9/21/2019
1
1
by Scott Juds – SumGrowth Strategies – Sept. 2019
2
Disclaimers
• DO NOT BASE ANY INVESTMENT DECISION SOLELY UPON MATERIALS IN THIS PRESENTATION
• Neither SumGrowth Strategies nor I are a registered investment advisor or broker-dealer.
• This presentation is for educational purposes only and is not an offer to buy or sell securities.
• This information is only educational in nature and should not be construed as investment advice
as it is not provided in view of the individual circumstances of any particular individual.
• Investing in securities is speculative. You may lose some or all of the money that is invested.
• Past results of any particular trading system are not guarantee indicative of future performance.
• Always consult with a registered investment advisor or licensed stock broker before investing.
9/21/2019
2
3
The Plan:
• Brief Summary of our Base Technology • How Artificial Intelligence Will Help• A Summary of How Merlyn.AI Works• The Merlyn.AI Strategies and Portfolios• Using Merlyn.AI within Sector Surfer• Why AI is “Missing in Action” on Wall Street?
• Let’s go Live Online and See How Things Work
Merlyn.AI Prudent Investing Just Got Simpler and Safer
4
Simply Type In
Merlyn.AILogistics:
9/21/2019
3
5
Merlyn.AI Builds On(and does not have to re-discover)
Other Existing Knowledge
It Doesn’t Have to ReinventAll of this From Scratch
9/21/2019
4
Merlyn.AI
Builds On Topof SectorSurfer
Academic Paper:
“Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency”
(1993)
Narasiman JegadeeshEmory University
Sheridan TitmanU. of Texas, Austin
Momentum in Market DataFirst Proved
Merlyn.AIAccepts This
Re-Discovery Not Necessary
9/21/2019
5
“the premier market anomaly” that’s “above suspicion.”
Momentum in Market Data
Academic Paper - 2008:
“Dissecting Anomalies”
Eugene FamaNobel Prize, 2013
Kenneth FrenchDartmouth College
Formally Confirmed
Merlyn.AIAccepts This
Re-Discovery Not Necessary
Signal-to-Noise Ratio
Controls the Probability ofMaking the Right Decision
Claude ShannonNational Medal of
Science, 1966
Proved
Merlyn.AIAccepts This
Re-Discovery Not Necessary
9/21/2019
6
Think Outside of the Box
Someplace to Start Designed for Performance
Matched Filter Theory
Design for OptimumSignal-to-Noise Ratio
J. H. Van VleckNoble Prize, 1977
Merlyn.AIAccepts This
Re-Discovery Not Necessary
Differential Signal Processing
Removes Common Mode Noise
(Relative Strength)
Wheatstone Bridge
Samuel H. ChristieRoyal Society 1836
5 Years Full Span
Merlyn.AIAccepts This
Re-Discovery Not Necessary
9/21/2019
7
Sectors Provide Power Strokes
Market Cycle
Economic Cycle
Merlyn.AIAccepts This
Re-Discovery Not Necessary
StormGuard-ArmorDetect the Onset of Bad Markets
Know When the Market is Safe: Risk-On vs Risk-Off
Merlyn.AIAccepts This
Re-Discovery Not Necessary
9/21/2019
8
StormGuard-ArmorDetect the Onset of Bad Markets
These Guys
Got us Here
Is There More?
9/21/2019
9
What About Selection Bias?Who Needs
XLV-Healthcare and XLE-Energy?
.
.
.
.
Merlyn.AI Is a Genetic Algorithm
Layered on Top of a Strategy
Why?To Remove Hindsight Selection Bias
Why?To Evolve its Set of Funds Each Month
Why?To Achieve Better Future Performance
Genetic Algorithm on Top
9/21/2019
10
Merlyn.AI Is a Genetic Algorithm
Layered on Top of a Strategy
This MeansYou Must Submit to a Higher Power
This MeansFund Selection is Automated
Merlyn.AI
Genetic Algorithm on Top
What is Artificial Intelligence?AI Definition
WikipediaArtificial intelligence (AI) is the ability of a machine to perceive its environment and take actions to maximize its chance of success at some goal.
Applied to InvestmentCan a machine perceive its environment (sector rotation, bull / bear markets) and take action to maximize returns and minimize risk? We say YES!
