Treasury Futures Trading Project
Hi all, To provide some background: Math major at non-target and intern at a startup looking to break into FICC S&T (Looking more into rates/rates derivs, but not super set on anything). I have been reading up on the recommended books and primers as well as networking. In my current role, I have been doing tons of data analytics in python and wanted to start applying it to something that is more adjacent to S&T. Also, I learn better by doing vs reading. Outline: Historical Data used: Treasury Yield Rates, TY Rates 14D Rolling average change Std Dev, CPI, GDP Growth, Unemployment, Fed Funds Rate, and Fed Balance Sheet Value. (Done) Do PCA on the Curve for slope, curvature, and magnitude. (Done) Cluster using scikit K-means to find steepening, flattening, increasing vol, and constricting vol regimes and then back testing. (WIP) Basically, just trying get signals for Short 2 year / Long 10 year, L 2yr/S 10 yr, buy straddle, and sell straddle. (Mostly just steepening/flattening, the options part would be cool to try but likely to be unsuccessful.) So that's the idea, but would like to hear from y'all on: A.) If I were to document my progress on Github or a blog, could that make up for my lack of S&T experience during recruiting? Or would it be looked over or go against me? B.) Any better ways to do this / problems that you might foresee. C.) Any recommendations for resources for learning more. Sorry in advance if anything is stupid or wrong, still learning.
I think this is a cool project to get more comfortable with python as it relates to markets and will definitely help with self development but won’t necessarily make or break you for recruiting. Joining clubs and staying up to date with market trends will be more helpful imo (assuming you’re recruiting for SA). I was on a rates desk over the summer and most of the coding I did was just fetching and cleaning data for the most part. Also built some dashboards and liquidity models.
pretty sweet, but I was once told that PCA/Regressions should not be where you find the 'juice'. Your idea or reasoning for displacement and how you think it will play out can be confirmed by them. Maybe in your backtests you saw a certain curve pair or fly or area in iro/swaption vol surface get displaced... you may think this is a rv opportunity but what if it keeps deviating? so much live supply/demand that makes a huge impact on prices. Also maybe look at some cash vs futures, swap spreads.. etc.
I got a few examples up in this repo: https://github.com/yieldcurvemonkey/Curvy-CUSIPs
would love to see a PR come in on a ust futures example
I will definitely take a look and see what I can do. Mind if I PM?
feel free
also not really related to OP's post but everybody is sleeping on Matlab - mathwork has some really nice FICC examples up
Care to share any gems?
All reasonable ideas and potentially good experience although I am not sure how much it will help with recruiting. Arguably treasury futures are not the ideal starting place for this sort of project as the quarterly roll and inherent cheapest to deliver optionality complicate things. And this is beyond the "10 Year T-Note" being much closer to a 7-year cash bond,. Looking at the underlying cash bonds is certainly a lot of the juice in treasury futures but I am not sure if there is high quality free data for that.
Minima cumque laudantium suscipit animi. Minima autem quibusdam ut incidunt illo eveniet. Facilis voluptas explicabo libero consequuntur pariatur unde.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...