caklim00 wrote: ↑
Tue Mar 24, 2020 9:55 am
Ok, so you would be paying interest then. Why not just use e-mini/micro e-mini for S&P 500 and ZB for treasury bond futures instead?
Please don't read this as a recomendation that anyone does this, but just wondering why pay interest buying ETFs on margin when one can just use futures instead?
Because in back tests, $3.2 million algorithm @ 3% margin interest rates > $2.5m SPY/TLT 3x levered quarterly re-balanced > $1.5m UPRO/TMF?
IBKR is really competitive with their margin giving libor + 0.50%. For 0.50% basis points it's absolutely worth doing spy/tlt on margin:
Because I've shown SPY/TLT is very tax efficient having a negative
tax drag on capital gains and a minor
tax drag of 1.27% from dividends. Futures are marked to market
at 60% LTCG/40% STCG. Futures are very tax inefficient
for buy and hold like this.
Futures are very tax risky.
Imagine being marked to market paying 60/40% on capital gains of say $300k in the peak 2007 year, then having 2008 come and now your position is $89k? You pay tax on your futures market value on December 31st of each year.
. Your capital losses will offset new gains of course, but in reality there's going to be huge tax withdrawals from the account to pay for that insane tax drag.
SPY+TLT on margin means every tax lot is long term capital gains, re-balancing is done efficiently with the extra cash from margin: you drift to 2.9x margin, re-balance assets & leverage to 3x, you've bought most of your new shares on margin to re-balance with instead of selling old shares.
Is all of this worth 50 basis points for the markup IBKR has over Libor? Good god yes.
No one should be considering futures for this portfolio in a taxable account. Maybe for a IRA at TD Ameritrade as they only mark up margins by 125%.
I've not even got into the algo trading side of futures:
QuantConnect does support trading futures but they have no data going back to 2003 for futures. Their implementation sucks. If I drop off the 3% margin interest simulation rate down to say 1.5% or 0.75%, it only affects the ending value to something like $3.5million. Need to make sure no bugs in trading futures.
If there's a bug in trading stocks in my algorithm, say it's spamming orders constantly or buys too much, that's a lot less dollar risk. Let's say I have a live bug that is spamming buy orders because there is lag on order fills (this is an actual bug I had early on in my algo trading hobby.) With equity it's an annoyance. With futures contracts it could be a bankruptcy.
Flash crashes in the futures are a real possibility still. The /ES futures went as little as 50-100 contracts showing at the bid/ask when usually it's been 10k contracts. SPY and TLT has remained liquid throughout the most recent volatility. Too much risk of bad fills in volatility using futures. I'd need to write a lot of extra code to price check the /ES and /ZB futures.
The bond futures are a different asset class than TLT. They are a fixed set of securities specified by the exchange. They have different convexity than TLT. TLT has bonds in the fund that would never
be delivered in a futures contract. This needs to absolutely be back tested to make sure it doesn't blow up in 2008. TLT's convexity is really interesting as it's very favorable in bullish bond markets and less risk in bear bond markets. The bond futures convexity acts more how one individual bond would trade instead, and so there may be more risk there.
That's why I'm not trading futures. I think it's a stupid idea in taxable
for anyone who wants to shoot for over a $1m adventure or doesn't need to realize gains for 20+ years.
In tax advantaged
, sure? Knock yourself out. It's either futures or UPRO/TMF, so futures will probably win out. With how margins are you'd probably need a 200k+ IRA to use the futures version and you'd need to watch it like a hawk to manage your leverage. At $177 for /ZB one contract has a notional value of $177,000. After hours /ES is 2442x50 - 122,100 notional. Micro futures are $5/tick so $12,220 notional.
So we can get our S&P leverage fine from micros, I won't bother with the math here. The bond future is the biggest problem. With 3x leverage the minimum account size for 45% bonds levered to 3x is: 177000/3/.45 = $131k. There's going to be a crap ton of variance in leverage. Say your portfolio goes down 20% to 104.8k, and stays there for the next quarterly rebalance. Now your bond allocation is .45 * 104.8k = 47.16k Now, with that one
futures contract, your bond leverage is 3.75x. (math: 177/47.16). Assuming you have perfect 3x leverage on the S&P 500 your new portfolio leverage is: 3.3375x (3.75*.45 + 3*.55). So now you're taking on more risk then you signed up for and probably need to get out of the future completely.
You wouldn't be able to use 2 full contracts until you're at 262k, 4 contracts at 524k, 8 contracts at 1 million, etc. You'd have some position of cash + bond fund + futures that's changing constantly due to the leverage. Honestly, I take back my 200k account recommendation. You won't get a smooth leverage curve until you're 1 million+ and have enough numerical fluidity to adjust contract sizes to leverage. $1 million at IBKR would be a $2m margin loan on SPY/TLT at perfect 3x leverage, and definitely qualify for benchmark + 50 basis point rates.
Thinking about future math with such large fixed contract sizes really hurts my brain. I wouldn't recommend replicating this strategy with futures until there is a micro for /ZB.
I think on I'll probably just do UPRO/TMF in my tax advantaged accounts as it'd still be beating 100% stocks to the ground, or just leave them alone in a more conservative portfolio (bucketing.) Honestly - do UPRO/TMF in tax advantaged until $1m or micro /ZB comes out, then finding someone like TDA who will let you have 125% futures margin in an IRA.