we have available to us, we can group them in different ways. (Location 482)
Tags: orange
No, diversification is the way to go. Apply your trading strategy on multiple markets, and your probabilities of success are much improved. (Location 503)
Tags: orange
This is the first thing that most people think of when designing a trading model. It’s the most obvious part, but it’s not necessarily the most important part. (Location 528)
Tags: orange
Remember that most professional trading models aim to hold a certain amount of portfolio risk on a position. (Location 548)
Tags: orange
Financial risk is about potential value variation per unit of time. (Location 557)
Tags: orange
The most common way to quantify risk is about measuring past volatility. (Location 571)
Tags: orange
If you are a risk manager or risk controller of a larger portfolio, of a number of portfolios or something as complex as a whole bank, risk can be very mathematically intensive. (Location 580)
Tags: orange
Ten percent in a month, ten percent in a year, or ten percent in ten years are all very different situations. (Location 585)
Tags: orange
The first issue has to do with correlation. This should be considered when you have multiple positions open, and has to do with how they relate to each other. (Location 595)
Tags: orange
If on the other hand, you hold a stock and a commodity, they could potentially be completely unrelated, and holding them both could be lower overall risk than only holding one of them. (Location 598)
Tags: orange
This is why the Sharpe Ratio remains one of the most commonly used comparison measurements. This ratio takes the overall returns, deducts risk free interest for the same period, and divides this by the standard deviation. (Location 609)
Tags: orange
From a valuation perspective, it does not matter if you have closed your position or not. The value of your position, and in extension of your portfolio, is not affected by whether or not you still have an open exposure. (Location 629)
Tags: orange
These methods are usually collected under the umbrella term Money Management. This is a term that you won’t likely hear in the financial industry, but which seems quite common in the hobby segment. (Location 639)
Tags: orange
From a logical point of view, if you missed the first trade, you should now buy the full 2,000 shares. That’s the only way that you end up with the same portfolio state as originally intended. (Location 664)
Tags: orange
The bad news is that your returns are likely to be less than 15% p.a. in reality over longer time periods. The good news is that if you can achieve that, you can make a lot of money in this business. But of course, performance need to be paid for, with volatility. (Location 729)
Tags: orange
Strategies with Sharpe of 3 or even 5 do exist, but they tend to be of the so called negative skew variety. That expression, referring to the shape of the return distribution, means that you win small most of the time, until you suddenly get hit with a big loss. For such strategies, you may see long periods of constant winning, only to suffer a sudden and sometimes catastrophic loss. (Location 753)
Tags: orange
Most people today view programming as some sort of specialized task for propeller heads. Programming is for programmers. The kind of people who do nothing but program all day. (Location 793)
Tags: orange
Python is to a large extent purpose built for finance. There are tools available which are designed by hedge fund quants and made available to everyone for free. (Location 814)
Tags: orange
There are some clear issues with Python of course. Given that most people using Python at the moment are hard core quants, documentation tends to assume that you already know everything and there is a clear aversion to anything user friendly. (Location 828)
Tags: orange
I want to briefly mention the useful but annoyingly spelled application Spyder, before moving into more detail about the equally annoyingly spelled Jupyter Notebook environment which we will be using for almost the entire remainder of this book. (Location 890)
Tags: orange
1).cumprod() (Location 1973)
Tags: pink