9/21/2019
11
Artificial Intelligence AlgorithmsTypes of machine learning algorithms[edit]
•Almeida–Pineda recurrent backpropagation
•ALOPEX
•Backpropagation
•Bootstrap aggregating
•CN2 algorithm
•Constructing skill trees
•Dehaene–Changeux model
•Diffusion map
•Dominance-based rough set approach
•Dynamic time warping
•Error-driven learning
•Evolutionary multimodal optimization
•Expectation–maximization algorithm
•FastICA
•Forward–backward algorithm
•GeneRec
•Genetic Algorithm for Rule Set Production
•Growing self-organizing map
•HEXQ
•Hyper basis function network
•IDistance
•K-nearest neighbors algorithm
•Kernel methods for vector output
•Kernel principal component analysis
•Leabra
•Linde–Buzo–Gray algorithm
•Local outlier factor
•Logic learning machine
•LogitBoost
•Manifold alignment
•Minimum redundancy feature selection
•Mixture of experts
•Multiple kernel learning
•Non-negative matrix factorization
•Online machine learning
•Out-of-bag error
•Prefrontal cortex basal ganglia working memory
•PVLV
•Q-learning
•Quadratic unconstrained binary optimization
•Query-level feature
•Quickprop
•Radial basis function network
•Randomized weighted majority algorithm
•Reinforcement learning
•Repeated incremental pruning to produce error reduction (RIPPER)
•Rprop
•Rule-based machine learning
•Skill chaining
•Sparse PCA
•State–action–reward–state–action
•Stochastic gradient descent
•Structured kNN
•T-distributed stochastic neighbor embedding
•Temporal difference learning
•Wake-sleep algorithm
•Weighted majority algorithm (machine l
Supervised learning
•AODE
•Artificial neural network
•Association rule learning algorithms
• Apriori algorithm
• Eclat algorithm
•Case-based reasoning
•Gaussian process regression
•Gene expression programming
•Group method of data handling (GMDH)
•Inductive logic programming
•Instance-based learning
•Lazy learning
•Learning Automata
•Learning Vector Quantization
•Logistic Model Tree
•Minimum message length (decision trees, decision graphs, etc.)
• Nearest Neighbor Algorithm
• Analogical modeling
•Probably approximately correct learning (PAC) learning
•Ripple down rules, a knowledge acquisition methodology
•Symbolic machine learning algorithms
•Support vector machines
•Random Forests
•Ensembles of classifiers
• Bootstrap aggregating (bagging)
• Boosting (meta-algorithm)
•Ordinal classification
•Information fuzzy networks (IFN)
Bayesian[edit]
Bayesian statistics
•Bayesian knowledge base
•Naive Bayes
•Gaussian Naive Bayes
•Multinomial Naive Bayes
•Averaged One-Dependence Estimators (AODE)
•Bayesian Belief Network (BBN)
•Bayesian Network (BN)
Decision tree algorithms[edit]
Decision tree algorithm
•Decision tree
•Classification and regression tree (CART)
•Iterative Dichotomiser 3 (ID3)
•C4.5 algorithm
•C5.0 algorithm
•Chi-squared Automatic Interaction Detection (CHAID)
•Decision stump
•Conditional decision tree
•ID3 algorithm
•Random forest
•SLIQ
Linear classifier[edit]
Linear classifier
•Fisher's linear discriminant
•Linear regression
•Logistic regression
•Multinomial logistic regression
•Naive Bayes classifier
•Perceptron
•Support vector machine
Unsupervised learning[edit]
Unsupervised learning
•Expectation-maximization algorithm
•Vector Quantization
•Generative topographic map
•Information bottleneck method
Artificial neural networks[edit]
Artificial neural network
•Feedforward neural network Logic learning machine
•Self-organizing map
Association rule learning[edit]
Association rule learning
•Apriori algorithm
•Eclat algorithm
Semi-supervised learning[edit]
Semi-supervised learning
•Active learning – special case of semi-supervised learning
Generative models
•Low-density separation
•Graph-based methods
•Co-training
•Transduction
Deep learning[edit]
Deep learning
•Deep belief networks
•Deep Boltzmann machines
•Deep Convolutional neural networks
•Deep Recurrent neural networks
•Hierarchical temporal memory
•Generative Adversarial Networks
•Deep Boltzmann Machine (DBM)
•Stacked Auto-Encoders
Other machine learning methods and problems[edit]
•Anomaly detection
•Association rules
•Bias-variance dilemma
•Classification
• Multi-label classification
•Clustering
•Data Pre-processing
•Empirical risk minimization
•Feature engineering
•Feature learning
•Learning to rank
•Occam learning
•Online machine learning
•PAC learning
•Regression
•Reinforcement Learning
•Semi-supervised learning
•Statistical learning
•Structured prediction
• Graphical models
• Bayesian network
• Conditional random field (CRF)
• Hidden Markov model (HMM)
•Unsupervised learning
•VC theory
How SumGrowth Will Use AITo Perceive the environment and take action to maximize success.
