So,
I didn’t know much about availability and borrowing costs on stocks apart from large caps. Here are some quick numbers from just the past ten days:
- For top 1000 Mkt Cap with over $5MM Daily Tot (60), seems like good availability and can currently borrow most to all of them for 0.25% to 2%…except in severe short squeezes where you can see 8-10% annual rates.
- For AvgDailyTot(60) $2MM to $5MM. Costs range from 0.25% to 5%. In current environment.
- For $1MM to $2MM. Using only SP1500. Can’t always borrow. But usually can (at least in this calm environment). Costs can raise up over 10%, on stocks with hi SI. Not sure can accurately backtest models in this liquidity range. MArgin rates will spike and availability fall during big market crashes. (I think).
Pretty clear that the margin costs should be pegged to the liquidity in some way (like a variable margin cost formula). Would want to look back at 2008 and 2002 as well.
There are also capacity constraints on the lower end. Sometimes not enough shares to sell an R2G.
Data points. Where’d I get these numbers? Some current short costs to borrow stocks:
Model 1. (all at IB). These are all big stocks. My systems either force stock to be in top 1000 in market cap and doing over $5Million dollars a day in daily liquidity over past 60 days. Sometimes both.
- BBRY. Reached past 8%+ costs when short squeeze on. I was shorting.
- FNV. Currently 0.25%. Reached 1.46% when I was shorting last week.
- FMX. Current. 0.72%. A few hundred K in shares available.
- EW. Current. 0.25%. 2Million shares available.
- TRQ 0.47%. A few hundred thousand shares available.
What happens is…if there is a big short squeeze, borrowing costs go way up. I was in BBRY during this. Otherwise, these large, liquid stocks have a few hundred thousand shares available at low rates. Likely pegged to interest. You also have to pay the dividends on the underlying stock, but models handle that.
If I lower liquidity and look at shorting stocks between $5M DailyTot(60) and $2M daily tot (60) - I get stocks like -
- RIGL. 0.25%. OVer 5 million share available.
- PVG. 1.8% with 400,000 shares.
- SEA. 2.8%. Has been has high as 3% in past 10 days. Only 20,000 shares available.
So, borrowing costs re rising a lot without short squeezes in place. If a system is using high short interest to find shortable stocks, they are paying much more likely…as those rates are rising. Some stocks can’t be shorted.
If we go 1MM to 2MM liquidity, get the following:
- IMI. 0.35%, 500,000 shares.
- TAHO. 100,000 shares available. 0.43%
- TRX. No data available.
- UEC. No data available. Checked NYSE and Nasdq.
Now, it’s possible I’m not finding these because I’m in the All Fundamentals and these are foreign stocks on foreign exchanges.
So…I switch Universes to SP1500 with same low liquidity constraints.
- SIGM. On the Nasdaq for 0.25% Over 2,000,000 shares.
- NANO. 0.35%. Over a million shares.
- DEL. 0.25%, 250,000 shares.
- PQ. 0.25%. over 3,000,000 shares.
- HZO. 0.25%, over 2,000,000 shares.
Bottom line, stocks within SP1500 with AvgDailyTot(60) between $1MM and $2MM seem to be pretty avail. Would need to run samples on past dates to make sure. And under short squeezes. But appears workable.
Now, I look at the top 10% of companies with outstanding short interest. With the $1MM min. liquidity (up to anything).
- BKS. .627%. 200,000 shares available.
- SWC. 0.6%. 800,000 shares.
- AIRM. .052%. 250,000 shares.
So, even at these levels of shorting, seems that can be playing. Each model builder would need to run tests for at least 30 of their stocks over 5 different time frames, but appears we can short more than I first thought.
If I force the sim to look at only stocks with liquidity between 1Million and 2 Million Avg Daily 50 and top decile short interest, I get:
- RICN. Cost is over 10.8%. Only 25,000 shares.
- FBP. .655%. 100,000 shares.
So…in some cases, get huge spikes in costs.