Tuesday, April 26, 2005

Bond boxes

Just above this post you should see an hourly bar chart of the June '5 treasury bond futures. I've drawn some lines on this chart to illustrate what in my view is the most basic and useful type of technical analysis - Nicolas Darvas' "Box Theory". This is the only sort of technical analysis I think you will ever need.

Darvas described the box theory as his own invention in his 1961 book "How I made 2 million dollars in the stock market" (Its a fun read and I strongly recomend it). While Darvas may have indeed come up with the box theory on his own, the basic principles have been known to good traders for as long as there have been speculative markets because they neatly fit the observed behavior of all markets.

Will studying the box theory make you a good trader? Probably not, because successful trading requires certain artistic skills and a certain emotional makeup that most people don't have. But if you are one of the few who are cut out for the speculative game, you will find the box theory very useful tool.

Darvas failed to explain his box theory clearly in his books. In this blog I hope to communicate the basic principles of the box theory by illustrating them with some real time examples.

Boxes are often referred to as trading ranges but I prefer the term "box" out of deference to Darvas and also because of its snappy pronunciation. The basic principles of the box theory are easy to state. First, the size of boxes tends to change only slowly. Secondly, boxes of the same size tend to stack on top of one another very neatly. Finally, boxes show the trend direction: an uptrend will show a series of boxes at higher and higher levels and a downtrend will show the reverse.


Let's turn to the bond example above and see how it illustrates these principles.


The weekly trend in the bonds turned up from the 109 low on March 23. Although I have not shown it on the chart, the 109 level was easily projected as a box low using the boxes that had developed in the downtrend from the 117 top in February 2005. The start of the uptrend was then shown when the market broke above the 110-12 level which was the projected top of the last box in the downtrend and the actual low of the second-to-last box in the downtrend.

The first multi-day box in the new bond uptrend developed during the first two weeks of April. At this point I must empasize that delinating and/or estimating the location of the top or bottom of a box requires a certain artistic skill. I am convinced this skill can be acquired through constant practice and I am also convinced that this is the ONLY way to do it!

In any event, the two horizontal solid lines delineate an actual box that is 46 ticks wide. When your think the trend is up you must be a buyer near the lower edge of any box. That bottom edge may have already been formed by a previous price extreme or its location may only be an educated guess made by assuming the new box will be the same size as the preceeding box.

In this case the intial 6 hour rally off of 109 was 46 ticks, so estimating a 46 tick drop from the employment number high of 112-19 would have made you buy near 111-05; the low that same day was 110-25. Note also that the subsequent rally off of 110-25 was exactly 46 ticks! I would then be inclined to say that the box extended from 110-25 on the low side to 112-07 on the high side. The trading that occurred previously between 112-07 and 112-19 was all "employment number frenzy" and the market spent little time in that zone. So I chose not to include it in the new box.

On April 8 the market dropped to 110-28, just 3 ticks from the bottom of the box and a good chance to buy if you are bullish (which I was).

On April 12 the bonds broke above the 112-07 estimated top of the box. How much further would they go? Your first guess has to be 46 ticks above 112-07, i.e to 113-21. But you also should keep in mind that when a box lasts a week or more like this one did, the market will usually hestitate for a few days once it gets 1/2 the box size (in this case 23 ticks) above the top of the box. Note the hesitation for a day or so near 113-00.

The bonds reached 113-18 on April 15, only three ticks shy of the estimated top of the projected 46 tick box. What to do? Well, the conservative thing to do is to sell longs and plan to repurchase on a break of near 46 ticks. In this instance you would leave some profit on the table, but remember, you've got to let other people make a living too!

In any case, the market hesitated only a few hours near 113-21 and rallied further to 114-20. Note that 113-21 plus 23 ticks (1/2 the box size) is 114-12. As the market reacts downward from 114-20 where would you buy back your longs (assuming you were still bullish)? Well there are two choices. The first is the previous (and incorrect) estimate of the top of the box at 113-21. This is a fairly agressive play but it makes sense if you are convinced the upside potential is still significant (as I did).

The more conservative answer is to buy a 46 tick break. This would entail buying at 113-06 but the market only got down to 113-09. This illustrates a basic point. If you think the market will stop at a price level identified by the box method, it is usually best to make the trade a few ticks ahead of the level (remember that you have competitors!).

Where is the market now according to the box theory? Well this morning the bonds came down hard from a high at 114-18 (top of the box is 114-20). Usually a move from one extreme of a box ends near the other. The low today is 113-22 but this is 13 ticks away from the low of a 43 tick box, so I would prefer the market to drop further before being a buyer again. (If the bonds are stronger than I think I will regret this!). In any case, as I write this the market has rallied 12 ticks from its 113-22 low, so if the reaction continues, a 12 tick box with high 113-22 would put the reaction low near 113-10. I think the subsequent rally should carrry pretty close to 115-06, the top of the highest projected box on the chart.

No comments: