# How Have Our Trading Edges Performed?

Last Updated on September 19, 2022 by Quantified Trading

As of writing (August 2022), we have published 19 different monthly trading edges in the following ETFs (or you can also trade any relevant futures contracts): SPY, QQQ, XLP, TLT, FXI, SIL, XLU, DAX, and HYG. The strategies are divided into swing trading strategies and overnight strategies. Most of them only trade the long side, but TLT and XLP have short strategies included.

**In this article, we backtest all our trading edges as a portfolio of trading strategies (we excluded the DAX trading edge of November 2021 to simplify the backtest). This is how we did it:**

Table of contents:

## How we did the backtest

When you are trading you face a “dilemma”: you only have a certain amount of capital, and you hopefully have many different strategies over a range of different assets. You need to ask yourself at least these two questions:

- How much capital do you allocate to each strategy?
- How much capital do you allocate to each asset class?

When you are about to take a position today, you don’t know how many signals you’ll get in the coming days. Because of this, and the allocations you make, live results can deviate significantly from your backtests. Moreover, some strategies don’t work well together, even if they look great on any independent backtest.

We do two backtests in this article: the first backtest is with a maximum of 4 positions at any time, and the second backtest is with a maximum of 3 positions. All backtesting is done in Amibroker (we even have a pretty extensive Amibroker course).

We have not done any strategy optimization or portfolio optimization. Our portfolio composition can be improved – a lot.

This is all there is to it. This also means we are having a lot of “idle” cash on the sidelines at any time (when there are few signals). As you’ll see in the backtests, the time spent in the market is pretty low.

Because all of the strategies were developed between 2012 to 2017, we divide the backtest into three samples: 2007-2017, 2018-2021 (out of sample backtest), and 2022. We separated 2022 because of the difficult market.

## Monthly trading edges – backtest 1

Let’s make our first backtest. We make the following assumptions:

- Max 4 positions at any time.
- We allocate 24% of our equity to a position to any strategy (not asset), except in the SIL strategy where we have only 3% (it’s illiquid).
- For example, if we have a signal in SPY for both an overnight until tomorrow’s open and one overnight position for the close, we allocate 48% of the equity on these two positions (24% each).
- Amibroker picks trades randomly if there are signals at the same time to avoid having more than 4 positions.

### In sample period 2007-2017

The equity curve looks like this:

There are 1015 trades, the average gain is 0.5%, CAGR is 9.6% (risk-adjusted 38%), max drawdown is 8%, the win rate is 71%, time spent in the market is 25%, and the profit factor is 2.4.

### Out-of-sample period Jan 2018 to Dec 2021:

The equity curve looks like this:

There are 363 trades, the average gain is 0.51%, CAGR is 9.6% (risk-adjusted 39%), max drawdown is 4%, the win rate is 71%, time spent in the market is 24%, and the profit factor is 2.3.

### The bear market of 2022

The strategies have held up pretty well during the bear market of 2022:

There are 78 trades, the average gain is 0.36%, CAGR is 11.8% (annualized, risk-adjusted 47%), max drawdown is 4%, the win rate is 65%, time spent in the market is 27%, and the profit factor is 2. The main reason for the drawdown in June is two losses in TLT and XLP. However, the few short trades have helped the portfolio a lot with an average of 0.8% per trade.

Some readers might say the equity curve of 2022 looks erratic. But that’s because this is a much shorter time interval. Every equity curve looks better the more time is compressed in the chart!

Let’s end this backtest with a summary of the annual returns per year for the whole period (2022 is lower than the above because of annualization):

## Monthly trading edges – backtest 2

The second backtest sets a maximum of 3 positions at any time. To simplify our backtests in this post, we make the following assumptions:

- Max 3 positions at any time.
- We allocate 33% of our equity to a position to any strategy (not asset), except in the SIL strategy where we have only 3% (it’s illiquid).
- For example, if we have a signal in SPY for both an overnight until tomorrow’s open and one overnight position for the close, we allocate 66% of the equity on these two positions (33% each).
- Amibroker picks trades randomly if there are signals at the same time to avoid having more than 3 positions.

