No. of Recommendations: 7
This may be what you mentioned. The thread is long. This is the first post.
As some of you know, I am in the fortunate position of not having to
worry about capital gains tax, but here's a post for those in the US of A
where not only do they hit you with capital gains tax, but it's a higher
rate for things with short holds like the typical approach using quant and/or timing.
First, put 100% of your money in RSP, the equal weight version of the S&P 500, permanently. No trading, ever.
That's gotta be tax efficient, right?
So far we have a strategy that is giving 14.1% CAGR and DDDD3 risk metric 9.51%.
The test period here is late January 24 1978 through last Friday.
Now, let's take some of the sting out of it.
We'll hedge when the omens are bad, using 75% short Nasdaq 100 futures.
These aren't too bad tax-wise, for two reasons. First, I understand the
profits on futures are taxed partly at the long term tax rate even for short trades.
And second, the hedging isn't where most of the profits lie in this strategy anyway.
The first omen we try is hedging when NH-NL signal is negative.
i.e., hedge when there have been more new 52 week lows than new 52 week highs on the Nasdaq in the last week or two.
I use a version "vote on recent days" which doesn't trade too often,
so we're 75% hedged about 70% of the time. 3.5 trades per year on average.
That changes our portfolio to CAGR 18.18%/year and risk metric 4.22%.
Why the number 75% for the hedge? It gives the lowest risk metric.
But, for the real steadiness, does anybody remember the Hot24/Hot30/Hot36/Hot42/Hot48 strategy?
The market tends to do very well on average around month ends in winter months.
That's basically the superior six days in the better winter months,
with the variations just being the number of month-ends you consider good.
First post on that here
http://boards.fool.com/for-the-lazy-man-28055010.a......
Most recent in the middle of this thread
http://boards.fool.com/how-robust-is-it-when-nh-nl......
Since returns are so good in those days (bear market or not), and since
it costs essentially nothing to put your hedges on and off with futures,
we make the extra rule ALWAYS go unhedged during the Hot42, bull or bear.
That's the 6 bullish days around the ends of the 7 months Oct/Nov through Apr/May inclusive.
6 days per month, 7 month ends, 42 days per year unconditionally unhedged.
(the superior six days here are the last four of one month and the first two of the next).
So, our final rule is:
Long RSP at all times, no trading, no tax ever* except some on the dividends.
(*until you retire and start liquidating, anyway).
Hedge 75% short Nasdaq 100 futures during the times that BOTH of the following are true:
- NH-NL signal is negative (more new lows than new highs lately on the Nasdaq)
- It's not one of the Hot42 days.
Note that the long RSP position is more than adequate for the security to do
the hedging with futures. You can be $100000 long and $75000 short with a $100000 portfolio.
It sounds like a lot, but it's much safer than 100% long!
This brings the final tally to CAGR 20.1% and DDDD3 risk metric 3.01%.
You're unhedged (100% long) 74.7% of the time and 75% hedged the other 25.3% of the time.
You'd think there would be tons of signals per year, but not so.
The NH-NL bullish periods (avg 3.54 sigs/yr) often line up with the Hot42
bullish periods (always 14 sigs/year), so a lot of them cancel out.
e.g., going from one bullish to both bullish back and forth creates no trades.
The final result is an average of only 7.2 signals per year (3.6 round trips).
Note, I haven't disclosed the gory details of the exact NH-NL signal I'm testing,
but it's just one variant of what's used around here. No real secret sauce.
I have a bunch of different variations based on the general idea of "how
many of the last N days had more, or substantially more, new highs
than new lows on the Nasdaq". I took the 5 best variants and made a
meta-model that's bullish only if all 5 of those 5 are bullish.
Thus it's not critically dependent on any one tuning parameter that way,
and it's a tiny bit risk averse because it goes bearish if ANY variant is bearish.
But the variant isn't that critical. If you use the most brutally
simple variant exactly as I defined in my last post on the subject...
http://boards.fool.com/maybe-mungo-will-reiterate-...... (end of the post)
...you get CAGR 18.65% and DDDD3 risk metric 4.17%, 8.9 signals/year on average.
Of course not many people have a feel for what a DDDD3 risk metric means.
After all, I made it up myself. But zero risk means every single rolling
year return had a return of at last 10% and every single rolling quarter was at least breakeven.
The Nasdaq 100 has had a metric of 14.8% so the risk figures above are truly wonderful.
A word is worth but one one-thousandth of a picture, so here's what the last 35.4 years looks like:
stonewellfunds.com/TaxHedge.jpg
The colour change divides the pre-discovery and post-discovery period.
I came up with this a while ago.
The downsides to this approach:
(2) The biggie is you do have to check the NH-NL daily when it's anywhere near a crossover.
In fact most of the time you don't have to go near a computer because
it's painfully obvious if you're in a bull or a bear lately, but sometimes you do.
Of course the upside is you never have to look at an individual stock again!
(2) Plus of course the downside that the timing signals will never work
as well in real life as they do in a backtest. However they're
probably still going to add a lot of steadiness to your portfolio with
at worst a tiny drag on long run performance, or (as I expect) a net improvement.
(3) Lastly, hedging is emotionally difficult. One side of every hedge is always a loss.
The majority of the individual hedge trades will not be profitable and
it is critical that you always pay attention to the rolling year returns of
both long+short, NEVER to the trade by trade returns from the hedges alone.
This is particularly problematic when the market has been rising strongly
and fairly steadily for a couple/few years (as now), and it starts to look like
hedging is a dumb idea and it goes out of fashion just at it should come into fashion.
But the long term matters and the short term doesn't.
As long as you print that out and paste it above your screen you'll be OK.
Jim
