Founders Sunday: Time Series Algorithms

Fire Drills, Analyst Spoofing, and more

Overview: for 2p.m.

• Pattern recognition

• Backtesting/ regression analysis

• Making an algorithm

The Goal: to make money

• Statistical Basis: how they are tested, used

• Intuitive starting points

• behavior, behavior, behavior

• Event driven: earnings, one time events, infrequent but known, regularly repetitive

• Signal to noise. the more regular and frequent, the more noise

• The less regular and frequent, the less data proof

• The sweet spot.

Early career examples

• bank robbers, purse snatchers, behavior

• Time Series first example.- PTJ

• Steve Cohen- earnings survey spoof

• Lunch time, dead zone during the Fix

Gold Now:

• Friday behavior last 4-5 weeks

• Friday behavior last 10-12 weeks

• waht si the pattern observed

• what is the pattern that can automated and traded DAILY ONLY

• We dont care why…. just make money

• But here is why anyway.. save money

• pattern Strengthening or weakening over time?

• how can we refine it: volume, OI, TA tools

• geopolitical risk- the behavior is the same, the events are new

• dont go home short/long with geopolitical risk

This recording will be available for Premium subscribers Monday/ Tuesday