Emailing Algo Convention 2019 PDF

www.dravyaniti.com www.algoconvention.
com
ALGO
CONVENTION
APRIL 20 & 21
HOTEL NOVOTEL IMAGICA, KHOPOLI. 2019
www.dravyaniti.com www.algoconvention.com
IMPROVISING
SYSTEMS
USING
FILTERS ALGO
TOM BASSO - FORMER CEO, CONVENTION
TRENDSTAT CAPITAL MANAGEMENT
INC. & “MARKET WIZARD”
2019 – DAY 1
Trading with Algos
By Tom Basso
Founder of Trendstat Capital
And
Founder of Enjoytheride.world
enjoytheride.world 1
Noise versus Data
Data
Data
Noise
Capture Information, Ignore Noise
1. Noise is relative to time frame

measured

measured
2. The more room you give for noise,
the later you will signal data. Over-
dampening

measured
the later you will signal data. Over-
dampening
the larger the risk per trade.
Simple Breakout Trend Following
Buy Buy
Weekly
Buy
Buy
Sell
Sell
Sell
Giving Noise More Room
Increasing the Trading Frequency
1. Start longer term and move

shorter
2. Don’t trade more than you have to
in order to be effective
Moving to Daily Data
Daily
Buy
Buy
Sell
Stop
Increase The Frequency More
1. Going intraday has a lot of issues

2. Advantages as well
Intraday data
5 minute bars
Buy
Buy
Buy
Sell
Sell
Noise versus Information
Information
Noise
Information
Information
Multiple Trend Following Models
Noise
Filters – Use Carefully!
1. Degrees of Freedom - In statistics,
the number of degrees of freedom
is the number of values in the final
calculation of a statistic that are
free to vary.
enjoytheride.world
14
free to vary.
2. Schwager: Degrees of Restriction
enjoytheride.world
15
free to vary.
2. Schwager: Degrees of Restriction
3. Less is more!enjoytheride.world
16
Filters – Problems in Paradise?
1.You want lots of trades to create
reliable results
enjoytheride.world
17
reliable results
2.The more filters you add, the less trades
you get
enjoytheride.world
18
reliable results
2.The more filters you add, the less trades
you get
3.One trade can be the difference over a
year.
enjoytheride.world
19
Over-Filtering - $100K
End Max Long
Value CAGR Mod- DD est # Win/ Win
Model $K % MAR Sharpe % DD Trades Loss %
Long, Keltner 200D 3X ATR
Stop, Ranked by Highest
ROC 200D 41803 28.9 0.67 1.13 43.2 41.9 340 4.48 40.3
Long, HiVol Filter 20/80, Mkt
Up, Keltner 200D 3X ATR
Stop, Ranked by Highest -
ROC 200D 89382 33.1 0.91 1.16 36.5 26.2 352 4.54 41.2
-
100D Range B/O
11134 22.0 0.59 0.97 37.4 45.8 301 7.40 37.9
-
SMA Cross 30/150
10647 21.7 0.51 0.95 43.0 44.6 308 6.86 38.0
-
RSI 14D>70
9249 21.0 0.48 0.90 43.3 59.9 291 5.86 35.4
enjoytheride.world
20
Position Sizing
Why do it and how to do it?
Successful Traders Size Their Positions – Why and How?

Available: enjoytheride.world, Amazon and iBooks store
Managing Position Risk
• All positions have different risk:
• Risk to stop loss
• Risk of moving quicker or slower = volatility
• Risk of bankruptcy or going to zero
• Risk of news
• Risk of the market
enjoytheride.world
22
• All positions have different risk:
• Risk to stop loss
• Risk of moving quicker or slower = volatility
• Risk of bankruptcy or going to zero
• Risk of news
• Risk of the market
• On a position basis, manage the first two
enjoytheride.world
23
Control position size by risk%/equity
Risk = points from where you are going to get

into a position to where you would stop out
and get out of the trade
Risk % = $risk of 1 unit of instrument/$equity

enjoytheride.world
24
Example:
Equity = $100,000
Percent risk allocation to a single position = 1%
Price to buy XYZ stock = $20.50, Stop loss price on XYZ =
$19.05
Risk $ = 20.50 – 19.05 = $1.45
Position size by risk = ($100,000 X 1%) / $1.45 = 689.66 shares,
rounded down to 689 shares
enjoytheride.world
25
Example:
Equity =$100,000
Percent allocation to a single position = 1%
Price to buy Gold futures in March = $1275, Stop loss on
March Gold = 1267.50
Risk $ = $1275 - $1267.50 = $7.50/contract X $100 per full
point move in gold = $750
Position size by risk = ($100,000 X 1%) / $750 = 1.33
contracts, rounded down to 1 contract
enjoytheride.world
26
Managing Position Volatility
What is volatility?
I use Average True Range (ATR)
ATR = x period avg of (range of price movement)
In days over say 21 days:
ATR = 21 day avg ( max(yesterday’s close,today’s
high)-min(yesterday’s close, today’s low))
enjoytheride.world
27
Example:
Equity = $100,000
Percent risk allocation to a single position = 0.75%
Volatility (ATR(21) of XYZ stock = $0.85
Position size by Volatility = ($100,000 X 0.75%) / $0.85 =
882.35 shares, rounded down to 882 shares
enjoytheride.world
28
Example:
Equity =$100,000
Percent allocation to a single position = 0.75%
Volatility (ATR(21) of Gold futures in March = $3.60
Volatility $ = $3.60 / contract X $100 per full point move in
gold = $360
Position size by Volatility = ($100,000 X 0.75) / $360 = 2.08
contracts, rounded down to 2 contracts
enjoytheride.world
29
Managing The Initial Position
• Which should you use risk or volatility?
enjoytheride.world
30
• How about both?
enjoytheride.world
31
• How about both?
• I always err to the conservative side, so I
always take the smaller position size.
enjoytheride.world
32
Example:
Equity = $100,000
XYZ Stock at $20.50 per share, stop loss order at $19.05 = $1.45 risk
Initial Size by 1% risk allocation = 689 shares
Percent risk allocation to a single position = 0.75%
Initial Size by 0.75% volatility allocation = 882 shares
Minimum of (689, 882) = 689 shares of XYZ for the order
enjoytheride.world
33
Example:
Equity =$100,000
Initial Size by 1% risk allocation = 1 contract
Initial Size by 0.75% volatility allocation = 2
contracts
Minimum of (1, 2) = 1 contract of gold for the
initial position
enjoytheride.world
34
Managing Ongoing Risk
• Your job doesn’t stop after getting into a
position!
• Everyday ask is this portfolio properly
positioned?
• As position matures conditions can change
enjoytheride.world
35
• Your job doesn’t stop after getting into a
position!
• Everyday ask is this portfolio properly
positioned?
• As position matures conditions can change
• You must change with the changing
conditions!
enjoytheride.world
36
Example:
Equity =$105,000
Price to XYZ stock = $25.00, Stop loss price on XYZ = $21.00
Risk $ = 25.00 – 21.00 = $4.00
Position size by risk = ($105,000 X 1.5%) / $4.00 = 393.75
shares, rounded down to 393 shares
enjoytheride.world
37
Example:
Equity =$105,000
Risk of gold contract currently = $10*100=$1,000
Position size by Volatility = ($105,000 X 1.5) / $1000 =
1.575 contracts, rounded down to 1 contract
enjoytheride.world
38
Managing Ongoing Vol
• Volatility can change dramatically during a
trade
• Not managing the volatility of a position can
allow it to move the portfolio
enjoytheride.world
39
Managing Ongoing Vol
• Volatility can change dramatically during a
trade
• Not managing the volatility of a position can
allow it to move the portfolio
• Trying to keep all the positions contributing
somewhat equally to the results
enjoytheride.world
40
Managing Ongoing Volatility
Example:
Equity = $105,000
Percent risk allocation to a single position = 1%
Position size by Volatility = ($105,000 X 1%) / $1.80 =
583.33 shares, rounded down to 583 shares
enjoytheride.world
41
Managing Ongoing Volatility
Example:
Equity =$105,000
Percent allocation to a single position = 1%
Volatility (ATR(21) of Gold futures in March = $10.00
Volatility $ = $10 / contract X $100 per full point move in
gold = $1,000
Position size by Volatility = ($105,000 X 1%) / $1,000 = 1.05
contracts, rounded down to 1 contract
enjoytheride.world
42
Combining the Ongoing Controls
• Same as initial take the smaller of the two
answers
• I run it every day and look for a “peel off”
trade
enjoytheride.world
43
Combining the Ongoing Controls
• Same as initial take the smaller of the two
answers
• I run it every day and look for a “peel off”
trade
• The answer can get to 0 position size in wild
markets, wars breaking out, currency
devaluations, world panics.
enjoytheride.world
44
Controlling Portfolio Risk
1. Market Math example of adding money
or withdrawing money
enjoytheride.world
45
2. “Lockstep” – What it is and how to deal
with it.
enjoytheride.world
46
2. “Lockstep” – What it is and how to deal
with it.
3. Find a amount of risk level you are
comfortable with if all your positions go
against you enjoytheride.world
47
Portfolio Selection
1. First look for zero correlation
enjoytheride.world
48
Portfolio Selection
2. Next use logic for items that
have nothing to do with each
other
enjoytheride.world
49
Portfolio Selection
2. Next use logic for items that
have nothing to do with each
other
3. Use equity size to determine
what you are able to trade
enjoytheride.world
50
Portfolio Selection - Example
enjoytheride.world
51
enjoytheride.world
52
enjoytheride.world
53
Portfolio Selection
Common sense goes a
long way on Portfolio
Selection!
enjoytheride.world
54
Mental Side of Trading
1.Self-Responsibility is essential
enjoytheride.world
55
2.Self-Awareness is very useful
enjoytheride.world
56
3.Mental States can be changed
enjoytheride.world
57
3.Mental States can be changed
4.Discipline is essential and can
only be achieved using the first
three items
enjoytheride.world
58
Questions for Tom
Tom can be found at:
Twitter: @basso_tom
Facebook: @enjoytheride.world
Educational website for traders:
www.enjoytheride.world
enjoytheride.world
59
TRADING
SYSTEMS ON
RENKO
CHARTS ALGO
PRASHANT SHAH & ABHIJIT CONVENTION
PATHAK - DEFINEDGE 2019 – DAY 1
SOLUTIONS
Trading Systems on Renko chart
Prashant Shah and Abhijit Phatak
www.definedge.com
Usual charts – Two dimensional
Renko Charts
• Origins in Japan
• Introduced by Steve Nison in his book
“Beyond Candlesticks”
• Only price
• Simple
• Objective
Construction
Hindalco: 10 Absolute brick-value
Brick-value
Hindalco: 1% brick-value
Bhel: 5 Absolute brick-value
Bhel: 5% brick-value
Brick-Value
ATR Brick-Value
Understanding of Renko charts
Formation of Bricks
Line chart
Renko chart – 0.25%
Renko chart – 0.50%
Features
• Diagonal plotting
• Every brick is made of two prices
• Either bullish or bearish
• Patterns can be designed with combination of
bricks
• Indicators are calculated on Renko bricks
Moving Average on Line chart
Moving Average on Renko chart
Data frequency is daily
Brick-value is important tool

