Good morning Chicago Quantum (SM) friends, and Chicago Quantum Net Score followers. Two nights ago, and into yesterday's trading day we ran a 3,500 stock analysis. All stocks in the NYSE, NYSE American, and two NASDAQ exchanges, after they pass data validation.
A few things happened that are relevant for our customers and colleagues:
1. For about a day, Yahoo Finance fed blank data into historical stock prices, but only the second time we downloaded them, and only from NASDAQ stocks (NYSE and NYSE American were ok).
Over the course of 3000+ stocks, we were ending up with 39 days of complete data, or worst case, 3 days of data (out of 253 days). We validate the data the first time, remove stocks that fail data validation, then we download and validate again. During the second step, historical price and volume data was missing from 13 stocks (we kept a record of the names and exchanges - they were all from NASDAQ). This self-corrected ~36 hours later. A long-term fix is to buy market data services with service level agreements.
2. We ran the analysis on our server, and with the system, code and template defaults it found the best portfolios. We went back and re-ran certain solvers for 2 hours...with no change. When we ran on our iMAC, we found good portfolios, but after re-running for an obnoxiously long time (many hours), we did find better portfolios. Processing power matters.
3. When we change our risk setting from medium to low, we get many more stocks (e.g., from two to seven), and we see companies added that better diversify out the risk. We are still in the range of 3,500 reducing down to a portfolio of 5-8 stocks with a low risk setting.
4. We ran on the D-Wave Systems Advantage 1.1 and it ran well. Out of ~15 runs, we hit a 4-stock portfolio that was in the ballpark of the good answers (not best, but good). Here are some D-Wave Systems Inspector pictures (attached below).
We also shared some skewness, kurtosis and variance data on individual stocks on Twitter. We also shared all the negative BETA stocks we found on Twitter. We picked up a few likes, but that was all.
Finally, today spent time recreating the 13F filings for the one stock we are long in: Summit Midstream LP (NYSE: SMLP). This is not simple when the stock changes CUSIP, goes through a 15:1 reverse split, and the filers do not file quickly...we are stuck looking at Dec 31 and Sept 31 reporting and trying to reconcile and guess. However, we did find the names of a few large money managers. Should we call/write them about our capabilities?
Net-Net, what did we learn?
Once our business grows (revenue), we need to:
- Buy more powerful server(s) to save analysis time
- Arrange to buy market data services with 'guaranteed' access
Keep adding solver capabilities to our platform (e.g., IBMQ, gekko, quantum walks on graphs, etc.)
It is nice to share data with the general population, but not sure anyone values it.
1. We are thinking of lowering our riskfree rate floor below 1%. Thoughts?
2. Should we add NASDAQ Capital Markets (SM) into our analysis, despite the compliance & listing requirement differences?
3. Can we simplify our pricing? Do we have too many options? Should we consider a subscription model?
Disclosure: I/we are long $SMLP (the CQNS model choose it months ago, and earlier in January), which caused us to do significant fundamental analysis.
Thank you again for reading.
Jeffrey Cohen, President, US Advanced Computing Infrastructure, Inc.
Strategic IT Management Consultant with a strong interest in Quantum Computing. Consulting for 29 years and this looks as interesting as cloud computing was in 2010.