Scribd
Upload
68 views21 pages

Exogenous Vs Endogenous Data

Technical trading models use historical stock price and volume data as endogenous factors to predict short-term price changes, while event-driven models use exogenous fundamental and macroeconomic data like earnings or interest rate changes. Both types of models require large amounts of input data. When building financial models, it is important to consider what data is available, assumptions made, and whether multiple factors could be combined.

Uploaded by

raghs4u
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
68 views21 pages

Exogenous Vs Endogenous Data

Technical trading models use historical stock price and volume data as endogenous factors to predict short-term price changes, while event-driven models use exogenous fundamental and macroeconomic data like earnings or interest rate changes. Both types of models require large amounts of input data. When building financial models, it is important to consider what data is available, assumptions made, and whether multiple factors could be combined.

Uploaded by

raghs4u
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/ 21

What do we want to model?

Understanding Exogenous
vs Endogenous Data
Learning Objectives
● Distinguish between Exogenous
and Endogenous factors

● Practice: What types of data to


use for technical and
fundamental trading models?

● Preview how you can model


build with BigQuery ML
Agenda
Exogenous and Endogenous
model factors

Asking the right questions of


your data

Case Study: Modelling CPU


Performance by Vendor with
BigQuery
Trading and investing models
have different goals
● Trading = predict relatively
short-term changes in the price
of an asset

● Investing = estimating the fair


value of assets relative to their
current price

● Both types of models require


lots of data as input factors
Endogenous factors depend on trading data

Technical Strategy Price data


Price trends and volatility: AAPL
traded higher 4 out of 5 days and
Model daily price volatility decreased to 1%

Size of orders at the bid and ask price,


Use historical stock prices, Order bid/ask spread: Average size
order book information and Book
increased to 3,000 shares. Bid-ask
spread declined to $0.14.
trading volume for AAPL to
predict future behavior Volume: AAPL traded to a record high
Volume on 145% of average daily volume.
data

Order Trading
Price data Endogenous models are created using the historical price and
Book Volume
volume data of AAPL stock. Endogenous models assume that
all available fundamental data is incorporated into the current
share price. Volume is used to distinguish price noise from
more durable price changes.
Exogenous factors depend on fundamental or macro data
Earnings Shock: AAPL quarterly
Event-Driven earnings and forward guidance
Strategy Model exceed analyst expectations

Supplier Shock: A major


semiconductor supplier shuts down
Affected by releases of operations
fundamental data affecting
AAPL and by unexpected Customer Shock: 30% tariffs are
changes in macro variables imposed on iPhones and watches

Quarterly Customer/
Macro Data
earnings and
guidance
Supplier
Shocks
Releases Exogenous variables affect share price but are not dependent
on share price.
Technical Model: Predicting changes in stock price for AAPL
using stock price and volume data alone

Are these factors Endogenous or Exogenous?

An unanticipated change in The U.S. Federal Reserve changes A competitor has major safety
consumer tastes for your product the cost of borrowing issues with their products

The price of critical components for


Shipping times from overseas grow
making phones and computers Quarterly earnings announcements
due to new ocean weather patterns
suddenly changes

Trading volume matched with price


to track CEO is replaced unexpectedly Trends in stock price volatility
Volume-Weighted-Average-Price
Technical Model: Predicting changes in stock price for AAPL
using stock price and volume data alone

Are these factors Endogenous or Exogenous?

An unanticipated change in The U.S. Federal Reserve changes A competitor has major safety
consumer tastes for your product the cost of borrowing issues with their products

The price of critical components for


Shipping times from overseas grow
making phones and computers Quarterly earnings announcements
due to new ocean weather patterns
suddenly changes

Trading volume matched with price Trends in stock prices measured by


to track Volume-Weighted Average CEO is replaced unexpectedly
momentum
Price
Event-Driven Model: Predicting changes in stock price for
AAPL based on release of fundamental and macro data

Are these factors Endogenous or Exogenous?

An unanticipated change in The U.S. Federal Reserve changes A competitor has major safety
consumer tastes for your product the cost of borrowing issues with their products

The price of critical components for


Shipping times from overseas grow
making phones and computers Quarterly earnings announcements
due to new ocean weather patterns
suddenly changes

Trading volume matched with price Trends in stock prices measured by


to track CEO is replaced unexpectedly
momentum
Volume-Weighted-Average-Price
Key takeaway: Exogenous and endogenous factors both drive
changes in share prices. Technical strategies tend to focus on
endogenous factors and event-driven strategies focus on
exogenous factors
Agenda
Exogenous and Endogenous
model factors

Asking the right questions of


your data

Case Study: Modelling CPU


Performance by Vendor with
BigQuery
Is there a “right” set of data
for financial modelling?
Ask yourself:

● What data do I have?

● What data do others have?

● Does the freshness of my data matter a


little or a lot?

● What assumptions is my model making?

● Is there a combination of things I could


model?
Is there a “right” set of data
for financial modelling?
Ask yourself:

● What data do I have?

● What data do others have?

● Does the freshness of my data matter a


little or a lot?

● What assumptions is my model making?

● Is there a combination of things I could


model?
Is there a “right” set of data
for financial modelling?
Ask yourself:

● What data do I have?

● What data do others have?

● Does the freshness of my data matter a


little or a lot?

● What assumptions is my model making?

● Is there a combination of things I could


model?
Is there a “right” set of data
for financial modelling?
Ask yourself:

● What data do I have?

● What data do others have?

● Does the freshness of my data matter a


little or a lot?

● What assumptions is my model making?

● Is there a combination of things I could


model?
Is there a “right” set of data
for financial modelling?
Ask yourself:

● What data do I have?

● What data do others have?

● Does the freshness of my data matter a


little or a lot?

● What assumptions is my model making?

● Is there a combination of things I could


model?
Agenda
Exogenous and Endogenous
model factors

Asking the right questions of


your data

Case Study: Modelling CPU


Performance by Vendor with
BigQuery
Case study: Predicting CPU performance across
vendors
If you knew the performance of brand new CPU chips
produced by leading semiconductor manufacturers, what
could you model?
● Performance vs competitor
Potential inputs in a
● Performance over time
model for valuing a
● Growth/slowing in improvements business
Case study: Predicting CPU performance across vendors

You are given the following raw inputs: Could you quickly predict an overall
performance score?
● vendor,
● model_name, benchmark_score
● max_mhz,
● nominal_mhz,
● cores,chips,
● channels,
● mem_gb,
● mem_speed,
● l1_cache_mem_kb ,
● l2_cache_mem_kb ,
● l3_cache_mem_mb ,
● os,
● compiler,
● sponsor
Topic Preview: Linear Regression with BigQuery Machine Learning

Model name
● Quickly test if you can accurately
Model type and what
column we’re predicting model the behavior in the data in
for minutes

● If you’re predicting a number, use


a forecasting model (like linear
Model inputs
regression)

● If you’re classifying (bucket A,


bucket B) use a classification
model (like logistic regression)

● Spend the most time


understanding the nature of your
data before modeling
The 80/20 rule of building machine learning models

Actually writing model code

Data collection, cleansing,


transformation, and
feature engineering

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