Minimum time-interval for Data
frequency
Time-frame
Renko brick value - Stocks
• 0.50% log brick: Short term
• 1% log brick: Medium term
• 3% - 5% log brick: Long-term

Price formation
Trend Identification
40-column EMA
Indicators
Back-testing
Nifty 500 Stocks
Since 1995
Back-testing
8 years data on One-minute

timeframe
March 2011 to Feb 2019
Brick distribution
Brick Zone
Position sizing rules to better
the performance
RENKO RS chart
RENKO RS chart
Renko Charts
• Objective price patterns: Entry and Exit levels
• Trading systems using price patterns and
Indicators
• Reduce whipsaws due to construction method
• Reduce trading signals and make them more
productive
• Possibility of Minimum time-interval data
frequency significantly reduces impact cost
• Brick-value is a useful tool
Straddles / Strangles
Nifty
11700 Straddle
11800 Straddle
11900 Straddle
Bank Nifty
29500 Straddle
30000 Straddle
30500 Straddle
Tata Motors
170 Straddle
210 Straddle
Thank You
www.definedge.com
MEAN
REVERSION
STRATEGY
FOR STOCKS ALGO
VISHAL MEHTA - HEAD INDIA CONVENTION
CHAPTER, CMT 2019 – DAY 1
Mean Reversion Trading
System for Stocks
Presented by : Vishal Mehta , CMT
Background : Vishal Mehta, CMT
Mentors & Journey
Trend Following
Ed Seykota
Systematic trading
Subhadip Nandy
Mean Reversion
Larry Connors
Mean Reversion Philosophy
Mean Reversion is a phenomenon that there is an underlying stable trend in the

price of an instruments and prices fluctuate around this trend . Therefore, values
deviating far from the trend will tend to reverse direction and revert back to the
trend.
Markets are trend following in long Term and Mean reverting

in Short Term.
Why Mean Reversion Trading Systems
• Lasts for few hours to few weeks.
• You Buy Low and Sell High or Sell High and Buy Low (Comfortable for
Traders)
• High Win to Loss Ratio
• Smooth equity curve.
• Low drawdown.
• Low Capital Deployment.
Trading Model V/S Trading System
Trading Model is designed to explore a potential market edge in the
market on historical data.
Trading System consists of many parts of trading likes of Number of

Symbols, Entry , Exit , Position Size, Stop Loss, Target Price etc.
Systematic Trader’s primary focus should be able exploit edge in the

market and with proper rules and position size they can exemplify the
returns.
Finding Holy Grail
Source : Piyush Chaudhry @piyushchaudhry

Mean Reversion Trading Systems
Mean Reversion Trading Systems
Mean Reversion Bear Trading
Day Trading RSI System
Rules for Short
Entry :
1) C > 200 SMA
1) RSI(2) > 50
2) Todays % Change > 3%
3) Short at 1% Limit next Day
Exit :
1) Exit at 3.25 PM
Performance Report
Important Numbers :
Net Profit : 13,10,511
Net Profit % : 436%
CAR : 21%
Max DD : 13%
CAR/MDD : 1.64
Back test Period : Sept 2011 – April 2018
Initial Capital : 1,00,000
Leverage : 10 Times
Max Position : 10
Capital per position : 1,00,000
Transaction cost : 0.05%

Equity Curve
Bear Phase Bull Phase Bear Phase Bull Phase

Stop Loss & Profit Target
Stop Loss
• Studies shows Stop Loss hurts the Mean Reversion Strategies
• Various ways of defining Stop loss :

• Risk Based Stop Loss
• Volatility Based Stop Loss
• Technical Stop Loss
• Interesting Way of putting Stop Loss based on Maximum Adverse Excursion

(MAE)
Stop Loss & Profit Target
Avg Winning % MAE : -0.77%
Std Winning % MAE : 0.73%
Avg Winning % MFE : 2.63%

Std Dev Winning % MFE : 1.62%
Ideal Stop Loss : Avg Winning % + 2 Std Winning % MAE

Ideal Stop Loss : 2.23% (Round up 3%)
Ideal Profit Target : Avg Winning %MFE + Std Winning %MFE

Ideal Profit Target : 4.25% (Round up 5%)
Performance Report
Important Numbers : After Stop Loss & Profit Target
Net Profit % : 341%
CAR : 18%
Max DD : 10%
CAR/MDD : 1.82
Important Numbers : Before Stop Loss and & Profit Target

Net Profit % : 436%
CAR : 21%
Max DD : 13%
CAR/MDD : 1.64
Backtest Report
• As they say I have never found a Bad Backtest Report.
• All Systems are output of some kind of optimization either Market
Observations or moving the parameters.
• All Due care is taken while running the system and backtest.
Please make your own due diligence before trading,

Past Performance is no guarantee of future
Performance.
When you think system is broke ?
• How can we differentiate between System Drawdown and System
Breaking
• Max DD if hit 2 times in Real time Trading hints to possible of System
breaking down.
Paradox of System trader
• Helps you to remove emotions from Trading
• Machine will take over man : Tesla car (its man and Machine)
• Itching to improve the performance after viewing the recent

transaction and stop loss
Benefits of System trader
• Back test result gives comfort while trading in live market during drawdown period
• Greatest benefit is to pull the trigger when you can not do discretionary
Successful Trader Vs Novice Trader
Source : Piyush Chaudhry @piyushchaudhry

Thank You
LinkedIn Twitter
Email : mastertrader21@gmail.com @vishalmehta29

Mobile : 9820757038
ANIL GHELANI
RAKESH PUJARA GAURAV RAIZADA VIVEK GADODIA DEVANG JHAVERI
HEAD OF PASSIVE INVESTMENTS
FOUNDER, COMPOUNDING AND PRODUCTS, DSP (MODERATOR)
COO, IRAGE MASTERTRUST CO-FOUNDER, DRAVYANITI
WEALTH ADVISORS INVESTMENT MANAGERS
INVESTMENT MANAGERS LLP CONSULTING CO-FOUNDER, DRAVYANITI CONSULTING
AIF IN INDIA : APNA ALGO CONVENTION

TIME AAGAYA 2019 – DAY 1
PANEL DISCUSSION
OPTION BACK-
TESTING OVER
EXISTING
SYSTEMS ALGO
SHUBHAM AGARWAL – CONVENTION
FOUNDER & CEO, QUANTSAPP 2019 – DAY 1
Option Backtesting
Option Back testing over existing

System
Mr. Shubham Agarwal, CMT, CFA, CQF

CEO, Quantsapp
Content
Section 1: Why Options? Edge over Futures
Section 2: Understanding your existing system with an

Option perspective
Section 3: Approximating the Optimal Strategy
Section 4: Defining Back-testing Parameters
Section 5: Yipee: Back-test your existing system
2
Option V/s Future
Why Options? Edge of Options over

futures
3
Case-study
Let’s Bet on Nifty

If it goes up by 100 points, I pay you Rs.80
If it goes down 100 points, you pay me Rs.20
Call Option is what

you need
4
Long Future vs. Long Option
Long Future Payoff Long Option Pay-off
Unlimited Profit Unlimited Profit
Profit Profit
Call Option is what

Price you need Price
-tive +tive
Price Price
-tive +tive
Limited loss
Loss Unlimited loss Loss
5
Cost Effective
Trade on Nifty
Target: 100 points Stop Loss: 20 points
Case 1: Buy Futures Case 2: Buy Call Option
Target Profit: Rs.7,500 Target Profit: Rs.4,500