### In sample period 2007-2017

The in-sample period is a bit better with 3 instead of 4 positions (as expected):

There are 985 trades, the average gain is 0.49%, CAGR is 12.5% (risk-adjusted 36.4%), max drawdown is 11%, time spent in the market is 34%, the win rate is 71%, and the profit factor is 2.3.

### Out-of-sample period Jan 2018 to Dec 2021

Again, a slight improvement compared to having 4 positions:

There are 352 trades, the average gain is 0.48%, CAGR is 11,7% (risk-adjusted 35.8%), max drawdown is 6%, time spent in the market is 32%, the win rate is 70%, and the profit factor is 2.2.

### The bear market of 2022

There are 75 trades, the average gain is 0.4%, CAGR is 17.1% (risk-adjusted 46%), max drawdown is 5%, time spent in the market is 37%, the win rate is 65%, and the profit factor is 2.1.

The value of having short strategies is easy to spot in 2022: There are just 10 short trades, but the average gain per trade is a solid 0.8% and make a huge contribution. (Please also see our short strategy bundle.)

As expected, the results improve because we have less idle capital on the sidelines. The annual returns are listed below (max 3 positions):

## Our opinion on the backtests and results

We are pretty happy with the results for 2022. The main reason is that we want uncorrelated returns to the overall stock market (S&P 500).

However, many traders have completely unrealistic expectations.

First, the most important thing about short-term trading (in our opinion) is to get uncorrelated returns to the overall stock market (see more about this below). Second, if you are able to get 10-15% unleveraged returns over a decade you are really good – we would even call you a maverick.

## What kind of return can you expect going forward?

As a rule of thumb, we always assume that the future return will be around 10 – 25% worse than the backtest.

Why so?

Because you have to factor in trading mistakes (fat fingers), behavioral mistakes, strategy deterioration, commissions, and slippage. For example, the best period was during the Covid mess in March and April 2020. Were you able to press ahead with the signals when the markets fell 10% in a single day? We can assure you most traders would not. Most traders would stop trading and wait.

Currently, commissions are close to negligible and so is slippage in all the above ETFs (except for SIL). Please check our article on slippage in trading.

However, the sad fact about trading is that most strategies deteriorate a little over time. This is a cost you have to factor in. Hence, always expect slightly lower returns in real trading.

## Why do we trade?

Why do we trade?

- It’s fun
- It’s challenging
- It’s profitable uncorrelated returns compared to buy and hold
- If you are good, you can leverage

Personally, we believe it’s wise to BOTH invest (buy and hold) and does short-term trading. The main reason for trading is to create uncorrelated returns. A good example of that is in 2008, 2018, 1Q 2020, and 2022.

If you are serious about trading, we strongly recommend the following articles abut correlation and trading:

- What does correlation mean in trading? (Trading strategies and correlations)
- Uncorrelated assets and strategies â€“ benefits and advantages (examples and backtests)
- Does your trading strategy complement your portfolio of strategies?
- Why build a portfolio of quantified strategies (including two strategies)

## Relevant articles

WeÂ end the article by listing some relevant articles about automated trading and backtesting:

- How To Optimize A Trading Strategy? â€“ (Example and Definition of Optimization & Backtesting)
- 26 trading lessons learned after 20 years of full-time trading
- How Jim Simonsâ€™ trading systems made 66% a year (The Medallion Fund)
- Can you get rich by quant trading? (Tips and tricks for quant traders)
- 8 pros and cons of quant trading (Quant trading strategies)
- Advantages With Mechanical Trading Strategies (The Parts Of A Mechanical Trading Strategy)
- Mechanical Trading Strategies Vs. Discretionary Trading Strategies
- News And Trading â€“ How Important Is It?
- Proprietary trading â€“ pros and cons (a personal experience)
- How to get started in trading (How can a beginner start trading)
- Survivorship bias in backtesting, trading, and investing (How To Avoid It)
- Disadvantages of backtesting (Why backtesting doesnâ€™t work)
- How to fail as a trader: 11 skills of the unsuccessful (inverse thinking)
- Out-of-sample trading tests explained (what is out-of-sample backtesting?)
- How to generate trading ideas (What are trading ideas?)
- Is It Possible To Make Money Swingtrading? My numbers 1st half 2013
- Is the stock market a zero-sum game?