High Frequency Trading Strategies

1
High Frequency Trading Strategies
MSE 444 Paul Merolla Erik Anderson Alexis
Pribula
2
Streaming feeds are inexpensive
3
Our data set SP mini futures (ES)?
1 minute bar of ES
Challenge Can we use the added information
from the order book to predict price movements
over the short term (secs to mins)?
4
Price Prediction Strategies

• Use an adaptive MMSE estimator based on recent
  historical data
• Moving window for each point
• Support Vector Machines
• Use last BBOs
• Use moving average of BBOs

Predict this value
Training window
5
Profit of market making strategy depend on
order-fill probability
P(edge clearing) 40
P(edge clearing) 20
Fees 0.35/share, 12 cancellation 2 second time
bins
6
Order Lifetime
1 bin 2 seconds
7
Higher Order Market Data Improves Returns
9.2 return, 3.2 volatility (9 days)?
P(edge clearing) 30
8
Example Trading Day for Market Making
Simulation parameters
Bin Depth
Train window
Date
10
10
2-21-08
Result summary
Edge Clearing Prob.
Profit w/ higher order
Profit w/ lower order
0.5
-1.8
30
Time Bin
1 bin 2 seconds
9
Another prediction approach SVMs(Artificially
Simple) Example

• Best Bid Price goes up
• Best Bid Price goes down

• Support Vector Machines (SVM) separates data with
  non-linear boundaries
• (Real data does not separate as well as this)?

Relevant Measurement 2 e.g. Inside Ask Rate
Relevant Measurement 1 e.g. Inside Bid Rate
10
ICA as pre-processing before SVM

• Let xnx1 be a random vector representing our
  data, e.g.
• x1 is the Inside Bid Rate
• x2 is the Inside Ask Rate
• x3, x4,
• Goal of ICA
• Find matrix B such that the components s1, s2,
  of
• spx1 B xnx1
• are as independent as possible
• are less in number than components of x (i.e.
  pltn)?

Then give spx1,the Independent Components
(ICs), as training data for SVM Rational Filter
s out irrelevant measurements SVM works better on
independent components
11
ICA SVM and Market Making
Market making strategy of buying/selling at
previous Best Bid/Ask Price every 2 seconds works
well but fails when shocks occur
Use SVMICA to predict direction of Best Bid/Ask
Prices 1 minute into the future. Use predictions
to stop market making strategy during market
shocks

• Not as prone to lose money during these shocks,
  however, we have not been able to get perfect
  separation on market shock days
• Either all shocks cannot be modeled with ICA
  SVM, or
• We dont have enough samples to learn good
  separation boundaries

12
ICASVM General Performance
Make order equal to last price with SVM ICA
control
Last Price alone
Conclusion SVMICA does similarly to other
methods on good days while providing some
protection on bad days