Call Option is what
Stop Loss: Rs.1,500 Stop Loss: Rs.1,500
you need
Margin: Rs.51,546 Margin: Rs.1,500
ROI: 14.5% ROI: 300%
6
Cost Effective
Call Option is what

you need
7
Reduced Losses
Trade on Nifty ( Gap Down)

Target: 100 points Desired Stop Loss: 20 points
Actual Stop Loss: 100 points
Case 1: Buy Futures Case 2: Buy Call Option
Target Profit : Rs.7,500 Target Profit : Rs.4,500
Desired Stop Loss : Rs.1,500 Desired Stop Loss : Rs.1,500

Call Option is what
Actual Stop Loss : Rs.7500 Actual Stop Loss you need
: Rs.1500
Initial Capital : Rs. 1 lac Initial Capital : Rs. 1 lac
Margin : Rs.51,546 Margin : Rs.1,500
ROI : -14.5% ROI : -100%

ROI(% Capital) : -7.5% ROI(% Capital) : -1.5%
8
Reduced Losses
Call Option is what

you need
Cut down losses
9
Future Market
Futures comes with power of leverage of ~4X
It amplifies return by 4 times
But….if stock losses 25% , you can

Gives
wipe out entire capital
10
Option-Why?
Edge #1: Options Provide Superior Risk Management Techniques
Increased Return
Call Option is what
you need
Expenses
Expenses
Return
Only Cash Using Options

Increase return to stay in game
11
Why Options?
Take trade for any forecast
Bullish V/s Bearish
Either-ways V/s Oscillation
Ede 2# Take trade for any forecast

Apart from taking Bullish and Bearish
bets options can also help us to trade
in Either-ways market (Volatile
market) or Oscillation Markets with
ease
12
Why Options?
Event Trading
Budget,
RBI Policy, Fed
Meeting
Demonetisation
Event
Trading
Mergers and
Result
Demergers
Edge 3# Event Trading

Options helps to take advantage during Events like Budget, Result season, RBI policy as
we can form strategies to take advantage of heightened volatility in the market.
World over event trading had proved to be high volume activity
13
Why Options?
Make money when market is sideways
Historically
data suggest
market is
consolidating
65-70% times
Edge 4# Make money when market is sideway

With linear characteristics of futures, one can not make money in Consolidating market due to
frequent stop losses however Option’s non-Linear characteristics comes as saviour to generate
money by selling options
14
Why Options?
Participate high risk entry point with
low risk trade
Maruti re-testing its earlier low of 6400 and reversing sharply. Future long
comes with risk of false move. 6700 April CE moved from 200 to 500 in 5
days
Edge 5#Participate in high risk trade

Options helps to participate in high risk entry point by taking low risk trade. If reversal takes place
you make immediate gain while if stock plummets down loss is limited to premium outflow
Isn’t it Powerful?
15
Why Options?
Natural Leverage
Edge 6# Natural Leverage:
Option Provides Natural Leverage

Nifty Future @ 11825 75 Market Value: 886875
Nifty 11800 CE @ 85 75 Total Premium:12750 ( Two lots)
So by buying 2 lot of Nifty ATM Call Option you are taking exposure of Future ( Not exactly
equal …
( Ex :16/4/19)
Leverage: ~65+ times
16
Why Options?
Create own Pay-off
Edge 7# Create your own Payoff:

Options helps to create own pay-off depending on forecast
Bullish Mildly Bullish Extreme Bullish Sideways Eitherways
Bearish Mildly Bearish Extreme Bearish
One can even Repair loss making strategy
17
Why Options?
Example
EXAMPLE
18
Understanding existing
system
Session 2 : Understanding your existing

system with Option perspective
19
Understanding existing system
Trend following System ideal PNL based on trend
Call Option is what

you need
20
Understanding existing System
Evaluating your System
Short term(3- Medium term

Duration Intraday
6 days) (10-15 days)
Call Option is what
you need
Trend
Trend
Based on following
Following Oscillating
Move medium
large moves
moves
21
Now over laying an Option Strategy on Futures
Buy : Long Call

Sell : Long Put
Is it right……..?
22
23
Similarly One Strategy fitting all market condition will not provide great
results
24
Session 3 : Approximating the Optimal
Strategy
25
Optimizing Option Strategy
Do you think Time plays a crucial input in

decision making ?
Call Option is what
you need
26
Strategy differ based on Time left to expiry
Start of expiry Mid of expiry Near Expiry
Long Bull - Call Ratio - Call

Call
Long Bear - Put Ratio-Put

Put
27
Do you think Speed of movement ( Volatility) too

plays a crucial role in decision making?
Call Option is what

you need
28
Strategy differ based on expected speed of

move ( Volatility)
Expecting lower Expecting higher Expecting

moves moves oscillation
Short Long Condor

Straddle Straddle
Short Long
Strangle Iron
Strangle
Butterfly
29
SEVEN WAYS TO
BUILD SUPERIOR
TRADING
STRATEGIES ALGO
LARRY CONNORS – CHAIRMAN CONVENTION
OF THE CONNORS GROUP
(TCG)
2019 – DAY 1
SEVEN WAYS TO
BUILD SUPERIOR
TRADING
STRATEGIES
LARRY CONNORS
Disclaimer:
Connors Research, LLC ("Company") is not an investment advisory service, nor a registered investment advisor or broker-dealer and does not purport to tell or
suggest which securities customers should buy or sell for themselves. The analysts and employees or affiliates of Company may hold positions in the stocks or
industries discussed here. You understand and acknowledge that there is a very high degree of risk involved in trading securities. The Company, the authors,
the publisher, and all affiliates of Company assume no responsibility or liability for your trading and investment results. Factual statements on the Company's
website, or in its publications, are made as of the date stated and are subject to change without notice.
It should not be assumed that the methods, techniques, or indicators presented in these products will be profitable or that they will not result in losses. Past
results of any individual trader or trading system published by Company are not indicative of future returns by that trader or system and are not indicative of
future returns which be realized by you. In addition, the indicators, strategies, columns, articles and all other features of Company's products (collectively, the
"Information") are provided for informational and educational purposes only and should not be construed as investment advice. Examples presented on
Company's website are for educational purposes only. Such set-ups are not solicitations of any order to buy or sell. Accordingly, you should not rely solely on
the Information in making any investment. Rather, you should use the Information only as a starting point for doing additional independent research in order to
allow you to form your own opinion regarding investments. You should always check with your licensed financial advisor and tax advisor to determine the
suitability of any investment.
HYPOTHETICAL OR SIMULATED PERFORMANCE RESULTS HAVE CERTAIN INHERENT LIMITATIONS. UNLIKE AN ACTUAL PERFORMANCE
RECORD, SIMULATED RESULTS DO NOT REPRESENT ACTUAL TRADING AND MAY NOT BE IMPACTED BY BROKERAGE AND OTHERSLIPPAGE
FEES. ALSO, SINCE THE TRADES HAVE NOT ACTUALLY BEEN EXECUTED, THE RESULTS MAY HAVE UNDER- OR OVER-COMPENSATED FOR THE
IMPACT, IF ANY, OF CERTAIN MARKET FACTORS, SUCH AS LACK OF LIQUIDITY. SIMULATED TRADING PROGRAMS IN GENERAL ARE ALSO
SUBJECT TO THE FACT THAT THEY ARE DESIGNED WITH THE BENEFIT OF HINDSIGHT. NO REPRESENTATION IS BEING MADE THAT ANY
ACCOUNT WILL OR IS LIKELY TO ACHIEVE PROFITS OR LOSSES SIMILAR TO THOSE SHOWN.
Connors Research, LLC

185 Hudson Street, Ste 2500
Jersey City, NJ 07311
Copyright © The Connors Group, Inc., 2019.

Larry Connors
Laurence Connors is Chairman of The Connors Group (TCG), and the principal executive officer of Connors Research LLC.
TCG is a financial markets information company that publishes daily commentary and insight concerning the financial markets
and has twice received an award by the Entrex Private Company Index for being one of the 10 fastest growing private
companies.
He has over 35 years of experience working in the financial markets industry. He started his career in 1982 at Merrill Lynch as
an Investment Advisor, and later moved on to become a Vice President with Donaldson, Lufkin, Jenrette (DLJ), where he
worked with the Investment Services Group from October 1990 to March 1994. Mr. Connors is widely regarded as one of the
leading educators in the financial markets industry. He has authored top-selling books on market strategies and volatility trading,
including Short-Term Trading Strategies That Work, and Street Smarts (with Linda Raschke). Street Smarts was selected by
Technical Analysis of Stocks and Commodities magazine as one of “The Classics” for trading books written in the 20th century.
His most recent book Buy The Fear, Sell The Greed; 7 Behavioral Quant Strategies For Traders was published this past
summer.
Mr. Connors has been featured and quoted in the Wall Street Journal, New York Times, Barron’s, Bloomberg TV & Radio,
Bloomberg Magazine, Dow Jones Newswire, Yahoo Finance, E-Trade Financial Daily, Technical Analysis of Stocks and
Commodities, and many others. Mr. Connors has also been a featured speaker at a number of major investment conferences
over the past two decades.
INTRODUCTION | LARRY CONNORS AND CESAR ALVAREZ
The 200 Day Moving Average Is Your
Best Friend!
Larry Connors
Short Term Trading Strategies That Work
Buy Pullbacks, Not Breakouts!
Larry Connors
Longer Term Prices Trend
There are so many great academic studies, along with successful money management firms
to confirm this. To go further, read Gary Antonacci’s book - “Dual Momentum”
Larry Connors
Beware of Strategies With
Too Many Trading Rules!
Larry Connors
Data Integrity Is Your Lifeblood.
Trust, But Verify!
Larry Connors
Trade Construction and Portfolio
Construction Are Vital To Your Success!
Larry Connors
SEE THE TRADE, TAKE THE TRADE!
Larry Connors
Buy The Fear, Sell The Greed
The 7 Rules To Build A Superior Trading Strategy
1. The 200 Day Moving Average Is Your Best Friend!
2. Buy Pullbacks, Not Breakouts! - (Oversold Short Term Equity Prices
Mean Revert)
3. Longer Term Prices Trend!
4. Beware of Strategies With Too Many Trading Rules!
5. Data Integrity Is Your Lifeblood. Trust But Verify!
6. Trade Construction and Portfolio Construction Are
Vital To Your Success!
7. See The Trade, Take The Trade!
Larry Connors
Special Offer
If you're looking to improve your programming
and testing skills and apply the same
techniques used by the majority of the major
trading firms in the world, I invite you to attend
a free webinar I'm holding next week.
We teach a 5 week course which promises to

