Investment Approach

Trading Signals

Trading signals are currently derived from 13 independent models*

Each model uses unique data elements to determine its trade signal, for example:

Relationships among specific currencies/equities/interest rates

Interest rate sensitivity of specific fixed income instruments

Multiple models focus on one market, unlike strategies where the same model is used on multiple markets

Very low correlations between sub-models

Each model is backed by extensive research & development in an attempt to identify data that is predictive of Treasury price movements

100% quantitative for trading signal generation

We believe the math is elegant but not complex

* The numbers of models can change over time

 

Mitigating Model-Specific Risk

Simple example of the interaction between models and the long/short decision making process

Data-driven, multi-model decision

Each sub-model is derived from a unique data set

Attempts to identify models that are robust with long-term sustainability

 

short/long switches

 

Sub-Model Correlations

Short Models 1-5

Long Models 6-13


Models 1 2 3 4 5 6 7 8 9 10 11 12 13
1 1.0                        
2 .071 1.0                      
3 .157 .04 1.0                    
4 -.025 -.03 .038 1.0                  
5 -.014 .088 .155 -07.9 1.0                
6 .013 .077 .033 .039 .017 1.0              
7 .097 .075 .039 -.069 .090 -.066 1.0            
8 .038 .010 .015 .027 -.001 0.74 -.026 1.0          
9 .016 .024 -.002 .023 .011 .018 0.0 -.004 1.0        
10 -.003 .019 .030 .008 .041 .055 -.074 .173 .013 1.0      
11 .144 .067 .137 .037 .072 -.025 .003 .034 .015 .024 1.0    
12 -.051 .039 .008 .036 .076 -.008 .007 -.066 -.0008 -.043 -.042 1.0  
13 .081 .048 .126 .12 .251 .008 .021 .184 .0007 .037 .289 -.114 1.0


10/19/2011 through 06/30/2017 (most recent data available)

 

  princeton logo