If I am short both TQQQ(+3x) and SQQQ (-3x). To rebalance, I always buy the leg that appreciated; if QQQ goes down, I have to buy SQQQ (ie short QQQ); if QQQ goes up, I have to buy TQQQ (ie long QQQ). So I will be buying high and selling low in each step, like delta hedging for a short option, so am short gamma. The other side then is long gamma. What you said is LETF holders are short gamma, the opposite.
I wasn't referring to holders of the etf. I was referring to the etf itself and its need to rebalance
I am confused why either of you think that these things have any "gamma". It's just an excess return index, so it's a martingale by definition.
Your right. It is not gamma. I am confused. If held in balance, at any point in time, they hold 2-3x the underlying index, which is delta1, so they must be linear too during the period. So my question is, if I am short TQQQ/SQQQ, which is referred to as a short straddle by Jonathan Kinlay, where does the gamma come from (hence theta = source of profit) ? If I rebalance daily, the portfolio is absolutely flat, there is no theta PL. By playing with Excel, all I see is PL coming from trending (compounding) vs sideway market (decay), without any rebalancing. I thus don't understand what Jonathan Kinlay was talking about in his Nobel-prize winning article.
Whatever, I don't get paid a premium to hold long or short, so there is no theta and so no gamma. Let's not talk greeks. They are delta1 products.
Ok, maybe gamma was an unclear choice. But the etf needs to buy on up days and sell on down days to rebalance NAVs with the reference index. I thought using delta and gamma terminology made things easier, apparently not
The "decay" comes from several sources (a) locking in mean-reversion which tends to be higher in more volatile markets. (b) hefty fees, which will probably be reflected in the borrow rates (c) slippage at the close - usually these things have a pre-close observation window (d) impact on the close price by their own trading If you simulate day-to-day, you can replicate their prices pretty close as long as you get the exact observation/execution logic right. If you simulate/trade IRL with less frequent re-balancing (e.g. start flat and rebalance to some sensible hysteresis band) you're mostly trading mean reversion with a tiny bit of other effects.
Yeah, that's correct. I just think gamma implies some sort of convexity which this thing does not have.
Right, the PL (short side) comes predominantly from mean-reverting moves (gyrations). Regarding convexity, how I see it is they are linear between rebalancing, but show convex payoffs if looked at over multiple rebalancing periods. Is it not. Or maybe I should just call it path-dependency.
It’s definitely path dependent. If you’re going through a period of positive autocorrelation, leveraged ETFs will actually show positive expectation and vice versa