Market Closure and Short-Term Reversal Discussion by Dong Lou London School of Economics December 3th, 2015 Imperial College Hedge Fund Conference Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 1 / 12
Summary Main Findings A twist on the daily reversal strategy Traditionally, we sort on the t-1 close-to-close return, and hold the long-short portfolios for a day This paper decomposes daily returns into an intraday component (9:30am-4pm) an overnight component (4pm-9:30am) We can then run four return predictive analyses OC OC, OC CO, CO OC, CO CO Run the same analysis across a number of countries and asset classes US, UK, France, Germany, Japan Equity, interest rate, commodity, currency Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 2 / 12
Summary Main Findings In the US equity market using the transaction price (TAQ) at 9:30am as the open price focusing solely on S&P 500 constituents over the period of 2011-2014 the CO OC strategy has a daily return of 36bp (t-stat of 11) in other words, a monthly return of 8% this return spread is even bigger on days with higher VIX Similar results in other developed equity markets Similar results in other asset classes (w/ smaller magnitudes) Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 3 / 12
Summary A Quick Summary An interesting empirical piece with lots of cool results lots of robustness checks results from non-us markets results from various asset classes My discussion will mostly focus on the equity market One big comment on the key result One smaller comment on the data Will be comparing to my own work with Polk and Skouras Apologies... Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 4 / 12
Comment #1 The First Minute Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 5 / 12
Comment #1 The First Minute If we skip one minute between the formation and holding periods, the return effect goes down by more than 2/3 If we skip by 15 minutes, it goes down by nearly 90% Note: the reversal effect remains significant Market microstructure concerns Is the price at 9:30am reliable? What s the volume in that minute? Is the price tradable (i.e., enough volume)? There s also an open auction for some stocks Does it make a difference using auction price vs. first traded price? What about the big-ask spread at open? It d be interesting to understand what happens in that minute Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 6 / 12
Comment #1 To Proceed... A placebo test intraday liquidity is hump-shaped, so lowest liquidity at noon repeat the analysis using the 12pm price (averaged over one minute) as the open price if there s something special about 9:30am, should see a weaker reversal pattern using the noon price Lou, Polk and Skouras (2015) try to get around these issues by using the VWAP over the first 30 min (ensure enough volume) examining the monthly reversal strategy (less s.t. bid-ask bounce) turns out there s a surprisingly persistent component in overnight/intraday returns that s unrelated to short-term reversal Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 7 / 12
Comment #1 Lou, Polk and Skouras (2015) Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 8 / 12
Comment #1 Lou, Polk and Skouras (2015) Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 9 / 12
Comment #1 Lou, Polk and Skouras (2015) Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 10 / 12
Comment #2 Data Issues For most tests, the paper uses CRSP for the US equity market Datastream for foreign equity markets It s unclear how CRSP or Datastream calculates the open price is this the open auction price or first traded price? what if the first trade happens at 3:30pm? what if there s no trade at all on that day? what is the volume associated with the open price? better to use TAQ (TFTH) for all the tests How do you deal with a missing open price? not uncommon for small cap stocks should not randomly skip a day one solution is to hold the position till the next open Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 11 / 12
Conclusions Conclusion An interesting paper Definitely recommended Maybe play down the return reversal effect in the first few minutes after market open Better to use TAQ (TFTH) for all the tests Dong Lou (LSE) Market Closure and Short-Term Reversal Brevan Howard Centre 12 / 12