Does delta-hedging change the transaction cost picture for option momentum strategies?

External question — Quantitative Finance Stack Exchange Author: ALi Saleh Original post: https://quant.stackexchange.com/questions/85743 License: CC BY-SA 4.0 — https://creativecommons.org/licenses/by-sa/4.0/ Adaptation: HTML converted to plain text; contact email addresses removed. Heston, Jones, Khorram, Li and Mo (JF 2023) document option momentum: trailing 6 to 36 month straddle returns predict future straddle returns, robust to delta hedging and to factor adjustment. Their transaction cost argument is cross sectional, showing that momentum profits are similar for high cost and low cost underlyings. The paper itself notes a median bid-ask spread of about 14 percent on their portfolios and acknowledges that costs could eliminate the profits if the mispricing lies inside the spread. I ran a fill level test of a long only variant to see whether that concern binds. Setup was fixed in advance: S&P 100 underlyings, 2013 to 2026, nearest ATM straddle on standard third Friday expirations, one month hold, 12 minus 1 formation since the paper documents lag 1 reversal. Long only top decile, equal weighted, no short leg. Entry at ask and exit at bid with no mid quote fills, plus 0.65 per contract commissions on both legs both sides. Liquidity screen of 500 open interest per leg and straddle spread under 15 percent of mid. Data was OPRA quotes, definitions and open interest. Result was 0.0963 net annualized Sharpe over 138 months, with a mean of 55.5 eligible names per month, so the outcome is not driven by thin breadth. My question is specifically about the delta hedged construction. The authors report that delta hedged call and put returns display very similar momentum patterns to the unhedged straddle version. Hedging removes the directional component, which should reduce the variance of the position, but it introduces rehedging costs in the underlying. Is there published work that evaluates the delta hedged version at effective fills rather than mid quotes, including rehedging costs? More generally, for a monthly held option position, does delta hedging typically improve or worsen the net of cost Sharpe once the underlying trades are counted? I am trying to determine whether the failure I observe is specific to the unhedged long only straddle, or whether it extends to the hedged construction as well.

External answer — Quantitative Finance Stack Exchange Author: Russlan Ramdowar Original post: https://quant.stackexchange.com/a/85787 License: CC BY-SA 4.0 — https://creativecommons.org/licenses/by-sa/4.0/ Adaptation: HTML converted to plain text; contact email addresses removed. tl;dr: hedging makes it worse, not better. your 0.10 Sharpe isn't a bug, it's the actual finding once you stop pricing at mid like it's free money. Here's the mental model. Unhedged straddle = one trade in, one trade out. All your pain is that one 14%-ish spread you already flagged. Done. Delta-hedging = you buy the straddle AND THEN you're rebalancing the underlying every day to stay flat. Sounds smart, strips out the directional noise, leaves you "pure" momentum signal etc etc. Except now instead of paying one spread, you're paying a tiny spread ~20 times a month (daily rehedge on monthly hold). Death by a thousand cuts basically. Two things stack on top of each other and neither is free: realized vs implied vol drag — every rehedge is basically buying/selling gamma priced off implied vol, but your P&L is driven by realized vol. If realized comes in lower (very common right after a momentum signal fires, vol mean-reverts), you just quietly bleed on every single rehedge. No spread even needed for this one, it's a pricing mismatch thing. actual bid-ask on the underlying, times however many rehedges you do. even a "cheap" 5bps underlying spread, paid 20x, is not nothing. it's a whole extra cost layer sitting on top of the option-leg cost you already modeled correctly. Heston/Jones/Khorram/Li/Mo's hedged-vs-unhedged similarity is a mid-quote finding. mid quotes are academic fantasy land, nobody trades there. the second you go to real fills (ask in, bid out, PLUS bid-ask on every rehedge trade) you've added a whole new friction layer that just doesn't show up in their headline number. Closest stuff I know of that pokes at this: Muravyev 2016 (JF), "Order Flow and Expected Option Returns" — shows a bunch of option return anomalies just disappear or flip once you use real quote/trade data instead of mid. closest analog to what you're doing, just not momentum-specific. So no, this isn't specific to your unhedged long-only construction — it's the general pattern. Delta-hedging will almost certainly make your net Sharpe worse at these fill assumptions, not better, because you're stacking underlying-leg friction on top of option-leg friction. If you do want to test the hedged version for real, the lever to poke at is rehedge frequency — try weekly instead of daily and see how much cost drag you save vs tracking error you give up. that tradeoff is the whole ballgame.