have you programming trading strategies in
Python. If you'd like to learn more, please sign
up here. If you cannot attend live, a recording
of the webinar will be emailed to you.
http://tradingmarkets.com/python-webinar-sign-up
Special Offer
Pre-order the book “The Alpha Formula”
by Chris Cain & Larry Connors.
Pre-order by April 30, 2019 and receive a

50% discount ($49.95/$24.95).
https://tradingmarkets.com/store/The-
Alpha-Formula-Pre-Order-p135397775
Use code ALGO1 to order.

Contact Me
Larry Connors
lconnors@connorsresearch.com
APPLYING
PROBABILITY
THEORY AND
GAME THEORY TO
TRADING AND
INVESTING ALGO
HIMANSHU PATIL – CO- CONVENTION
FOUNDER, ARDENT 2019 – DAY 1
GAME THEORY AND ITS
APPLICATION
Game theory and its application in Poker.
Possible Adaptation to Equity Markets.
 Game theory is the process of modeling the strategic interaction
between two or more players in a situation containing set rules
and outcomes.
 Originally, it addressed zero-sum games, in which one person's
gains result in losses for the other participants.
 It has applications in all fields of social science, as well as in logic
and computer science.
 Today, game theory applies to a wide range of behavioral
relations, and is now an umbrella term for the science of logical
decision making in humans, animals, and computers.
WHAT IS GAME THEORY??

HISTORY OF GAME THEORY
• Game theory did not really exist as a unique field until John
von Neumann published the paper On the Theory of
Games of Strategy in 1928
• In 1950, the first mathematical discussion of the prisoner's
dilemma appeared, and an experiment was undertaken by
notable mathematicians Merrill M. Flood and Melvin
Dresher, as part of the RAND Corporation's investigations
into game theory. RAND pursued the studies because of
possible applications to global nuclear strategy.
• Around this same time, John Nash developed a criterion for
mutual consistency of players' strategies, known as Nash
equilibrium, applicable to a wider variety of games than the
criterion proposed by von Neumann and Morgenstern. Nash
proved that every n-player, non-zero-sum (not just 2-player
zero-sum) non-cooperative game has what is now known as
a Nash equilibrium.
 Connect Four (James D Allen)
 Tic-Tac-Toe or Naughts and Crosses
 Checkers (Jonathan Schaeffer and his team)
 Checkers is till date most complex Game Yet Solved.
 Any Game that involves Chance cannot be Solved.
EXAMPLES OF SOLVED GAMES

BRANCHES OF GAME THEORY
 Lets Watch the Video to understand the Basics of poker
 https://www.youtube.com/watch?v=JOomXP-r1wY
POKER: THE BASICS

 Poker is a Zero Sum Game (Ignoring Rake)
 Poker Models are a Combination of behavioral relations and
Probability
 Most Common used Technique to Start GTO in Poker is using Nash
Equilibrium Strategies.
 GTO Poker in Theory is Unexploitable i.e. It guarantees a minimum
Win rate.
GAME THEORY IN POKER ( GTO

POKER)
Poker is solved. GTO solved
Strategies Exist for Poker.
Pros Play only GTO.
There is a Formula to Play GTO
Poker.
MYTHS ABOUT GTO IN
POKER.
We will
Explain
it More
in
Equilab
GENERAL HAND STRATEGY IN ONE

PICTURE
 Now lets consider some Hand Examples where Game theory can
Help us take better Decisions.
 Following Slides have examples of tough decisions made by me
Correctly
 Hand 1:
http://www.pokerhandreplays.com/view.php/id/9559427
 Hand 2:
 Hand 3:
EXAMPLES OF GTO
BASIC SPOTS
 http://www.pokerhandreplays.com/view.php/id/9559444
(Hand1)
 http://www.pokerhandreplays.com/view.php/id/9559450
 As seen above, sometimes a hand will be played right and yet

lose.
SOMETIMES U PLAY RIGHT AND YET

LOSE
SKILLS PYRAMID
Different Levels in Poker
 Assuming having spent time in learning fundamentals and having
good grasp of basic knowledge, the minimum time required to be a
pro should be
 Level 200(bb2), Start BR: 4000,End BR:10000, Hours Required: 300
 Level 500(bb5), Start BR: 10000, End BR: 25000, Hours Required: 300
 Level 1000 (bb10), Start BR: 25000, End BR 75000, Hours Required: 500
 Level 2500 (bb25), Start BR: 75000, End BR 200000,
Hours Required: 500.
 Total Hours = 300+300+500+500 = 1600.
 Assuming 8 hours a day, it would take 200 days to become an Entry
Level Pro in Poker on Indian sites. ( 4 tabling and 5 bb/100 WR).
HOW LONG DOES IT TAKE TO BE

ENTRY LEVEL PRO IN POKER
 Zero Sum Games( Ignoring Rake and Brokerage)
 Decisions Made by participants are based on incomplete
information
 Profitability exists on exploitation of mistakes by a Group of opponent
population
 Strategies Revolve around a Chance(Probability) of Occurrence of
a particular event or set of events
 Mathematical Models work on large sample
 Easy to enter, Hard to practice with proficiency
 Variance
SIMMILARITIES BETWEEN POKER AND

TRADING
 Improper Bankroll
 Inability to adapt with maturing markets
 Weak Fundamental Knowledge
 Indiscipline in pressure situations
 Impatience to overcome Variance
 Grossly overestimating their skill
TOP REASON FOR FAILURE IN POKER

AND TRADING
SKILL DEVELOPMENT GRAPHS
POKER LEVEL OF THINKING
 Asset pricing
 Mergers and acquisitions (M&A)
 Capital structure
 Corporate governance
APPLICATIONS IN FINANCE
 Game theory can function as a great tool (Filter) to aid the basic
Fundamental strategy that individual understands and employs
(Technical, Fundamental, macro, mixed etc)
 Mathematical aspect can help create mixed strategies which
can function under all weather portfolios (Extremely complicated
and high level)
 Any Direct Game theory Model to Investment does not exist
(atleast to my knowledge and I can be wrong)
APPLICATION OF GAME THEORY IN

INVESTING
 Game Theory Optimal (GTO) in Poker is applied using Nash
Equilibrium
 Equity Markets can never be in Equilibrium, so Nash Equilibrium
Cannot Be used to Markets
 Adjustments in Poker are made on basis of opponent tendency,
which can be different for different people. But individual
character is a dominant theme in selection of strategy
 Individual Character is virtually non-existent, Movements in
Markets is a result of Dominant strategy of entire population. Also
since the Number of opponents in Markets are so wide, prudence
in ignoring individual characteristic is optimal
DIFFERENCE IN APPLICATION OF GAME

THEORY IN MARKETS AND POKER
 Keynesian and Post-Keynesian economists, and their under-
consumptionist predecessors criticize general equilibrium theory
specifically, and as part of criticisms of neoclassical economics
generally.
 Specifically, they argue that general equilibrium theory is neither
accurate nor useful, that economies are not in equilibrium, that
equilibrium may be slow and painful to achieve, and that
modeling by equilibrium is "misleading", and that the resulting
theory is not a useful guide, particularly for understanding of
economic crises
CRITICISM OF EQUILIBRIA
Factors Cause Implications
Steep Discount in Price NSEL Default Change in Management CASE STUDY 1: MCX
(2014)
Fall in Revenue CTT Negative
Provided the fundamental MCX was leading Bullion trading

Valuation Discounted recover post crisis company in 2013 trading at 1550+
price towards end of 2012.
Market Reaction Panic Over Correction to scams
2013 saw a scam that impacted its
No possible increase in revenues (NSEL Default) which
Industry reaction Protected Sector Competetion
triggered a extreme fall in
Price at bargain due to revenues and hence share price
Likelihood of recovery Medium to High Recovery Risk (about 89% peak to trough)
Min 2 to 1 Max 4 to 1 over 5
R/R - years ( at about 450)
Factors Cause Implications
Fall in price Failure of Buy back Loss of confidence