FWPT: Forward WalkProgressive Tuning
Adaptively changing the algorithm based on the past character of the data. Walks through out-of-sample data for its buy/sell decisions.
StormGuard - Armor
Employs Fuzzy Logic to evaluate a composite of 12 measures of the market’s character to determine current investment safety.
FWPP: Forward WalkProgressive Picking
Uses a Genetic Algorithm to evolve the candidate funds in a populationof momentum strategies to eradicate remnants of hindsight selection bias.
Old
Old
NewMerlyn.AI
9/21/2019
12
Removing Hindsight Selection Bias
Employ a Genetic Algorithm“Forward-Walk Progressive Picking”
Funds Picked with In-Sample Data, But Used with Out-of-Sample Data
How Our Genetic Algorithm Works
9/21/2019
13
How Our Genetic Algorithm Works
Genetic Evolution:MutationCrossover
Human
Consider This Analogy to Humans
Strategy
What are the Gene Mutations
9/21/2019
14
Where do the Gene Options Come From?
Scraped From ETF.com
How Our Genetic Algorithm Works
Each month mutant and crossover children are made.
Children
Population of 12
Only children with better “fitness” than their parents
survive by replacing the parent.
Population of 12
9/21/2019
15
Evolution History for Member-1
How Our Genetic Algorithm Works Evolution History for Member-6
9/21/2019
16
AIGenetic Momentum Portfolio Manager
Monthly Choice Made by SOS Vote(Like a Strategy-of-Strategies)
300x Time Lapse
9/21/2019
17
Merlyn’s Magic Portfolio
Merlyn’s Magic Portfolio
9/21/2019
18
(Play Merlyn video from desktop now)
But First…
This is SectorSurfer
9/21/2019
19
Needs vs Wants?
Why are you Here Today?1. Afraid Current Savings Won’t Be Enough?2. Advisors Only Set Me Up for Average -1%3. I Only Trade for the Fun and Challenge4. If I Had an ETF that did 15% Bull or Bear
and I could chase kids and travel I would not be here or look at screens all the time.
ETF Tax Efficiency?
An Exchange In-Kind Is a Non-Taxable EventOriginally Designed for Moving an Accounts to Another Brokerage
Market Makers Exchange A Basket of Stocks for ETF SharesIndividuals Buy and Sell Only ETF Shares – It is Not Like Mutual Fund Ownership
Market Makers and Exchange In-Kind for the Current Basket When Shares are Created or Destroyed.
Thus Trades Occur Within the ETF Change the Basket, BUT the Investor Gets Long Term Tax Treatment.
9/21/2019
20
Merlyn.AI Corp. NewsFounded Jan 2019, Raised $2.5M, Exclusive License from SGS to Create & Market Merlyn ETFs
Merlyn.AISGS License
InvestorsETF Sponsor
SolactiveCalculatorPublisher
MarketingCable CNBC
Advisor SharesG.Adwords
Articles
SGSSectorSurferAlphaDroidMAI Indexes
Alpha ArchitectETF Advisor
Exemptive ReliefWeb ServicesCompliance
US BankCustodian
RBCMarket Maker
NYSE
QuasarDistributor
Mktg. Approval
FINRASEC
Bull-Rider Bear-Fighter
Index
9/21/2019
21
MAI Bull-Rider Bear-Fighter Index
Solactive – Index Publisher
9/21/2019
22
MAI Indexes(Bull-Rider Bear-Fighter Will Soon be Here)
Bull-Rider Bear-Fighter IndexGoogle Search – Stay Up To Date
9/21/2019
23
Why are AI Funds M.I.A. on Wall Street?
hello
Oct. 2017
June 2018
Why are AI Funds M.I.A. on Wall Street?
AIEQ$119M
9/21/2019
24
Why are AI Funds M.I.A. on Wall Street?
AIIQ$3.7M
Why are AI Funds M.I.A. on Wall Street?
BUZRIP Mar. 2019
Used AI to Find Patterns in Twitter Messages
9/21/2019
25
hello
August 2018
Why are AI Funds M.I.A. on Wall Street?
A Distinction Without a Difference
$7.2M AUM Ave. After 18mo.
Why are AI Funds M.I.A. on Wall Street?
Too Many Simultaneous Problems to Solve!
• Discover that Signal-to-Noise Ratio is the Problem & Solve It.
• Learn How to Detect a Bear Market & How to Treat it Differently
Neural Network Future Leaks Into the Past
• Neural Networks are a very complex matrix of connections
9/21/2019
26
We Will Go Live After BreakTo See How Things Really Work
(Do A Live Online Demo of Everything)
52
Merlyn.AI Prudent Investing Just Got Simpler and Safer
by Scott JudsSeptember 2019