Here's an outline of one way to play the long dated treasury yield curve using ladder assuming a $10,000 allocation amount. Looking at the US 30 year Treasury yield over more recent years and with Japan's recent 2% yield on their 30 year http://www.bloomberg.com/markets/rates-bonds/government-bonds/japan/ I've opted for a 5% top, 2% bottom ladder settings.
Whilst using that 30 year as a baseline, for extra volatility I'd actually buy/sell TMF (3x leveraged version of long dated treasury's).
With current US 30 year yields at around 2.92% http://www.bloomberg.com/markets/rates-bonds/government-bonds/us/ the ladder is close to the 2.88% rung level that indicates $6000 cash reserves. So buy $4000 of TMF at a recent price of $70.75 (out of hours price as I write this).
As the 30 year yield moves around, so that ladder will indicate either to increase or reduce cash i.e. we either buy or sell stock (TMF) to raise or reduce the cash reserve to the ladder indicated level.
I've levelled the ladder to trade $1000 at each step/rung. And the d'Alembert amount simply indicates a round trip up one step, down one step (or down one step, up one step) gross profit = $95.96 i.e. the price change between adjacent steps is 9.6% and we trade $1000 across that range = $96 (with a bit of rounding). Two trade costs per cycle (one buy, one sell) at perhaps $12.50 per trade = $25 costs leaving 96 - 25 = $71 net profit per cycle (pair of one buy, one sell trades).
This chart (which is live so it will change over time), shows some recent modest levels of volatility in yield
and with TMF scaling up the actual swings the actual gross profit could be 3x the figures I outlined above, maybe 96 x 3 = $288 less $25 trading costs = $263 net per step cycle.
We need to keep a rein on downside losses, so keep all volatility capture gains separate and if the core ladder losses reach $10,000 we dump out (stop loss), leaving the difference between volatility capture gains achieved whilst trading less the capital loss as the overall profit or loss. We also need to keep an eye on whether the model is still valid, i.e. if the economic circumstances change to look like yields might start to rise (inflation, rising interest rates etc.), then that could also be used as a stop/exit signal.
Its also useful to have another ladder that tracks an asset where the prices tend to move counter direction to each other, preferably scaled to somewhat similar magnitudes as well. Stocks do a pretty good job of that as this (also live) chart shows
i.e. the Dow's price moves somewhat in alignment with long dated treasury yields (and treasury prices move in the opposite direction to the yield). With two ladders and one adding as the other reduces and visa versa, the risks of a sharp move outside of the ladder top (or bottom) are reduced as losses and gains might be much the same from the two ladders and that cancel each other out, whilst both ladders might be generating volatility capture gains in the interim. In concept you could use the one ladder to do both i.e. have TMF as the 'stock' and Dow (or other stock index) as the 'cash'. But I find it better to run with two ladders so that each can be levelled to its own volatility/characteristics and periodically realign the two back into phase with each other.
A similar type setup could be made using AIM or LD-AIM, but as Conrad said the other day whichever we feel most comfortable with is the best choice and for me that's ladder (I like seeing all of the trade price levels and amounts to be traded up front so that limit orders can be left to do the work in my absence).
If someone told me that long treasuries were a bargain last year and that I would get a 30% return I would have called them crazy. (I guess that is why I should let AIM control my investing)
If I already owned them I would be selling some based on that 30% rise using AIM.
I will get in to long bonds the next time interest rates invert.
Market Data 
Markets 
AI