Standard procedure to CASE STUDY 2:
Failure of Buyback oppossed by Lender oblige to lender request
Likely Below value due to PC JEWELLERS (2018)
Price reaction analyis Extreme overreaction
impact on fundamentals Likely Nil Should help price move up • A normal Jeweller stock with
Normal Fundamentals with
Management Reaction High Debt Focus on Debt reduction decent track record
additional data quaterly reports look promising • Quoting over 400 at start of 2018
RR - min 3 to 1, Max 5 to 1 • Lost 75% of its share price.
Brokerage reports - recommend Buy
 Creating portfolio of Mixed Technical Strategies in Order to have
a Minimum Win ratio at markets in all phases (All weather)
 Applying Game theory For strategic acquisition of
customers(Investors)
 Many More
OTHER POSSIBLE IDEAS WORTH

EXPLORING
USING ALGOS TO
TIME THE MARKET ALGO
ANUPAM SINGHI – CEO, WILLIAM CONVENTION
O’NEIL INDIA
2019 – DAY 2
April 2019
CAN SLIM Strategy
OUR INVESTING METHOD
• The CAN SLIM method developed by Mr. O’Neil is based on profiling the best performing stocks of all
time and identifying common characteristics between them. The result is a system that blends
fundamental, technical, and market timing elements.
• Using these principles, we have developed a group of proprietary metrics that identify the best stocks in
the best industry groups. Our goal is to own the truly exceptional companies of a market cycle that are
capable of large returns.
• The result is a growth-oriented approach that emphasizes companies with outstanding earnings growth,
strong relative price strength, impressive returns on capital and stock charts showing patterns of
accumulation consistent with our historical precedents.
1965 – FIRST SET OF COBOL PROGRAMMERS STARTED
WORK IN CREATING THE FIRST SET OF ALGO’S
ONEIL BY THE NUMBERS
70,000 350 70 54 50
Global equities, Major Countries covered Years serving the Research analysts
funds and indices institutional by our institutional covering
in our proprietary clients around the international investor international
database globe database community markets
PRODUCTS
O'Neil Capital
Management is a private
fund management
company with investment
interests in real estate,
printing, digital media,
newspaper, brokerage,
investment advisory, and
information technology
services.
William J. O’Neil
"The best way to determine the direction of the market is to follow,

interpret, and understand what the general market averages are doing
every day. This is the most important lesson you can learn. Don't let
anyone tell you that you can't time the market.”
WHAT IS MARKET TIMING?
Market timing is the strategy of making buy or sell decisions of financial

assets (often stocks) by attempting to predict future market price
movements. The prediction may be based on an outlook of market or
economic conditions resulting from technical or fundamental analysis. This
is an investment strategy based on the outlook for an aggregate market,
rather than for a particular financial asset.
Source:Wikipedia
STOCK SELECTION ALLOCATION
MARKET TIMING
EXECUTION
THE IMPOSSIBLE DREAM? TIMING THE MARKET
Trying to time the market” is the #1 mistake to avoid. People that think
they can predict the short-term movement of the stock market — or
listen to other people who talk about (timing the market) — they are
making a big mistake,”
-Warren Buffett
"Far more money has been lost by investors preparing for corrections or
trying to anticipate corrections than has been lost in the corrections
themselves.“
-Peter Lynch
BUY AND HOLD
BEST ARE TIMING THE MARKETS
• "The best way to determine the direction of the

market is to follow, interpret, and understand what
the general market averages are doing every day. This
is the most important lesson you can learn. Don't let
anyone tell you that you can't time the market.” William J. O’Neil
• Berkshire is holding over $109 billion in cash? The last time Buffett held
that much of Berkshire’s assets in cash was in the years leading up to the
financial crisis from 2003 to 2007
• Advancement in computing and predictive analytics

MARKET TIMING APPROACHES
Non-financial Indicators
 Spurious Indicators (Whether the NFC or AFC team wins the Super Bowl)
 Feel Good Indicators (Hemline index)
 Hype Indicators-“cocktail party chatter”
Financial Indicators
 Past Prices - (January Indicator)
 Trading Volume(selling climax, put-call ratio)
 Changes in Volatility and future returns
 Valuations
 Normal Ranges (Mean Reversion)
 Short term Interest Rates
 Business Cycles
 Intrinsic Value Models
M FOR CAN SLIM
• Begins with Follow-Through Day • Distribution days contained within 4-6
• Leaders act very strong • Overhead Supply increases
• Distribution Days contained within 3-4
• Index holds its bottom for at least 3 days • Index loses key support levels
• Follow-Through Day is awaited • Distribution days pile up over 6
MAJOR PARAMETERS USED FOR MARKET
CONDITIONS
Factors used to identify Tops

• Distribution Day
• Threshold breaches
Factors used to identify bottoms

• Follow Through day
• Past X days high
Understanding Distribution Days
• Down day on major

index (loss of 0.2% or
more)
• Higher volume
D
• Distribution Count D
Understanding Follow-Through Day
• Strong up day on a
major index (gain of
1.5% or more)
• Higher volume
• Strong action in leading
stocks F F
DISCREET TO CONTINUOUS
-0.2% VOL
x 2 + bx +c =DD
DDCOUNT
DD REMOVE AFTER X DAYS
1/n Everyday
CHAMELEON US : BACKTEST
CHAMELEON US : LIVE PERFORMANCE
CHAMELEON INDIA ALGORITHM: BACKTEST RESULTS
CHAMELEON INDIA: YEARLY PERFORMANCE
Year Chameleon Nifty return Excess return
2001 26.8% -20.3% 47.1%
2002 22.4% -1.6% 24.1%
2003 22.4% -13.4% 35.8%
2004 93.3% 81.1% 12.2%
2005 54.6% 14.9% 39.8%
2006 76.9% 67.1% 9.8%
2007 8.4% 12.3% -4.0%
2008 86.0% 23.9% 62.1%
2009 121.5% -36.2% 157.7%
2010 113.5% 73.8% 39.7%
2011 21.0% 11.1% 9.9%
2012 29.6% -9.2% 38.8%
2013 24.7% 7.3% 17.4%
2014 37.1% 18.0% 19.1%
2015 24.3% 26.7% -2.4%
2016 10.1% -8.9% 19.0%
2017 23.3% 18.5% 4.7%
2018 5.8% 10.2% -4.4%
2019 16.6% 12.10% 4.5%
Portfolio starts 12 Jun 2006 and ends 13 Mar 2019
CHAMELEON INDIA: DRAWDOWN STATISTICS
Benchmark
Rank Start Date End Date Portfolio Drawdown Duration (Days)
Drawdown
1 Mar 5, 2001 Nov 26, 2001 -23.60% -32.80% 266
2 Dec 6, 2006 Jul 12, 2007 -22.40% -10.90% 218
3 Nov 11, 2008 Apr 2, 2009 -19.40% -13.10% 142
4 May 23, 2006 Aug 30, 2006 -18.40% -17.70% 99
5 Jan 14, 2004 Jul 22, 2004 -17.90% -29.90% 190
6 Jul 7, 2011 Dec 2, 2011 -17.30% -17.80% 148
7 Mar 8, 2005 Jul 26, 2005 -17.10% -12.30% 140
8 Jul 24, 2007 Oct 10, 2007 -16.40% -11.80% 78
9 Oct 30, 2000 Feb 23, 2001 -16.00% 0.00% 116
10 Oct 23, 2015 Mar 21, 2016 -14.60% -16.00% 150

CHAMELEON INDIA
CHAMELEON US : BACKTEST
CHAMELEON US : LIVE PERFORMANCE
Questions?
Discipline from Data
Machine Learning in the Stock Market
Patrick Erickson
Director of Data Science
2019-04-21 patrick.erickson@ocmcapital.com
Machine Learning
Machine learning is a set of tools and techniques for finding patterns in data with
minimal human guidance.
Two main types:

• Supervised – Model learns to predict some target variable
• Unsupervised – Model extracts patterns, with no regard for prediction
Machine Learning is Not New
• K-Means Clustering – 1957
• Logistic Regression – 1958
• Artificial Neural Networks – 1958
• Hidden Markov Models – 1959
• Support Vector Machines – 1963
Machine Learning Has Matured
• It works in the real world
• It’s practical and fast
• Barrier to entry is low
An Algorithm in 4 Steps
1. Acquire data
2. Clean, normalize, and abstract
3. Learn to make useful predictions
4. Build a decision framework
Challenges
• Noise
• Messy data
• Competition
• Truth is a moving target
Why Machine Learning?
• Disciplined prediction
• Discover less-obvious patterns
• Improvements are repeatable
• Build and adapt faster
An Algorithm in 4 Steps (ML)
1. Acquire data (NLP)
2. Clean, normalize, and abstract (K-means, Gaussian Mixture)
3. Learn to make useful predictions (Neural Net, SVM, HMM, Regression)
4. Build a decision framework (Reinforcement Learning)
Financial Markets Are Special
• The markets are complex adaptive systems, full of feedback loops and non-
linear effects.
• Outliers are everywhere and sometimes they are the only things that matter.
• Date and time matter. You cannot take a random sample in the usual manner.
• Your training examples are not independent!
What Really Matters
• Build for out-of-sample performance
• Knowing the true strength of your predictions is just as important as making
good predictions
• The pipeline is more important than the model
• Some types of errors are more important than others
• Use the right tool for the job
Questions?
Patrick Erickson
Director of Data Science
2019-04-21 patrick.erickson@ocmcapital.com
PCG Proposition
1
About us – Our Reach
• One of India’s leading broking houses and a subsidiary of Kotak Mahindra Group
• The first broking house to be backed by a bank
• Largest branch network amongst all broking houses in the country
• Our numbers speak :
- 1,325 branches including the franchisee network
- Present in 350 + cities
- 13 Lakh customers trust us as a broking partner
- Manage 6 Lakh + trades in a day
Nos. as on March 31, 2018

