Scribd
Upload
259 views27 pages

AdaptiveStrategies MattRadtke 02oct2018

The document discusses creating adaptive trading strategies. It defines quantified trading strategies and the process of backtesting and optimization. Adaptive strategies change rules based on market conditions rather than using fixed rules. The document outlines a 9-step process to create an adaptive strategy, including defining market regimes, testing variations, and comparing the adaptive strategy's performance to static strategies. It finds that while not a panacea, adaptive strategies can potentially generate better future results than a single best static strategy.

Uploaded by

Pradeep Arora
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
259 views27 pages

AdaptiveStrategies MattRadtke 02oct2018

The document discusses creating adaptive trading strategies. It defines quantified trading strategies and the process of backtesting and optimization. Adaptive strategies change rules based on market conditions rather than using fixed rules. The document outlines a 9-step process to create an adaptive strategy, including defining market regimes, testing variations, and comparing the adaptive strategy's performance to static strategies. It finds that while not a panacea, adaptive strategies can potentially generate better future results than a single best static strategy.

Uploaded by

Pradeep Arora
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 27

Creating

Adaptive
Trading
Strategies
Matt Radtke
A few details about my background:
• BS in Computer Science
• 25+ years of software industry experience
• Started trading stocks and options in 2008
• Worked for Connors Research in various roles 2012-Present
• Independent AmiBroker consultant

Introduction: Matt Radtke


• Quantified Trading Strategies
– Definition
– Back Testing
– Optimization
– Challenges

• Adaptive Strategies
– Definition
– Rationale
– Creation Process
– Verification
• Q&A

Agenda
Quantified
Trading
Strategies
A fully quantified trading strategy has precise rules for finding entry
and exit signals, ranking trade candidates (setups), entering new
trades, managing existing trades, and exiting each trade.

Quantified rules can be back tested using tools such as AmiBroker or


TradeStation. They can also be fully automated for live trading if
desired.

Quantified Strategies: Definition


To produce valid back test results, it is essential to use a disciplined
approach:
1. Define a set of quantified rules for ranking, entering, exiting, and
managing each trade.

2. Apply those rules consistently to all members of your trading


universe, as that universe existed at the time of the trade.
3. Manage capital realistically

Back Testing
When defining a set of rules, we typically don’t know in advance
which threshold values (parameters) will generate the best
performance. For example, after a Setup occurs, should we enter the
next day on a 1%, 2% or 5% limit order?
The solution to this problem is to test many different variations of the
strategy, with each variation including a unique set of strategy
parameters. AmiBroker refers to this process as “optimization”.

Optimization (AmiBroker)
The most common objection to back testing / optimization is that it is
simply an exercise in curve fitting.
Even if you are careful to avoid curve fitting per se, it can be difficult
or impossible to select the strategy variation(s) that will perform well
in the future, because results can be quite sensitive to the time
period over which you tested.
One way to alleviate this problem is to use Adaptive Trading
Strategies.

Challenges
Adaptive
Trading
Strategies
While a typical static trading strategy uses a fixed set of rules and
parameters, an adaptive strategy changes in response to current
market conditions or other factors.
For example, instead of defining one of our Setup rules as:
RSI(2) < 10
We might use a threshold of 5 in bear markets and 20 in bull markets.

Adaptive Strategies: Definition


By focusing on a single market regime or condition at a time, we
increase the likelihood that the future performance of the strategy
will more closely resemble the past behavior under similar market
conditions.
Two implicit assumptions are:
1. We have a quantified definition of market regime
2. We are able to generate a sufficient amount of data for each
market regime from which to draw conclusions

Rationale
The process for creating an adaptive strategy is somewhat involved,
but worth the effort. Below is a summary of the steps, which we will
then illustrate with an example.
1. Define market regimes
2. Define the trading strategy
3. Using a baseline date range, test different variations of the
strategy using a wide range of parameter values
4. Summarize interesting metrics by market regime

Summary of Creation Process


5. For each market type, select the “best” set of strategy parameters
6. Build an adaptive strategy that utilizes the best parameters for
each regime. Note that for certain market types, the best
approach may be to stop trading!
7. Test the adaptive strategy over a new (out of sample) date range
8. Test all variations of the original, static strategy over the same
date range
9. Compare the performance of the adaptive strategy to the best
variations of the static strategy

Summary of Creation Process, continued


Definitions for market regimes will vary with the type of trading you
do. For example, a day trader probably won’t care whether the SPY is
currently above or below its 200-day moving average.
Market Conditions will often incorporate both a price component and
a volatility component. For this test, we have four market types:
1. Volatile Bear
2. Quiet Bear
3. Volatile Bull
4. Quiet Bull

Exact rules can be found in the spreadsheet.

Step 1: Define Market Conditions


This is a simple mean reversion strategy based on RSI(2).
A Setup occurs when all of the following conditions are true:
• Stock is a member of the S&P 500
• Average daily Turnover (Close*Volume) for past 21 days > 1M
• RSI(2) < [5,10,15...25] for at least [1,2,3] Days
• C > MA(200)
• HV(100) > [15,25,35]

Step 2: Define Trading Strategy


Buy the stock on the day after the Setup occurs, using a [1,2,4,6] %
Limit Order below the Setup close.
Trade candidates are ranked by highest HV(100).
Up to 10 open positions are allowed.
Each position is allocated 10% of current equity.

Sell the stock when RSI(2) > [60,70,80]

Step 2: Define Trading Strategy


To generate data specific to a variety of market
regimes, I ran an AmiBroker Optimization over the
time period 1/1/2000 – 12/31/2012.

Let’s look at those results in the spreadsheet.

Steps 3 & 4: Run Optimization


For this illustration, we’re going to focus on some basic performance
metrics: Compound Annual Return (CAR), Max Drawdown (MDD), and
CAR/MDD.

Let’s look at how the best variations performed, and also review each
market regime.

Step 4: Summarize Metrics by Market Type


Step 5: Select Best Parameters for Each Market Type
When creating the adaptive version of the strategy, keep the
following in mind:
• All parameters, including scaling rules, stops, profit targets, exit
types/thresholds, etc. are determined by the market regime on the
Setup day.
• If you are writing code for a tool such as AmiBroker or TradeStation,
it will take some extra work to track the parameters for each trade
properly even if the regime changes while the trade is open.
• Trading an adaptive system without some kind of automated scan
can be extremely challenging. Create a reliable tracking system!
Step 6: Build the Adaptive Strategy
To test the adaptive version of the strategy, I ran an AmiBroker
back test over the date range 1/1/2013 – 12/31/2017.

We’ll look at those results momentarily.

Step 7: Test the Adaptive Strategy Over Recent Dates


I also ran an AmiBroker Optimization over the same date range:
1/1/2013 – 12/31/2017.

Let’s see how the results compare…

Step 8: Run the Full Optimization Over Recent Dates


Highest CAR, 2013-2017

Step 9: Compare the Adaptive Results to Static Results


Highest CAR, 2000-2012

Step 9: Compare the Adaptive Results to Static Results


As we’ve seen, using an Adaptive strategy is not a panacea.
However, it does provide an opportunity to generate better results on a
going-forward basis than simply selecting a strategy variation that uses
static parameters.
From my research, it appears that one of the most important components
of building a successful Adaptive strategy is defining the market regimes.
• Each regime should be well-represented within the back test
• Regimes should not change too often

Conclusion
Q&A
Please feel free to reach out to me with any questions, comments,
observations, or other feedback.
• Email: quantforhire@gmail.com
• Web Site: https://QuantForHire.com

Thanks for attending!

Contact Information

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy