By Jeffrey Cohen, President, US Advanced Computing Infrastructure, Inc.
I am working on make the run go faster without giving up value.
We already cut the solver run time by ~70% and get the same CQNS answer. That is a good start. In fact, today we added back 10% of our horsepower in our solvers (so maybe it will be 63% faster, just to be sure we don't miss any good solutions).
Now, the hard part. We need to speed up the data collection process. This has always been the sloppy part of our code. Lots of if, then, else statements. Lots of calls. Lots of data (and memory) to be able to write out a huge spreadsheet that we only use occasionally. This spreadsheet is great when we are doing fundamental analysis, but not when we are in a hurry to 'see the picks.'
So, our current step is to see what data and output is superfluous.
We love our BETA, market cap, dividends. We use them in our math. Also, Chicago Quantum is the home of the Negative BETA stock list :-).
However, there were a few hypotheses we tested a few years ago. Things like price and volume spikes. Stock splits. Do we really need them? Maybe not in the run_really_fast parameter setting.
We created a detailed spreadsheet that we use for fundamental analysis. It has valuation info. Likely most of that can go.
So, here is the initial list of variables and outputs that can go. This will simplify the run_really_fast parameter setting run.
Data to cut from inputs and outputs:
1. Cash Flow From Operations
2. Long Term Debt
Last day volume as a % of average
There will be more...this is the first part of the list.
Your feedback is welcome! Please comment below or directly contact me at +1.312.515.7333 (cell) or firstname.lastname@example.org (email).
Leave a Reply.
Full-time investment advisor and student of the financial markets.