Technology Platforms
Get access to multiple trading platforms that enhance trading experience of every customer. Below
are the various platforms that our customers can use as per their convenience:
Kotak Stock Trader KEAT Pro X – High Website – To TradeSmart Terminal

– Mobile App – Speed Online Trade - login delivers streaming
Stay Connected Trading Platform through Kotak news and analysis,
with Markets Securities Website with trade insights,
anytime . and includes most
www.kotaksecuriti
powerful trading
es.com
tools — all at a single
destination
Product Offerings
Free Intraday Trading (4,999):
- Annual subscription fee Rs.4,999, renewable yearly
- Zero account opening fee
- Low brokerage on delivery trades – 0.12%
- Access to research on website
- Trade through multiple platforms
- Minimum Margin – 25L – Cash or 50L Cash/Stock/Both
Free Intraday Trading (2,999):
- Annual subscription fee Rs.2,999, renewable yearly
- Zero account opening fee
- Low brokerage on delivery trades – 0.29%
- Access to research on website
- Trade through multiple platforms
- Minimum Margin – 10L – Cash or 20L Cash/Stock/Both
Happy Day Trading: 100% off on brokerage in online trading on certain days in the month. For details refer
goo.gl/ZvYHzc
Options Trading – Kotak Advantage
Pricing Leverage Platform
Super multiple Online –

Free Intraday
options – Nifty / Trade Smart
Trading
Bank Nifty KEAT Pro
Limit of 10x for Offline –

buying options &
Customized plans ODIN with
70x for selling
options Dealers Assistant
Offers & Value Added Services
Super Multiple for Options:

- Get a higher multiple on Nifty and Bank Nifty Options
- Get up to 10x for buying options and 70x for selling
- It helps to limit loss through a Stop Loss Order.
- All Super Multiple orders are "auto-square off" orders with a Stop Loss Price which the
customer can define. The position will be automatically squared off as a market order at
3.10 P.M.
Custom Algo Development
Custom Development
HNI / Professional Traders Chart connectors
Customized Algo
Convert signals to orders
development
Technical based strategies Position management

Execution Algo Suite
The system allows the user to execute slice order method for financial trading :-
Bidding With Timer

Specify the start time at which order generation will start.
Time Interval (TWAP)

Will slice your orders in particular time slice and fire the orders accordingly. It is a trading algorithm based on weighted average
price used for execution of bigger orders without excessive impact on market price.
Participation of Volume ( POV )

Will participate in % terms with respect to volume
Liquidity Seeker
Will place IOC orders in the market when your desired price buyer/seller will arrive
Jobbing Strategy: Will automatically place Buy/ Sell orders with respect to parameters placed in Initial difference and continuous
difference tab
For eg. : Assume value 2 is placed in Initial diff tab and 5 is placed in continuous diff tab. If price is trading at Rs 100, then Algo will
automatically buy at 98 and sell at 103 continuously
Dynamic Delta Hedging: Will Hedge your delta in terms of absolute and % change in underlying
Custom Algo Development
Understand client requirements
Algo development
Get exchange approvals
Train client on usage
Go Live
Typical Time Frame – 30 to 90 days

Chart Connectors
• Provide Order Management Engines to clients

• These engines convert signals into orders from third party tools like
• Excel
• Charting Tools, etc.
• Eliminate the need to manually track signals and place orders
• Once placed, each order can be manually modified & cancelled
• Order types
• Limit/Market
• Single Leg/Bracket/Cover, etc.
Typical Time Frame: One time development of about 45 to 60 days

Thank You
MOVERS
CRUDE OIL
TRADING
SYSTEM ALGO
ALOK DHARIA – FOUNDER, NAV CONVENTION
FINSTRAT & DR.PRASHANT
MULLICK – CO-FOUNDER, MOVERS 2019 – DAY 2
CRUDE OIL
ALGO TRADING
MOVERS QUANT
ALOK DHARIA
PRASHANT MULLICK
MOVERS Quant | Algo Convention Apr-2019

AGENDA
CRUDEOIL BASICS PREPROCESSING
Types and Uses for crude Data Collection and Storage - MYSQL
Types of Spread Trading Resampling - Pandas/Python
Term Structure of Crude and products EDA - Exploratary Data Analysis - Plotly
Limitations to trading structures
MODEL BUILDING MOVERS

Types of Trading Models Momentum & Volatility Enhanced
Trend Identification and Entry/Exits Returns System
Parameter Identification Multi Asset Class - Multi Time Frame
Model Testing Automated Trading System
Optimization of parameters
Performance Metrics

WHAT IS CRUDEOIL?

Demand For Crude Oil
TRANSPORTATION ECONOMIC DEMAND
Movement of Goods and People Plastics and Petrochemicals
Gasoline, JetFuel, Diesel, Residual Oil Naphtha ad LPG
WEATHER DEMAND POWER STACK

Heating Demand for Homes Power Generation
Propane, HeatingOil, LNG Diesel, Residal Fuel oil

TYPES OF CRUDE OIL

DIFFERENT TYPES OF SPREAD TRADES
LOCATIONAL ARBS REFINING SPREADS

WTI-BRENT (Quality differential + Gasoline Crack or Diesel Crack
Transporation costs) Spread between end product and raw
material
TIME SPREADS QUALITY/BTU

WTI first month - Second Month SPREADS
Storage Economics
WTI-WTS (Same Location but difference
Term Structure Analysis
in quality of the two crudes based on
product slate)
WTI-NG (Based on BTU Content)

Term Structure - WTI

Term Structure - Diesel

Term Structure - Gasoline

Term Structure - GasCrack

Term Structure - HeatCrack

Term Structure - WTI-Brent

MCX WTI L1-L2

Distribution of L1-L2

MCX Spread Vs US INV

Types of Algo Trading Models
• Low Latency - High Frequency Trading Systems
• High Latency - Rules Based Trading Systems
• Intraday Trading Models
• Swing Positional Trading Models
Add a little bit of body text
• Trend Following Momentum

• Mean Reversion
• Sideways
• Volatile
• Non-Volatile
Algo Model Selection
• Speed of Execution - Technology, Capital
• Cost Structure - High Frequency Higher cost
• Intraday Vs Positional - Capital Available for trading
Add a little bit of body text
and Risk Management

• Style of Trading - Contrarian or Trend Following

Data - Preprocessing
Intraday
1,3,5
Positional
15, 60, Daily

Data - Resampling Code

EDA - Plotting Code using Plotly Python

EDA - MCX Crude L1 60 min

DONCHIAN CHANNEL
• Donchian channel - highest high and lowest low of the
last n periods
• Donchian channel used for measuring the volatility of price
• Non-Volatile Stable Prices -Add
Donchian
a channel narrow
• Volatile Unstable Prices - Donchian
heading channel wider
• Price > highest n periods high - long
• Price < lowest n periods low - short
* https://en.wikipedia.org/wiki/Donchian_channel

Donchian Channel Calculation Code

EDA - MCX Crude Donchian Channel

Algo Code Entry

Algo Code Exit

Algo Code Output

Performance Metrics Code

Equity/Drawdown Curve and Metrics

Model Testing & Optimization
INSAMPLE/OUTSAMPLE TESTING
Use Train Data for a specific period to test model on test data
Use trending phase to model and sideways phase to test
Rolling Period model building and testing
OPTIMIZATION
Design of experiments using combination of different parameters
CURVE FITTING
Generic Rule - Is the behavior likely to change and if so can you explain the change in behavior?
Answer No - curve fitting

MOVERS
QUANTITATIVE TRADING
MOmentum Long/short directional trading
Volatility Reactive to market volatility
Enhanced Leveraged trading

Returns Primary focus on risk
System Automated trading system

MOVERS
INTRADAY INDICES POSITIONAL INDICES

Nifty / BankNifty intraday long/short Nifty / BankNifty futures & options
futures & options trading positions taken overnight
POSITIONAL EQUITY POSITIONAL COMMODITIES

Single Stock futures & options positions Commodity futures positions taken
taken overnight overnight

MOVERS Positional Commodities
COMMODITIES POSITIONAL
Seven most liquid and widely traded Positions initiated at market open and
instruments on the MCX. unwound 15 minutes before market close
Gold, Silver, Crude, Copper, Lead, Trading Window: 10AM to 11:30PM
Nickel, Natural Gas
LONG/SHORT SYSTEM DRIVEN

FUTURES All trading signals generated by our
proprietary algorithm and traded using
Liquid contracts of either the full size or
automated order firing to the terminal
commodity minis selected for trading
using near month futures

COMMODITY CONTRACTS TRADED
Gold (Gold Mini Contract Size: 100 gms)

