Nestor Sulikowski

Let’s use SAMOA to simulate a buy-and-hold strategy using 3 month T-Bills. For more information please make sure you read the following related blogs:

• SAMOA’s Booking System
• Buy and Hold: CBOT-TY futures
• Treasury Carry Calculations

Our strategy will buy a 13w TBill using an initial capital of $10,000,000, and will roll the position to the newest issue whenever there is a new auction. We chose to roll on issue date of the newest TBill.

• First, we need a list of rolls:

NOTES:

1. OTR, PriceOTR: The On-The-Run 13W TBill on DateS, with it’s price
2. Switch: “YES” when a roll needs to be made
3. OLD, PriceOLD: The old 13W, with it’s price (on the day of the switch)

• Now we can define the strategy with the language JScript as follows:

var K = 10000000;
function event_OpenBook() {
  context.AddBookTag("RepoLongs","OFF");
  context.AddCapital(K, "Initial K");
  //Initial Buy
  var q = Math.Floor( K / (vars.PriceOTR/100) );
  context.AddTrade(null, vars.OTR, q, vars.PriceOTR, "Initial Buy");
}
function event_OpenDay() {
    //Mark the book
    context.AddMark(vars.OTR, vars.PriceOTR);
    if (vars.PriceOLD!=null) context.AddMark(vars.OLD, vars.PriceOLD);
}
function event_CloseDay() {
    if (vars.Switch == "YES") {
        context.Liquidate("Switch: sell");
        //Switch: buy
        var q = Math.Floor( book.NAV / (vars.PriceOTR/100) );
        context.AddTrade(null, vars.OTR, q, vars.PriceOTR, "Switch: buy");
    }
}
function event_CloseBook() {
    context.Liquidate("Liquidate");
}

NOTES:

1. The line: context.AddBookTag("RepoLongs","OFF") turns off the automatic financing of longs in the repo market.
2. The rolls are executed inside the event “event_CloseDay”.

• When fed into ssRunStrategyS(…) we get the following trades:

NOTES:

1. One “Initial Buy” trade, followed by “Switch:sell” / “Switch:buy” rolls, followed by a “Liquidate” trade
2. Note that we the NAV grows, so does the Q (so that we stay fully invested)

• The returns are the following:

• When compared to a 3M TBill Total return index, we get an almost perfect match (for the months where the index is available):

NOTES:

1. The differences hover between –9 bps and 7 bps, which can be attributed to discrepancies in the dates chosen for the rollings.

• If we modified the strategy to roll always the same amount (and not stay fully invested), we would get the following differences with the index:

NOTES:

1. As expected the index outperforms (-1 bp to 40 bps), b/c the index stays fully invested.