Silver (Silver Mini Contract Size: 5 kgs)
Crude Oil (Crude Oil Contract Size: 100 bbls)
Copper (Copper Contract Size: 1 metric ton)
Lead (Lead Mini Contract Size: 1 metric ton)
Nickel (Nickel Contract Size: 250 kgs)
Natural Gas (Natural Gas Contract Size: 1250 mmBTU)

SUPPORTED PLATFORMS/BROKERS
ZERODHA KITE CONNECT API

Sharekhan TRADETIGER API
KOTAK Securities ODIN Client API
UPSTOX UPSTOX API

SYSTEM MANAGEMENT
Data Capture
- Multiple Data Sources / Automated Switching
Signal Generation
- Real-Time Notifications on Telegram/Email
Execution:
- Manual Monitoring of Executed Trades
- Real Time Automated Notification of Errors During
Execution.
MOVERS
MOmentum Long/short directional trading
Volatility Reactive to market volatility
Enhanced Leveraged trading

Returns Primary focus on risk
System Automated trading system

POSITION SIZING / RISK
Fixed ₹ Risk per Trade

Lots
Position Size (Lots) proportional to Inverse Volatility
System dynamically calculates size with market volatility

Volatility

Commodities Backtest - 2011/18
7 Commodities: Gold, Silver, Copper, Crude, Lead, Nickel & Natural Gas
Notional Capital: ₹ 3,50,000 per commodity
Fixed ₹ Risk per Trade
Each commodity traded on its own Notional Capital
System dynamically calculates size with market volatility
Entry/Exit signals as per MOVERS system logic

Backtest Results – 2018



Meet the Team
MOVERS QUANT GROUP
ALOK DHARIA PRASHANT MULLICK ANAND DHARIA

@alok_dharia @vohicapital @navfinstrat

Let's talk!
WAYS TO REACH US
PHONE
+91 97690 80842
EMAIL / TWITTER
moversquant@gmail.com / @moversquant
OFFICE
F49-50, Moongipa Arcade, D.N.Nagar, Andheri West,
Mumbai, Maharashtra 400053
• Highest rated trading app on Playstore.
• High Reliability.
• Giga Trading Engine helping all the way!
• Completely Personalized experience.

Web.stocknote.com
{ API }
{ API }
Web Mobile API
High Reliability – Separate servers for each platform

No Platform Dependency – Just Plug & Play
Dedicated API Support Team
Pricing
Stocknote Trade API Stocknote Data API

Rs. 2000/- Per month Rs. 2000/- Per month
Get Cashback of Rs.4000/- per month on brokerage for 6 months.
For Algo Convention attendees

Completely Free for first 20 registrations.
IMPROVISING
RETURN/RISK
RATIO OF A SYSTEM
USING MACHINE
LEARNING ALGO
HRISHABH SANGHVI – FOUNDER, CONVENTION
HL INVESTRADE, FOUNDER, RODEO
& FOUNDER, ARQUE 2019 – DAY 2
Machine Learning
for Traders
Hrishabh Sanghvi
1 Avoiding Losers
Which system would you prefer?
What is the ML Filter really doing?
◉
Backtest Stats
◉
Which system would you prefer?
Avoiding biases – Look-ahead
Avoiding biases – Hyper Parameters
What happens behind the scenes?
What happens behind the scenes?
Reality check?
Reality check?
Reality check
◉
Let’s just size the positions
Let’s just size the positions
For Naysayers
2 Machine Learning
Algorithmic Trading
Machine Learning
ML ‘learns’ & gets ‘smarter’
Machines are smarter and better
than humans
“
Machine learning can predict
previously unseen events, a.k.a.
“black swans”
“
The more data you have, the more
likely you are to hallucinate
patterns
“
AI is only for the technology elite
“
Do-It-Yourself
3 Machine Learning
Machine Learning is Powerful
Many ways to screw up ML
“
Components of a
4 Trading System
How to trade?
Advantage
◉ ◉
◉ ◉
◉
An Institutional Trading System
Arque.Tech - Different by Design
Services
◉ ◉
◉
Services
◉
◉
◉
The Team
Thanks!
Any questions ?
◉
◉
A QUANTITATIVE
ANALYSIS OF
VOLATILITY
BREAK-OUTS ALGO
SHUBHADIP NANDY – CONVENTION
INDEPENDENT QUANTITATIVE
TRADER 2019 – DAY 2
A quantitative
analysis of
volatility breakouts
2
Hello!
I am Subhadip Nandy
A full-time independent quantitative derivatives trader.
You can find me at @SubhadipNandy
3
What is quantitative
analysis ?
Quantitative analysis (QA) is a technique that seeks
to understand behavior by using mathematical and
statistical modeling, measurement, and research.
Quantitative analysts aim to represent a given

reality in terms of a numerical value.
4
Any form of quantitave analysis starts

with market observation.
Today we will try to check whether an

intraday move of the Open to Close has
any significance in predicting future
direction of an underlying
5
To be a bit more specific
⊸ How much do we move intraday

⊸ What move is statistically
significant
⊸ Does this move predict future
direction
Lastly, can the observations drawn

be used in devising a trading system.
6
OPEN to CLOSE
If Close < Open - Negative Day
If Close > Open - Positive Day
We plot this difference in percentage terms

7
Open – Close | Running Chart

8
Let’s review some concepts –

Standard Deviation
A quantity expressing by how much the members of a group differ from
the mean value for the group.
How to calculate the standard deviation of a set of numbers:

⊸ Work out the Mean (the simple average of the numbers)
⊸ Then for each number: subtract the Mean and square the result.
⊸ Then work out the mean of those squared differences.
⊸ Take the square root of that and we are done!
9
Normal Distribution and SD

10
Mean +/- 1.5SD = 84.1% of all

occurences
11
12
13
14
Normal distribution dictates

Mean +/- 1SD Mean+/- 2SD Mean +/- 1.5SD
Will cover 68.2% of Will cover 95.5% of Will cover 84% of
the data, ie, 173 the data, ie, 243 the data, ie, 214
days ina year ( 255 days in a year ( 255 days in a year ( 255
trading sessions) trading sessions) trading sessions
Significant Days = Significant Days = Significant Days =
255-173 = 82 255-243 = 12 255-214 = 41
15
We now divide each day’s price

action into the action of bulls and
bears
Open to High Open to Low
After market After market
opening, the bulls opening, the bears
took the price from took the price from
the Open to the the Open to the
highest point of the lowest point of the
day day
16
Dissecting a price bar

17
Who won the day ?

18
We now calculate these intraday bull and

bear moves for the last 5 trading sessions
We calculate the mean / average of the

individual moves
We calculate 1.5SD of this data
We add the 1.5SD to the average
We get a Positive Deviation value for the

bull side, and a Negative Deviation value for
the bear side
19
We now have two figures, one on

the bullish and one on the bearish
side. If the market now makes a
move more than this figure, whether
on the upside and downside , the
move is statistically significant
20
Any bull move which makes the
markets close above the value of
Open + Positive Deviation( mean +
1.5SD) suggests that a statistically
SIGNIFICANT bullish BREAKOUT has
taken place
Any bear move which makes the

markets close below the value of
Open - Negative Deviation ( mean +
1.5SD) suggests that a statistically
SIGNIFICANT bearish BREAKOUT has
taken place
21
22
23
24
25
26
27
28
29
30
31
This method helps to undertand when a

significant intraday move has taken place and
might provide clues to the future direction of the
instrument
This knowledge can be used as a building block

in designing a volatility breakout system
The system will require specific entry rules, stops

and money management
32
This method can also be used with your existing

method of analysis as another layer which
triggers the trade
The 5 day rule and the 1.5 SD are just that, rules.
It’s not written in stone. Feel free to play with
these two parameters and see what happens
The signals generated are better used in options

strategies depending upon your knowledge of
options science
33
34
Thanks!
Any questions?
You can find me at @SubhadipNandy &
quantgym@gmail.com & +91 97480 52739
www.quantgym.com
WHAT WE
LEARNED FROM
THE KAGGLE TWO-
SIGMA NEWS
SENTIMENT
COMPETITION ALGO
DR. ERNEST CHAN – MANAGING CONVENTION
MEMBER OF QTS CAPITAL
MANAGEMENT, LLC 2019 – DAY 2
What we learned from Kaggle
Two Sigma News Competition
Ernie Chan, Ph.D. and Roger Hunter, Ph.D.
QTS Capital Management, LLC.
The Competition
• Kaggle hosts many data science competitions

– Usual input is big data with many features.
– Usual tool is machine learning (but not required).
• Two Sigma Investments is a quantitative hedge fund with
AUM > $42B.
– Sponsored Kaggle news competition starting Sept, 2018, ending
July, 2019.
– Price, volume, and residual returns data for about 2,000 US
stocks starting 2007.
– Thomson-Reuters news sentiment data starting 2007.
– Evaluation criterion: Sharpe ratio of a user-constructed market-
neutral portfolio*.
Our Objectives
• Does news sentiment generate alpha?
– Find out using normally expensive, high quality data.
• Does machine learning work out-of-sample?
• Does successful ML == successful trading
strategy?
• How best to collaborate in a financial data
science project?
• Educational: example lifecycle of trading
strategies development using data science and
ML.
Constraints
• All research must be done in cloud-based Kaggle
kernel using Jupyter Notebook.
– Only 4 CPU’s, limited memory and slow.
– Kernel killed after a few idle hours.
– Cannot download data for efficient analysis.
– Cannot upload any supplementary data to kernel (E.g.
ETF prices).
– Poor debugging environment (it is Jupyter Notebook!)
– Lack of “securities master database” for linking stocks
data.
Features
• Unadjusted open, close, volume, 1- and 10-
day raw and residual returns.
– Jonathan Larkin[1] designed PCA to show that
residual returns = raw returns - β* market returns
= CAPM residual returns
[1] www.kaggle.com/marketneutral/eda-what-does-mktres-mean
• News sentiment, relevance, novelty, subjects,

audiences, headline, etc.
– Numerical, categorical, textual.
Target and Evaluation Criterion
• Target(t, s): Open-to-open 10-day residual return
from day t+1 to t+11 for stock s (given features
available up to 23:59:59 UTC on day t.)
• Prediction(t, s): Predicted sign(Target(t,s))
• Pos(t, s): Prediction(t, s)*Capital_Weight(t, s)
• Evaluation: Winner has highest
𝑚𝑒𝑎𝑛( 𝑠 Target(t, s) ∗ Pos(t, s))
𝑠𝑐𝑜𝑟𝑒 =
𝑠𝑡𝑑( 𝑠 Target(t, s) ∗ Pos(t, s))
=Sharpe Ratio of zero-beta portfolio of stocks hedged
with market index.
Data Issues and Cleansing
• Lack of “securities master database” – need to
create our own unique id (uid).
– Otherwise impossible to merge price and news data!
• Need to create our own split/dividend
adjustment price series for “fractional
differentiation” [2].
[2] Lopez de Prado, “Advances in Financial Machine Learning”
• Bad price data prior to 2009.

• How do we know if there are errors in news data?
Creating uid
• assetName = company name
– assetName of a company already set to its most recent by data
vendor.
• assetCode = ticker symbol
• Many assetCode → One assetName
• One assetName → Many assetCodes
• T-Mobile → (PCS.N, TMUS.N, TMUS.O)
– Ticker changes over time.
– Red ticker is most recent assetCode, used as our uid!
• Alphabet → (GOOG.O, GOOGL.O)
– 2 classes of stocks co-exist.
– Need to differentiate them due to different price (but not news)
data!
Creating uid
• If two assetCodes for same assetName co-
existed contemporaneously
– Use both as uids.
• If two assetCodes for same assetName didn’t
co-exist contemporaneously
– Just a ticker change.
– (We checked price and time gap to confirm this.)
– Use most recent assetCode as uid.
Bad Price Data
• Kagglers’ consensus: Many errors before 2009.
• Kagglers[3] checked all returns, and changes of
prices and volumes over threshold.
[3] www.kaggle.com/danielson/cleaning-up-market-data-errors-and-stock-splits
• They replace bad open, close, volume with

correct.
– Correct numbers from outside sources.
• They interpolate residual returns.
• We clip target residual returns to [-1,1]
News Data Errors
• Time series plots of statistics of numerical
news features show no structural breaks.
• No obvious way to check categorical features.
News Features
• 2 important numerical features:
– Sentiment ([-1, 1])
– Relevance ([-1, 1])
• We combine these features and take 5-day
moving average of product: movavg(s*r)
+1 if movavg(s*r) > 0
• Prediction(t, s) =
-1 if movavg(s*r) < 0
Naïve News Strategy
• Buy and hold for 10 days if Prediction(t, s)=+1
• Short and hold for 10 days if Prediction(t, s)=-1
• Hedge any beta exposure with market index.
• Equal capital allocation.
• Result on validation set:
– CAGR=2.3% (“alpha”)
– Sharpe Ratio=1
• Result on test set:
– CAGR=1.8% (“alpha”)
– Sharpe Ratio=0.75
News Strategy: validation set
News Strategy: test set
Categorical News Features
• Single value: E.g. headlineTag=‘BUZZ’
• Set of values: E.g. audiences={'O', 'OIL', ‘Z’}
• E.g. headlineTag has 163 unique values,
audiences has 191 .
• Ordinal feature or one-hot encoding?
• Many stocks have multiple rows per day.
• Combine daily features with one-hot and OR.
• Use LightGBM for features selection.
Features Selection
• Problem with LightGBM feature importance
– Uses training data, not validation data
– Hence assetCode and assetName are picked [4]!
[4] Larkin, www.kaggle.com/marketneutral/https://www.kaggle.com/marketneutral/the-fallacy-of-encoding-assetcode
• Solutions: MDA (CV or OOS) [5] or use non-

constant features.
[5] Chan and Hunter, www.kaggle.com/chanep/assetcode-with-mda-using-random-data
Audiences
• Use only 50 most common categorical values
• headlineTag, etc. similarly unimportant.

Price Features
• We have also created simple features based on
prices and volumes only. For e.g.
– Past 10-day residual returns.
– Lagged past 10-day residual returns.
– Fractionally differentiated price series.
– Change in volume.
• Use logistic with L1/L2 regularizations to predict
signs of future returns.
• Capital allocation: “risk parity”
– Inversely proportional to past volatility of returns.
Price Strategy
• Result on validation set:
– CAGR= 17.2% (“alpha”)
– Sharpe Ratio= 1.2
• Result on test set:
– CAGR= 2.7% (“alpha”)
– Sharpe Ratio= 0.28
Price Strategy: validation set
Price Strategy: test set
Conclusion
• For both news and price strategies, alpha and Sharpe ratio
significantly lower in test set than validation set.
• News strategy does not require training and hence little scope for
overfitting.
– Large “variance” likely due to alpha decay of news sentiment.
– Beckers, 2018 (JPM) meta-study of news sentiment research found
average information ratio of news sentiment strategies to be less than
0.5 from 2008-2017! (Performance roughly 1 2 of 1998-2017.)
• Price strategy’s Sharpe ratio deteriorated more in test set.
– Likely due to overfitting, despite simple, regularized logistic regression
model.
– We can’t rule out regime change either.
– Simple technical features do not work.
– Insights into specific market inefficiencies still required!
Thank you for your time!
More details at
www.epchan.com
ALGO
CONVENTION
APRIL 20 & 21
HOTEL NOVOTEL IMAGICA, KHOPOLI. WE WILL SEE YOU IN
CONVENTION 2020

Emailing Algo Convention 2019 PDF

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Emailing Algo Convention 2019 PDF

Загружено:

Авторское право:

Доступные форматы

www.dravyaniti.com www.algoconvention.

1. Noise is relative to time frame

1. Noise is relative to time frame

1. Noise is relative to time frame

1. Start longer term and move

1. Going intraday has a lot of issues

Successful Traders Size Their Positions – Why and How?

Risk = points from where you are going to get

Risk % = $risk of 1 unit of instrument/$equity

Prashant Shah and Abhijit Phatak

Brick-value is important tool

• 0.50% log brick: Short term

• 1% log brick: Medium term

• 3% - 5% log brick: Long-term

8 years data on One-minute

Mean Reversion is a phenomenon that there is an underlying stable trend in the

Markets are trend following in long Term and Mean reverting

Trading System consists of many parts of trading likes of Number of

Systematic Trader’s primary focus should be able exploit edge in the

Source : Piyush Chaudhry @piyushchaudhry

2) Todays % Change > 3%

3) Short at 1% Limit next Day

Back test Period : Sept 2011 – April 2018

Initial Capital : 1,00,000

Capital per position : 1,00,000

Transaction cost : 0.05%

Bear Phase Bull Phase Bear Phase Bull Phase

• Various ways of defining Stop loss :

• Interesting Way of putting Stop Loss based on Maximum Adverse Excursion

Avg Winning % MFE : 2.63%

Ideal Stop Loss : Avg Winning % + 2 Std Winning % MAE

Ideal Profit Target : Avg Winning %MFE + Std Winning %MFE

Important Numbers : Before Stop Loss and & Profit Target

Please make your own due diligence before trading,

• Itching to improve the performance after viewing the recent

Source : Piyush Chaudhry @piyushchaudhry

Email : mastertrader21@gmail.com @vishalmehta29

AIF IN INDIA : APNA ALGO CONVENTION

Option Back testing over existing

Mr. Shubham Agarwal, CMT, CFA, CQF

Section 1: Why Options? Edge over Futures

Section 2: Understanding your existing system with an

Section 3: Approximating the Optimal Strategy

Section 4: Defining Back-testing Parameters

Section 5: Yipee: Back-test your existing system

Why Options? Edge of Options over

Let’s Bet on Nifty

If it goes down 100 points, you pay me Rs.20

Call Option is what

Long Future Payoff Long Option Pay-off

Unlimited Profit Unlimited Profit

Call Option is what

Loss Unlimited loss Loss

Target: 100 points Stop Loss: 20 points

Case 1: Buy Futures Case 2: Buy Call Option

Target Profit: Rs.7,500 Target Profit: Rs.4,500

ROI: 14.5% ROI: 300%

Call Option is what

Trade on Nifty ( Gap Down)

Target Profit : Rs.7,500 Target Profit : Rs.4,500

Desired Stop Loss : Rs.1,500 Desired Stop Loss : Rs.1,500

Initial Capital : Rs. 1 lac Initial Capital : Rs. 1 lac

Margin : Rs.51,546 Margin : Rs.1,500